gooseworks-ai/goose-skills / skills/video/render-street-interview
render-street-interview
VerifiedWrite a street-interview ad from a complete inspected commercial interaction and current brand facts. Render product guessing, or prepare mic-only, prepared-sample or visible-task conversation script/prompt previews. Premise, actions and ad connection vary; capture rules stay fixed. Conversation media delivery is unverified.
tests/fixtures/episodes/grimwater-ep2fix.json
View on GitHub{
"_comment": "EPISODE 2 REPAIR. The user approved episode 2 (4820-4822) for its microphone, cast, grading and naturalness and asked for THAT fixed rather than replaced. So take A is seed 4820 ITSELF, the approved footage, untouched beyond the grade every take gets. Only the two takes that were wrong are new: 4834 and 4835, both generated with --scene-ref refs/scene/sceneref-4820-t5-noface.png so they inherit 4820's own microphone, street and light rather than re-deriving them from words -- which is the finding the whole run turned on (SKILL.md 27, 30). Episode 2's defect was that each take invented its own mic and its own street; the repair couples all three to 4820's.",
"episode": "ep2fix",
"brand": "Grimwater",
"slug": "gw",
"title": "What's in the can",
"_output_comment": "The output name carries all three seeds ON PURPOSE. build_looks.py's look filenames do not carry a seed and a seed-4816 build silently overwrote seed 4815's shipped deliverable once. This name cannot collide with episode 1 (4817-4819), episode 2 (4820-4822), any street-gw-*.mp4, or any 4815/4816/4823-4828 file.",
"output": "episode-gw-ep2fix-seed4820-4834-4835",
"takes": [
"grimwater-4820",
"grimwater-4834",
"grimwater-4835"
],
"brand_layer": {
"logo": "assets/grimwater/logos/grimwater-logo-3840.png",
"end_card": [
"NOTHING TO FEAR.",
"IT'S ONLY WATER."
]
},
"_grade_comment": "Not stored, same as episode 2. The episode is graded ONCE on the concatenated cut and the black lift is solved against that file by fit_grade.py at build time. A grade constant measured on one take outlives the take it was measured on (SKILL.md 19), and grading three takes separately is three curves across two joins, which is a visible step.",
"_ambience_comment": "Not stored either, and this one has bitten twice (SKILL.md 13 and 21a): the speech-free window belongs to a FILE. build_episode.py finds it on the finished concatenation and asserts it is speech-free before cutting a bed or an end-card room tone from it.",
"_drop_comment": "TWO SHOTS OF SEED 4820 COME OUT, both of the woman in the grey raincoat. The user saw a subject 'holding a second microphone' at episode-2 7.5s. Checked against the source: the object in her left hand is her PHONE, which is a configured prop in brands/grimwater-4820.json. It is not a microphone -- and at playback size, dark, vertical and held at chest height next to the interviewer's mic, it reads as one, which is what matters. The viewer's reading is the defect. It is in BOTH of her shots (4.54-6.33 and 6.33-12.10), not just the second, so both go and she comes out of the episode; her line 'Beer. That's a beer.' is recorded as cut. The lesson for the configs is the real finding: do not give a subject a dark handheld prop. A phone next to a microphone is a second microphone.",
"drop_shots": {
"gw-single-seed4820.mp4": [
[
4.54,
6.33
],
[
6.33,
12.096
]
]
},
"coupled_by_scene_ref": "refs/scene/sceneref-4820-t5-noface.png",
"_coupled_comment": "Seed 4820 predates the 'name the city' change, so its location clause differs from 4834's and 4835's by the two words ', in London' and the string comparison refuses the episode. The takes are in fact coupled more tightly than any episode that has passed that check: 4834 and 4835 were generated FROM A CROP OF 4820 ITSELF, so they inherit 4820's actual street, microphone and light rather than a description of them. Verified by eye across nine frames: same red brick mansion block, same green scaffolding, same bus shelter, same cycle lane, same mic. The build prints the differing clauses every time and check C reports them."
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |