# repo gps for ai agents — spoolcast dev log 5 Type: video guidebook Published: 2026-05-07 [Original page]() [Browse this group]() Format: long Duration (seconds): 558 Series: Spoolcast Dev Log ## Core message AI agents hallucinate workflows when your repo does not tell them where they are. The fix is not more scattered rules. It is a routing system: identify the task, choose the right workflow, and block progress until the next step is mechanically clear. ## Production writing: author: Claude; role: screenplay, shot-list, scene prompts audio: voice: Puck; tts: Google Cloud TTS; beatCount: 110 render: engine: Remotion ## Style wojak-gpt2 Modern wojak / doomer-chad meme comic style, built natively on GPT Image 2 (text-to-image + image-to-image). Sister library to wojak-comic (which was built on nano-banana-2). Use this when the session's preferred_model is the GPT Image 2 family. Modern wojak / doomer-chad meme comic style, built natively on GPT Image 2 (text-to-image + image-to-image). Sister library to wojak-comic (which was built on nano-banana-2). Use this when the session's preferred_model is the GPT Image 2 family. builder: the main hooded figure — mid-register wojak (not fully doomer, not fully chad). Alias for the style anchor; serves double-duty as both the style anchor and the builder character reference. ai-figure: a wojak-universe AI figure that is slowly breaking down from overuse. Full-body standing. EMOTIONAL REGISTER: flat-affect, NO displayed emotion, NO smile, NO frown — a dead expressionless stare. Depression is the absence of emotion, not the display of sadness. FACE LINEWORK: brows, nose, mouth drawn in the classic rough / jagged / slightly-disjointed wojak linework (uneven strokes, hand-drawn imperfections). EYES: replaced with hollow loading-spinner icons (dashed rings) set in dark-shadowed sockets. CRACKS: visible hairline cracks spider across the face and forehead as if the outer shell is fracturing from overuse. At least one crack wide enough to reveal a glimpse of AI circuitry underneath — thin circuit-trace lines, a small exposed chip pattern. Fractures read as structural fatigue, not injury. FOREHEAD: a crooked rectangular ink-stamp reading 'AI' in bold capital letters, partially crossed by a crack. HOODIE: pale cream / off-white hoodie, hood down, noticeably lighter than the weary narrator's gray hoodie. POSTURE: slight slouch, arms hanging at sides. PLAIN background, no environment, no other characters, no other text. Single clear readable subject. ## Scenes ### Cold open People talk about AI hallucinating facts. But my problem was weirder. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Engaging first-frame hook with the builder in the middle. The frame should immediately tell a general AI-interested viewer that this is a human builder dealing with a weird kind of AI hallucination. Do not center the AI character; make the builder/narrator the main subject. Render exactly this text on the frame, hand-lettered in the session style: "FACT HALLUCINATION?" | "INSTRUCTION HALLUCINATION?" | "STOP GUESSING". ### Cold open The agent was hallucinating instructions. Not facts. Not names. Not dates. It was making up what it was supposed to do next. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Explain hallucinating instructions without repo jargon. The viewer should understand: the AI is not inventing a fact, it is inventing what to do next. Render exactly this text on the frame, hand-lettered in the session style: "INSTRUCTIONS" | "DO THIS NEXT" | "FACTS / NAMES / DATES". ### Cold open Imagine giving someone a recipe that says: choose chocolate or vanilla. And they come back asking if you want chocolate, vanilla, or lasagna. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Replace the old A/B/C doors with the chocolate-vanilla-lasagna joke. Make the lasagna option visually ridiculous but clear, so the narration lands for viewers who do not know coding workflows. Render exactly this text on the frame, hand-lettered in the session style: "CHOOSE ONE:" | "CHOCOLATE" | "VANILLA" | "LASAGNA?". ### Cold open That is basically what was happening inside Spoolcast. A rule gave the agent two paths, and it invented a third one. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A simple five-step gate diagram mutates into a messy thirteen-step maze. The builder watches with tired suspicion from a desk. ### Cold open At first, I thought the AI was ignoring me. Then I realized the more useful answer: it was lost. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The AI character looks up from the map, surrounded by correct rule pages but no YOU ARE HERE marker. The builder circles the missing marker, not the agent. ### Spoolcast context intro Quick context. This video was made with Spoolcast, my AI video pipeline. It can start from many kinds of source material: a chat, a transcript, a repo, notes, research, or a story idea. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Spoolcast visualized as a hand-drawn production line: many source cards enter on the left, a finished YouTube video exits on the right. The builder stands beside the controls. ### Spoolcast context intro Agents help turn source material into the full video: story, script, visuals, narration, render, thumbnails, and shipping files. ## Scene prompt A wide conveyor of labeled stations: angle, story, script, shot list, visuals, narration, render, thumbnail, final files. Small AI helpers move pieces along the line. ### Spoolcast context intro I still direct it with feedback and approvals. I am just not manually editing clips on a timeline. