Skip to content

feat: add harness-agnostic ytstudio agent skill#36

Merged
jdwit merged 2 commits into
mainfrom
feat/agent-skill
Jun 17, 2026
Merged

feat: add harness-agnostic ytstudio agent skill#36
jdwit merged 2 commits into
mainfrom
feat/agent-skill

Conversation

@jdwit

@jdwit jdwit commented Jun 17, 2026

Copy link
Copy Markdown
Owner

Closes #29.

Authors and publishes a harness-agnostic ytstudio agent skill, following the open SKILL standard. Per the issue discussion, it lives in this repo at skills/ytstudio/ (no separate repo), and the bundled reference is generated from the CLI so it cannot drift.

What's included

skills/ytstudio/
├── SKILL.md                          # frontmatter + concise instructions
├── references/
│   └── reference.md                  # full command reference (generated, committed)
└── assets/
    └── upload-sidecar.example.yaml   # example upload sidecar
  • SKILL.mdname: ytstudio / one-line description of when to use it. Body uses progressive disclosure: prerequisites (install, one-time OAuth, profiles), the two rules an agent must know (-o json for parseable output; mutations are dry-run until --execute), per-command-group usage, quota awareness, and a recommended agent workflow. Points at references/reference.md rather than dumping every flag. No MCP, nothing harness- or vendor-specific.
  • references/reference.md — the full command reference, generated from the live typer app.
  • assets/upload-sidecar.example.yaml — example sidecar mirroring the videos upload schema.

Keeping the reference in sync

docs/reference.md is built on the fly by mkdocs and git-ignored; the skill copy is checked in (it travels with the skill), so it can drift. Two guards, both from the same typer source as the docs:

  • uv run python scripts/build_skill_reference.py regenerates it.
  • tests/test_skill_reference.py re-renders and fails if the committed copy is stale, so the existing pytest CI job catches drift.

Docs

  • README gains an Agent skill section linking the skill and the sync docs.
  • New docs/agent-skill.md page (added to the mkdocs nav) explains usage and the sync workflow.

Verification

  • uv run pytest — 254 passed (includes the new drift test).
  • uv run ruff check / ruff format --check — clean.
  • uv run mkdocs build --strict — clean (no broken links).
  • Independent adversarial review cross-checked every command/flag/enum in SKILL.md against the typer dump and source, validated the frontmatter against the spec, and checked the sidecar against upload_pipeline.py: no defects.

Follow-up (not in this PR)

Listing on skills.sh is an external action: once merged, point the registry at the skills/ytstudio/ subpath.

jdwit added 2 commits June 17, 2026 17:15
Author a model- and harness-agnostic SKILL.md (open SKILL standard) at
skills/ytstudio/ that teaches any skill-aware agent to drive the CLI:
prerequisites, the dry-run/--execute and -o json rules, and per-group
usage with pointers rather than a full dump.

Bundle the full command reference at skills/ytstudio/references/reference.md,
generated from the live typer app via scripts/build_skill_reference.py and
guarded against drift by tests/test_skill_reference.py. Add an example upload
sidecar asset, a docs page, and README/docs links.

Closes #29
@jdwit jdwit merged commit e0d9bf4 into main Jun 17, 2026
7 checks passed
@jdwit jdwit deleted the feat/agent-skill branch June 17, 2026 15:43
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Publish a harness-agnostic ytstudio agent skill (SKILL.md)

1 participant