feat: add .github/copilot-instructions.md#2088
feat: add .github/copilot-instructions.md#2088tmeschter wants to merge 1 commit intomicrosoft:mainfrom
Conversation
Covers versioning (NBGV), building (Gulp pipeline), running tests (validators, unit/trigger/integration), adding new skills, CI/CD workflows, PR guidelines, and repo layout. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
There was a problem hiding this comment.
Pull request overview
Adds a repository-level instructions file intended to give AI agents enough context to contribute effectively (layout, build/versioning rules, validation/test commands, and skill authoring workflow).
Changes:
- Introduces
.github/copilot-instructions.mdwith repo layout + build/versioning guidance. - Documents local validation commands (tokens/frontmatter/references/tests) and maps CI checks to local equivalents.
- Provides a checklist-style “Adding a new skill” workflow and lists available repo-local agent skills.
Show a summary per file
| File | Description |
|---|---|
.github/copilot-instructions.md |
New agent-facing repo instructions covering layout, build/versioning, validators/tests, CI mapping, and skill-authoring workflow. |
Copilot's findings
Comments suppressed due to low confidence (3)
.github/copilot-instructions.md:171
- The "Plugin Version Check" CI row lists a behavioral reminder rather than an actionable local command. Since the repo has a dedicated script for this check, consider updating the local equivalent to something like running
npm ciinscripts/andnpm run checkPluginVersionPr(at least whenplugin/**/plugin.jsonchanges).
| Plugin Version Check | `plugin.json` versions remain `0.0.0-placeholder` | Ensure you never edit version fields |
| Skill Tests | Unit and trigger tests for changed skills | `cd tests && npm test` |
.github/copilot-instructions.md:159
- The guidance about using folded YAML (
>-) for long frontmatter descriptions conflicts with existing repo skill-authoring guidance: descriptions should be inline double-quoted strings, and>-is explicitly discouraged/unsupported (skills.sh incompatibility). Please update this line to match the repo convention (usedescription: "..."even when long, and keep it within the 1024-char limit).
- Use progressive disclosure: frontmatter → SKILL.md → references
- Descriptions over 200 chars in frontmatter must use folded YAML (`>-`)
- Markdown links must not escape the skill directory (validated by `npm run references`)
.github/copilot-instructions.md:168
- The local equivalent for the ESLint CI job is incomplete: CI runs typecheck+lint for both
tests/andscripts/, but the table only lists thetests/commands. Update the local equivalent to include runningnpm run lint/npm run typecheckinscripts/as well (or point to a single root command if preferred).
| ESLint | Linting + typechecking `scripts/` and `tests/` | `cd tests && npm run lint && npm run typecheck` |
| Token Analysis | Token counts and limits for markdown files | `npm run tokens check` |
- Files reviewed: 1/1 changed files
- Comments generated: 1
| ### Unit and Trigger Tests (from tests/) | ||
|
|
||
| ```bash | ||
| cd tests | ||
| npm install | ||
| npm test # Run all tests | ||
| npm test -- --testPathPatterns=<skill-name> # Run tests for a single skill |
There was a problem hiding this comment.
The "Unit and Trigger Tests (from tests/)" section implies you can run cd tests && npm test directly, but the test suite reads skills from output/skills/ (git-ignored) and will fail unless you build first. Consider adding a note here (or prepend npm run build from repo root) so the instructions are correct for a fresh checkout.
This issue also appears in the following locations of the same file:
- line 157
- line 167
- line 170
Description
Adds a
.github/copilot-instructions.mdfile that gives AI agents (Copilot, Cursor, Claude, etc.) the context they need to work effectively in this repo.Covers:
npm run buildpipeline0.0.0-placeholderconvention, never manually edit versions.token-limits.json.github/skills/table with invocation guidanceEvery section was reviewed for agent relevance — human-only content (prerequisites, local Copilot CLI usage, marketplace publishing details) was excluded to keep the file focused and token-efficient.
Checklist
cd tests && npm test)npm run test:skills:integration -- <skill>)USE FOR/DO NOT USE FOR/PREFER OVERclauses: confirmed no routing regressions for competing skillsRelated Issues
N/A — new file, no skill files changed.