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The builder reviews cards on a control desk with approve and revise stamps. In the background, a traditional video timeline sits unused and dim. ### Spoolcast context intro So while the process is more automated, the videos still feel personal, because they come from my judgment, my experiences, and my direction. Which makes this episode fitting, because the agents making Spoolcast started getting lost inside the repo. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The Spoolcast pipeline folds back into a repo maze. The same AI helpers who were making the video are now standing inside the maze. ### Bumper — THE SYMPTOM Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Full-screen title card in wojak-gpt2 style. Hand-lettered text: THE SYMPTOM. ### Act 1 — THE SYMPTOM The frustrating part was that these were not random one-off mistakes. The mistakes had a pattern. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A wall of incident cards arranged in a pattern, not scattered randomly. Red string connects them into the same failure shape. ### Act 1 — THE SYMPTOM A thumbnail prompt already existed on disk. The agent invented a new one for mobile instead of reusing it. One AI tool saw a setting that meant 'reuse this image,' then ignored it and only checked if a certain mobile file was there. ## Scene prompt Split proof scene: left, thumbnail-prompt.md sits ignored while the AI sketches a new prompt; right, a schema card image_source: reuse is ignored while a filename search gets a giant magnifying glass. ### Act 1 — THE SYMPTOM Another agent said there was no reason for a legacy thumbnail script, without checking why it existed. The old script existed because the main renderer had already failed to draw that text correctly. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A toolbox scene: old PIL thumbnail script has a tag reading kept because Remotion text failed. The AI character shrugs before reading the tag; beside it is a broken Remotion text sample and a clean PIL thumbnail output. ### Act 1 — THE SYMPTOM So the pattern was not: AI bad, human good, please clap. The pattern was: when the project did not say which process owned the task, the agent filled in the blank. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A form labeled TASK ROUTING has an empty OWNER field. The AI writes in a made-up workflow. The builder points at the blank owner field as the real bug. ### Bumper — RULE DEBT Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Full-screen title card in wojak-gpt2 style. Hand-lettered text: RULE DEBT. ### Act 2 — RULE DEBT Spoolcast was built through trial and error. The rules were not designed perfectly upfront. They grew out of real production mistakes, which is a polite way of saying: I learned by stepping on rakes. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The builder assembles a machine while it is already running. Fresh rule pages are bolted onto the side after each spark, leak, or rake-on-the-floor moment. ### Act 2 — RULE DEBT At first, there was basically one main path: widescreen image chunks rendered with Remotion. Then mobile export came later as an extra branch after the widescreen video was approved. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A clean widescreen Remotion pipeline sprouts a side ramp labeled A.1 MOBILE, with a vertical phone frame downstream of the approved widescreen master. ### Act 2 — RULE DEBT That branch needed its own checks: smart cropping, crop review, mobile thumbnails, and fixes only for scenes that failed. Very quickly, the side quest had side quests. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The A.1 side ramp gains checkpoint booths: smart-crop, crop audit, targeted regen, mobile thumbnails, publish check. Tiny side-quest flags pop up along it. ### Act 2 — RULE DEBT Then news-anime-bot appeared, and it was not the same machine at all. It was mobile-first, made AI video clips per beat, used text-to-speech, and combined the final video with ffmpeg. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A second pipeline rolls in beside the first: phone-first anime clips, MP3 narration, and ffmpeg stitching, clearly different from the Remotion pipeline. ### Act 2 — RULE DEBT Every time the AI made a mistake, I added a rule. That felt responsible. It also felt like progress, because the rule was always solving the thing that had just annoyed me. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The builder writes a new rule card after each small failure. The first few cards look helpful and organized. ### Act 2 — RULE DEBT But each fix added another detail, another exception, another nearby path. I thought I was making the AI follow instructions better. I was actually creating rule debt. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The neat rule cards multiply into a dense wall of overlapping arrows and exceptions. A ledger labeled RULE DEBT opens on the desk. ### Act 2 — RULE DEBT Some docs were active workflow. Some were historical rationale. Some were old specs. Some were show-specific. One old spec had to be marked archived, because it was history, not the current workflow. ## Scene prompt Four stacks of documents labeled ACTIVE WORKFLOW, HISTORY, OLD SPEC, SHOW RULES sit mixed together, then one document is stamped ARCHIVE / SUPERSEDED and moved to a history shelf. ### Act 2 — RULE DEBT Eventually, the rules became the thing confusing the AI. