Repository navigation
docs(skills): make watch description state when to invoke - #154
Open
thomaswillner wants to merge 3 commits into
Open
thomaswillner wants to merge 3 commits into
thomaswillner wants to merge 3 commits into
Conversation
Merged
1 of 3 tasks
|
Thanks @thomaswillner — this is in the community fork frinsen/claude-video (main, after v0.3.0): merged as authored — the description now says when to invoke, which is what hosts match on. You're credited in AUTHORS.md; the disposition of every open upstream PR is in UPSTREAM.md, and the whole set is offered back here in #237. |
Conflict was the whole frontmatter block of skills/watch/SKILL.md, where both sides changed adjacent lines. Kept both intents: - upstream's version bump 0.2.0 -> 0.3.0. This is load-bearing, not cosmetic: test_release_versions_match_canonical_skill asserts SKILL.md's version equals the version in .claude-plugin/plugin.json and .codex-plugin/plugin.json, both of which are already 0.3.0 on main. - this branch's description, which states WHEN to invoke the skill rather than restating its mechanism. That is the entire point of the PR. The gemini/WhisperX capabilities upstream added to the description are not lost: they are documented at length in the body (23 references), which merged cleanly and is where test_skill_documents_and_bundles_the_gemini_engine looks for them. Keeping them out of the description is the change under review, not an omission. Verified: no conflict markers repo-wide; frontmatter parses as YAML (description 205 chars, spec max 1024); net diff against upstream/main is the single description line; both tests that read SKILL.md pass. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
main advanced twice since the last merge (de8b5b5 -> 7f288b4). The relevant commit is 4576273, "keep SKILL.md frontmatter within the Agent Skills spec", which restructured the frontmatter: top-level version, argument-hint, homepage, repository, author and user-invocable were dropped, version moved under metadata, and a new test now asserts the top-level key set is a subset of name/description/license/allowed-tools/metadata/compatibility. That restructure left the description untouched, so this PR is not superseded -- but the previous resolution's top-level version key no longer exists upstream. Resolution therefore takes upstream's frontmatter shape wholesale and applies only this branch's description to it. Verified against the tests as they now stand, not as they were: - top-level keys are {allowed-tools, description, license, metadata, name}, a subset of the allowlist - metadata.version 0.3.1 matches both .claude-plugin and .codex-plugin plugin.json (the version regex is now indented for metadata) - the gemini needles still resolve in the body - description 205 chars, spec max 1024 - no conflict markers repo-wide; net diff against upstream/main is the single description line Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The
watchskill description currently spends most of its length on implementation mechanism — yt-dlp, ffmpeg, the Whisper fallback — rather than on when an agent should invoke it.Per the Agent Skills specification, the description is the only thing an agent sees before deciding whether to load a skill. Naming the tooling does not help that decision; naming the user intents does. The mechanism is already documented in the skill body, which loads on invoke.
Before
After
Shorter, and it now lists the distinct intents that should trigger it. One line changed; no behavioural change.