Style gate
The online VERIFY half of the style-replication loop — the first-span-feedback discipline with a different oracle (match against the Group C exemplar corpus in references/targets.json) and a different sensor (a sidecar-free Style Profile of the DELIVERED span-0 pixels via profile-clip, never our plans). On a levered-field mismatch (|z| > SG_Z vs the corpus distribution) it runs a bounded knob-first fix loop — moves.py maps the worst mismatches to CLAMPED ephemeral env deltas (JUMP_CUT_SEG, JUMP_CUT_MAX_GAP, SPEED, SWITCH_SPACING) in work/<id>/style.env, the entrypoint-owned SG_RERENDER_CMD re-renders span 0 with them, and the fix is accepted only when accept.py passes (no checked field worse, at least one improved, no new mismatch) AND span 0's grade.json did not regress (reuses first-span-feedback's check_grade.py). Accepted knobs steer spans 1..N (the entrypoint sources style.env before fan-out) and land in references/accepted.jsonl; mismatches no knob can reach are SURFACED for an operator-approved durable code change. Self-arming — no corpus targets or n < SG_MIN_REFS → exact no-op, so the stock pipeline is unchanged until a corpus is learned. NON-FATAL everywhere, idempotent (.sgmeta), STYLE_GATE=0 force-disables.From its SKILL.md
npx -y skills add jperrello/C0BALT_CUT --skill style-gateAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
2.0 KB, 231 tokens by cl100k_base, as published. Nobody here has run it
style-gate
bash .claude/skills/style-gate/style-gate.sh <work_id> <clip_stem> <saved_mp4> [--pane <p>]
Emits work/<id>/<stem>.style.json (the match verdict: checked fields,
mismatches ranked by |z| each with lever + direction, diagnostic rows) and,
on an accepted move, work/<id>/style.env (the knob overrides spans 1..N
inherit) + a references/accepted.jsonl record.
Env: STYLE_GATE (1; hard no-op without references/targets.json),
SG_Z (1.5), SG_MAX_ITERS (1), SG_MAX_MOVES (2), SG_MIN_REFS (3),
SG_RERENDER_CMD (entrypoint-owned), STYLE_REFS (default references/).
Vision is off for gate profiling (PROFILE_VISION=0 internally) — the gate
only scores deterministic levered fields, so it needs no model call.
What ships with it: 6 files
19.9 KB alongside SKILL.md, 6 of them executable
- accept.pyruns1.5 KB
- match.pyruns2.4 KB
- moves.pyruns1.8 KB
- style-gate.shruns6.6 KB
- test_gate.shruns3.7 KB
- test_match.pyruns3.8 KB