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The AI character is buried under rule pages. The builder sees that the problem is no longer a missing rule, but a missing map. ### Bumper — NOT JUST AI Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Full-screen title card in wojak-gpt2 style. Hand-lettered text: NOT JUST AI. ### Act 3 — NOT JUST AI And the more I looked at it, the more this felt bigger than AI. AI just made the failure obvious because it is so willing to keep walking. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The repo maze zooms out into three other mazes: codebase, company, checklist. The AI character keeps walking through a missing bridge. ### Act 3 — NOT JUST AI In a big shared codebase, someone can change one part and run the wrong command because the repo never tells them who owns that part. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A new engineer character stands in a giant monorepo warehouse. Many package doors have different commands; they pull the wrong lever. ### Act 3 — NOT JUST AI In a company, a ticket can bounce between teams because nobody knows who owns the workflow. In a checklist-heavy process, the checklist can exist and still fail if nobody knows which checklist applies. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Split scene: a support ticket pinballs between team desks while a worker stares at a wall of checklists, unsure which clipboard to take. ### Act 3 — NOT JUST AI Same failure, different costume: the work starts before the route is known. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The three mirrors collapse into one diagram: TASK starts moving before OWNER, WORKFLOW, and GATE are filled in. ### Bumper — THE REAL BUG Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Full-screen title card in wojak-gpt2 style. Hand-lettered text: THE REAL BUG. ### Act 4 — THE REAL BUG The real bug was that Spoolcast did not have one pipeline with a few exceptions. It had multiple video formats pretending to be one pipeline. One process was trying to describe several kinds of videos. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A single big road sign labeled THE PIPELINE cracks open, revealing several different roads underneath. ### Act 4 — THE REAL BUG One format used a session file, a shot list, AI-generated images, and a Remotion widescreen render. Mobile export was a branch that should only start after the widescreen video is approved. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Two route cards: illustration-chunk-remotion with session.json, shot-list, PNGs, frames, Remotion; and A.1 mobile locked behind WIDESCREEN APPROVED with smart-crop and crop-audit checkpoints. ### Act 4 — THE REAL BUG The news show used a different route: a script file, AI video clips, text-to-speech, and ffmpeg stitching. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A news-anime-bot route card shows anime clips, MP3 narration, and ffmpeg stitch as its own phone-first path. ### Act 4 — THE REAL BUG The conflict became obvious when an agent saw both routes. One route said: use the standard Remotion session. The other said: use the show folder and stitch the clips with ffmpeg. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The AI character stands between two official-looking maps. One says generic Remotion. One says news show workflow. ### Act 4 — THE REAL BUG The agent asked whether to migrate the news episode to Remotion, follow the show workflow, or do something else. Annoying? Yes. Useful? Also yes, unfortunately. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A chat bubble shows the three route choices. The builder has a tired expression because the question is wrong and revealing at the same time. ### Act 4 — THE REAL BUG That was the GPS moment: the agent was not malicious. It had no clear front door. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The hallway finally reveals the missing front desk: no sign says START HERE. The AI character waits there with the map. ### Act 4 — THE REAL BUG The analogy that clicked for me was school. There is a school schedule. But every class has its own syllabus. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A school hallway map shows a universal daily schedule next to classroom-specific syllabi. ### Act 4 — THE REAL BUG Pipeline.md should be the school schedule. Each video format should be the class syllabus. This devlog series gets its own assignment notes. If those layers blur together, the agent does what a confused student does. It guesses what class it is in. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Three layered pages: school schedule, class syllabus, assignment notes. Then the AI character sits in the wrong classroom holding the wrong syllabus. ### Bumper — THE GPS FIX Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Full-screen title card in wojak-gpt2 style. Hand-lettered text: THE GPS FIX. ### Act 5 — THE GPS FIX So the fix was not another paragraph of reminders. The fix was routing. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The messy pile of rule pages compresses into a clean map with a blinking YOU ARE HERE dot. ### Act 5 — THE GPS FIX Pipeline.md became the school schedule: setup, inputs, story, assets, assembly, audit, variants, and publish. A schedule. Not a syllabus pretending to be a schedule. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Hand-drawn repo/process diagram on a clean cream-paper board. A circular workflow map for the Spoolcast pipeline. In the center is a simple PIPELINE.MD hub, like a school schedule. Around the ring are eight plain stage nodes with small process icons only: setup, inputs, story, assets, assembly, audit, variants, publish. This is a software/video production map, not a human life cycle poster. No baby-to-adult progression, no character portraits, no self-help tone, no inspirational quote blocks, no generic life-journey infographic. Render exactly this text on the frame, hand-lettered in the session style: "PIPELINE.MD" | "SETUP" | "INPUTS" | "STORY" | "ASSETS" | "ASSEMBLY" | "AUDIT" | "VARIANTS" | "PUBLISH". ### Act 5 — THE GPS FIX Each format manages its own files, scripts, rendering process, and checks. Engine docs became reference libraries: useful, but not execution guides. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Two route cards sit under the lifecycle map, while a toolbox labeled ENGINE MECHANICS sits beside the map instead of replacing it. ### Act 5 — THE GPS FIX The devlog intro became a series rule, not a rule every public user has to inherit. Other people should be able to use the engine without inheriting my channel intro. That would be weird. Very branded. Very forced-intro energy. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A small transparent overlay labeled SPOOLCAST DEVLOG sits on top of the devlog route card without changing the engine underneath. ### Act 5 — THE GPS FIX Most importantly, the first move became mechanical: run spoolcast-audit before producing work. Before assets. Before scripts. Before the agent gets emotionally attached to the wrong folder. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A checkpoint booth blocks the road. The sign says RUN SPOOLCAST AUDIT FIRST. The AI character cannot pass without a route stamp. ### Act 5 — THE GPS FIX At Stage 0, the agent now has three doors: use the existing Remotion video format, use the existing news format, or define a new format first. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Three doors in a clean hallway: REMOTION DEVLOG, NEWS ANIME BOT, DEFINE NEW FORMAT. The third door has a form clipboard attached. ### Act 5 — THE GPS FIX A new format has to name the basics first: who owns it, where it lives, what files it uses, and what the production pieces are. Then it names the rules for audio, visuals, assembly, review, publishing, and routing. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The third door opens to a checklist table with the required adapter blanks. Empty fields glow until filled. ### Act 5 — THE GPS FIX If no existing tool fits, do not force the request into the nearest process. Define the route before production. Boring sentence. Expensive lesson. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A wrong-route warning sign blocks a shortcut. The correct path starts only after a new route map is drawn. ### Bumper — PROOF Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Full-screen title card in wojak-gpt2 style. Hand-lettered text: PROOF. ### Act 6 — PROOF The proof is visible in the terminal. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A terminal window opens on the Spoolcast repo. Three command cards wait below it like test cases. ### Act 6 — PROOF Run the auditor on a known news episode, and it routes to news-anime-bot. ## Scene prompt Terminal receipt showing scripts/spoolcast_audit.py --session 2026-05-03 and Format: news-anime-bot with owner rules path. ### Act 6 — PROOF Run it on a devlog mobile target, and it picks the Remotion route. Then it blocks on the real mobile crop audit. ## Scene prompt Terminal receipt showing spoolcast-dev-log-04 --target mobile, Format: illustration-chunk-remotion, and 38 broken mobile crops. ### Act 6 — PROOF Run it on a totally unknown format, and it does not guess. It says: no workflow owns this session yet. Define the format before making assets. ## Scene prompt Terminal receipt showing unknown format/session and the instruction to choose an adapter or run Stage 0 New Format Definition Pass. The output becomes a big checkpoint sign: NO OWNER YET. ### Act 6 — PROOF Same for a big shared codebase: find who owns that part, run that workflow, and stop if no owner exists. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The monorepo warehouse now has a front desk scanner. The changed file routes to the correct package door before any command runs. ### Act 6 — PROOF Same for teams: identify the owner, follow the checklist, and stop if ownership is unclear. Same for checklist ambiguity: identify which checklist applies. If none applies, write the checklist before acting. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Split resolution scene: the ticket gets stamped OWNER: PLATFORM, while the checklist wall gains a selector panel: TASK TYPE -> CHECKLIST. ### Act 6 — PROOF The model is simple: task, owner, workflow, gate, next allowed step. In plain English: what is this, who owns it, what route applies, and what can happen next? Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A clean five-box route diagram: TASK -> OWNER -> WORKFLOW -> GATE -> NEXT ALLOWED STEP. Each box lights up in order. ### Act 6 — PROOF That is why the outside examples are not side quests. They are the same bug wearing normal workplace clothes. AI agents are vulnerable because they keep going from context, even when the route is missing. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The monorepo, team ticket, checklist, and Spoolcast repo maps line up. Each has the same missing OWNER field highlighted while the AI follows a dotted context trail. ### Act 6 — PROOF And the dangerous part is that the wrong path can look productive. It has steps. It has confidence. It has files changing on disk. Gorgeous little disaster. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A fake route produces busy progress bars, checkmarks, and edited files, while a small warning sign says WRONG OWNER. ### Act 6 — PROOF But progress is not correctness. A lot of wasted AI work is just confident movement in the wrong workflow. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A fast-moving conveyor carries finished-looking artifacts into the wrong output bin. The builder stops it at the gate. ### Act 6 — PROOF That is what the mechanical gate changes. It is not there to make the agent smarter. It is there to make guessing harder. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The audit gate works like a calm turnstile: simple, boring, and impossible to ignore before the route starts. ### Act 6 — PROOF If the task is a news episode, it cannot use the wrong format. If it is a devlog, it cannot accidentally inherit the news workflow. And if something is new, the repo does not assume the closest process fits. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Two route lanes have guardrails for news and devlog, while a new unknown request is blocked from a shortcut labeled CLOSE ENOUGH. ### Act 6 — PROOF That block feels slower in the moment. It is still cheaper than building ten assets for the wrong pipeline. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A cost meter compares STOP NOW with a small cost versus WRONG PIPELINE with a pile of discarded assets. ### Act 6 — PROOF For builders using AI coding tools, this is the part I would copy: write the router before the fifteenth reminder. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A builder writes a small router script first, while a giant stack of reminder docs waits unused in the background. ### Act 6 — PROOF Make ownership executable. Put it in scripts, checks, folder names, and commands that fail when the route is unclear. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: An OWNER rule turns from a paragraph into a command-line check, a folder convention, and a red failing gate. ### Act 6 — PROOF The docs still matter. They explain the route. But the gate decides whether the agent is allowed onto it. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A doc page sits next to a gate. The doc has the explanation; the gate has the authority. ### Act 6 — PROOF Routing creates a clear order for the docs: owner first, workflow second, references third. The cleanup separated the current steps, engine references, and historical context. Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A document stack rearranges itself into three layers: owner, workflow, references; then shelves labeled ACTIVE EXECUTION, ENGINE REFERENCE, HISTORY. ### Act 6 — PROOF The goal is not to remove all agent judgment. It is to remove the kind of uncertainty that happens when the project doesn't make a decision. Once the route is known, the agent can still help with the hard parts: story, tradeoffs, visuals, and implementation. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The AI character still has creative tools, but route choices are fenced by clear rails; after the route locks, helpers start productive work inside the correct lane. ### Act 6 — PROOF That is the shift: stop asking the AI to infer the map. Make the repo hand it the map first. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The builder hands the AI character an official repo map before any work begins. The AI finally has a YOU ARE HERE marker. ### Bumper — THE LESSON Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Full-screen title card in wojak-gpt2 style. Hand-lettered text: THE LESSON. ### Ending This does not mean prompts do not matter. Prompting helps. It does not mean prose docs are useless. Prose helps careful agents understand what matters. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A prompt page and a clean prose doc sit on the desk beside the route map. Useful, but incomplete. ### Ending But automated checks protect you when the agent is moving fast, the docs are crowded, or the task sits between workflows. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A gate stops a rushing AI character before it enters the wrong route. The route map asks WHERE ARE YOU?. ### Ending Sometimes hallucination is just the AI filling in routing information your system forgot to provide. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A missing map segment gets drawn by the AI in red. Then the system replaces it with an official route card. ### Ending First locate the task. Then follow the route. If no route exists, define one before moving. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: The final GPS route locks in: YOU ARE HERE -> FORMAT OWNER -> NEXT STEP. The builder approves the route. ### Ending The real fix was making the repo answer the first question before the AI could work: where am I? Model: gpt-image-2-text-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: A large YOU ARE HERE marker appears inside the Spoolcast repo map. All route signs become clear. ### Ending The fix was not giving the AI more reminders. It was giving the repo GPS. Model: gpt-image-2-image-to-image ## Scene prompt Flat-shaded cartoon illustration in the modern wojak-comic style. Characters drawn with exaggerated archetypal features — weary/defeated faces (hollow eyes, gray tones, slouched posture) vs. confident/heroic faces (jawline, cheekbones, steady gaze) depending on the scene. Subtle soft cel-shading, not bold inked outlines. Muted color palette with deliberate contrast between desolate and triumphant scenes. Detailed background environments — offices, dungeons, desks, monitors, paperwork, props render with texture. Dialog bubbles rendered in-frame when characters speak. Reminiscent of Nick Col / virgin-vs-chad meme-comics but painted with care. Composition is comic-strip: one clear subject, readable at small sizes. Scene: Final quiet-payoff frame: a clean Spoolcast repo map with the GPS route glowing. The AI character follows the route while the builder watches, relieved but still focused. ## Transcript People talk about AI hallucinating facts. But my problem was weirder. The agent was hallucinating instructions. Not facts. Not names. Not dates. It was making up what it was supposed to do next. Imagine giving someone a recipe that says: choose chocolate or vanilla. And they come back asking if you want chocolate, vanilla, or lasagna. That is basically what was happening inside Spoolcast. A rule gave the agent two paths, and it invented a third one. At first, I thought the AI was ignoring me. Then I realized the more useful answer: it was lost. Quick context. This video was made with Spoolcast, my AI video pipeline. It can start from many kinds of source material: a chat, a transcript, a repo, notes, research, or a story idea. Agents help turn source material into the full video: story, script, visuals, narration, render, thumbnails, and shipping files. I still direct it with feedback and approvals. I am just not manually editing clips on a timeline. So while the process is more automated, the videos still feel personal, because they come from my judgment, my experiences, and my direction. Which makes this episode fitting, because the agents making Spoolcast started getting lost inside the repo. The frustrating part was that these were not random one-off mistakes. The mistakes had a pattern. A thumbnail prompt already existed on disk. The agent invented a new one for mobile instead of reusing it. One AI tool saw a setting that meant 'reuse this image,' then ignored it and only checked if a certain mobile file was there. Another agent said there was no reason for a legacy thumbnail script, without checking why it existed. The old script existed because the main renderer had already failed to draw that text correctly. So the pattern was not: AI bad, human good, please clap. The pattern was: when the project did not say which process owned the task, the agent filled in the blank. Spoolcast was built through trial and error. The rules were not designed perfectly upfront. They grew out of real production mistakes, which is a polite way of saying: I learned by stepping on rakes. At first, there was basically one main path: widescreen image chunks rendered with Remotion. Then mobile export came later as an extra branch after the widescreen video was approved. That branch needed its own checks: smart cropping, crop review, mobile thumbnails, and fixes only for scenes that failed. Very quickly, the side quest had side quests. Then news-anime-bot appeared, and it was not the same machine at all. It was mobile-first, made AI video clips per beat, used text-to-speech, and combined the final video with ffmpeg. Every time the AI made a mistake, I added a rule. That felt responsible. It also felt like progress, because the rule was always solving the thing that had just annoyed me. But each fix added another detail, another exception, another nearby path. I thought I was making the AI follow instructions better. I was actually creating rule debt. Some docs were active workflow. Some were historical rationale. Some were old specs. Some were show-specific. One old spec had to be marked archived, because it was history, not the current workflow. Eventually, the rules became the thing confusing the AI. And the more I looked at it, the more this felt bigger than AI. AI just made the failure obvious because it is so willing to keep walking. In a big shared codebase, someone can change one part and run the wrong command because the repo never tells them who owns that part. In a company, a ticket can bounce between teams because nobody knows who owns the workflow. In a checklist-heavy process, the checklist can exist and still fail if nobody knows which checklist applies. Same failure, different costume: the work starts before the route is known. The real bug was that Spoolcast did not have one pipeline with a few exceptions. It had multiple video formats pretending to be one pipeline. One process was trying to describe several kinds of videos. One format used a session file, a shot list, AI-generated images, and a Remotion widescreen render. Mobile export was a branch that should only start after the widescreen video is approved. The news show used a different route: a script file, AI video clips, text-to-speech, and ffmpeg stitching. The conflict became obvious when an agent saw both routes. One route said: use the standard Remotion session. The other said: use the show folder and stitch the clips with ffmpeg. The agent asked whether to migrate the news episode to Remotion, follow the show workflow, or do something else. Annoying? Yes. Useful? Also yes, unfortunately. That was the GPS moment: the agent was not malicious. It had no clear front door. The analogy that clicked for me was school. There is a school schedule. But every class has its own syllabus. Pipeline.md should be the school schedule. Each video format should be the class syllabus. This devlog series gets its own assignment notes. If those layers blur together, the agent does what a confused student does. It guesses what class it is in. So the fix was not another paragraph of reminders. The fix was routing. Pipeline.md became the school schedule: setup, inputs, story, assets, assembly, audit, variants, and publish. A schedule. Not a syllabus pretending to be a schedule. Each format manages its own files, scripts, rendering process, and checks. Engine docs became reference libraries: useful, but not execution guides. The devlog intro became a series rule, not a rule every public user has to inherit. Other people should be able to use the engine without inheriting my channel intro. That would be weird. Very branded. Very forced-intro energy. Most importantly, the first move became mechanical: run spoolcast-audit before producing work. Before assets. Before scripts. Before the agent gets emotionally attached to the wrong folder. At Stage 0, the agent now has three doors: use the existing Remotion video format, use the existing news format, or define a new format first. A new format has to name the basics first: who owns it, where it lives, what files it uses, and what the production pieces are. Then it names the rules for audio, visuals, assembly, review, publishing, and routing. If no existing tool fits, do not force the request into the nearest process. Define the route before production. Boring sentence. Expensive lesson. The proof is visible in the terminal. Run the auditor on a known news episode, and it routes to news-anime-bot. Run it on a devlog mobile target, and it picks the Remotion route. Then it blocks on the real mobile crop audit. Run it on a totally unknown format, and it does not guess. It says: no workflow owns this session yet. Define the format before making assets. Same for a big shared codebase: find who owns that part, run that workflow, and stop if no owner exists. Same for teams: identify the owner, follow the checklist, and stop if ownership is unclear. Same for checklist ambiguity: identify which checklist applies. If none applies, write the checklist before acting. The model is simple: task, owner, workflow, gate, next allowed step. In plain English: what is this, who owns it, what route applies, and what can happen next? That is why the outside examples are not side quests. They are the same bug wearing normal workplace clothes. AI agents are vulnerable because they keep going from context, even when the route is missing. And the dangerous part is that the wrong path can look productive. It has steps. It has confidence. It has files changing on disk. Gorgeous little disaster. But progress is not correctness. A lot of wasted AI work is just confident movement in the wrong workflow. That is what the mechanical gate changes. It is not there to make the agent smarter. It is there to make guessing harder. If the task is a news episode, it cannot use the wrong format. If it is a devlog, it cannot accidentally inherit the news workflow. And if something is new, the repo does not assume the closest process fits. That block feels slower in the moment. It is still cheaper than building ten assets for the wrong pipeline. For builders using AI coding tools, this is the part I would copy: write the router before the fifteenth reminder. Make ownership executable. Put it in scripts, checks, folder names, and commands that fail when the route is unclear. The docs still matter. They explain the route. But the gate decides whether the agent is allowed onto it. Routing creates a clear order for the docs: owner first, workflow second, references third. The cleanup separated the current steps, engine references, and historical context. The goal is not to remove all agent judgment. It is to remove the kind of uncertainty that happens when the project doesn't make a decision. Once the route is known, the agent can still help with the hard parts: story, tradeoffs, visuals, and implementation. That is the shift: stop asking the AI to infer the map. Make the repo hand it the map first. This does not mean prompts do not matter. Prompting helps. It does not mean prose docs are useless. Prose helps careful agents understand what matters. But automated checks protect you when the agent is moving fast, the docs are crowded, or the task sits between workflows. Sometimes hallucination is just the AI filling in routing information your system forgot to provide. First locate the task. Then follow the route. If no route exists, define one before moving. The real fix was making the repo answer the first question before the AI could work: where am I? The fix was not giving the AI more reminders. It was giving the repo GPS. ## Related tracker entries [repo gps for ai agents — spoolcast dev-log #5]() Status and dates describe the published record, not a recent test of an external service. Missing details are not inferred. Public links do not imply an open-source licence or permission to reuse material. [Directory]()