Refine repo context boundaries and GitHub workflow guidance#20
Refine repo context boundaries and GitHub workflow guidance#20
Conversation
AGENTS.md now generates PR description templates and issue update patterns dynamically from actual GitHub history via gh api. Learned workflow rules from session artifacts are injected directly into the workflow section. Repos with richer history get richer agent guidance — the virtuous cycle. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
There was a problem hiding this comment.
Pull request overview
This PR enhances dev.kit agent so AGENTS.md can dynamically learn guidance from (1) prior local session lesson artifacts and (2) live GitHub repo/user history, aiming to produce repo-specific PR/issue update guidance on each run.
Changes:
- Add GitHub PR/issue comment pattern detection utilities (via
gh api+jq+awk) to support dynamic guide generation. - Inject “Learned from prior sessions”, “PR description guide”, and “Issue update guide” sections into generated
AGENTS.md. - Extend
dev.kit learnflow template rendering to cover additional template IDs; bump version and update changelog.
Reviewed changes
Copilot reviewed 5 out of 5 changed files in this pull request and generated 4 comments.
Show a summary per file
| File | Description |
|---|---|
| package.json | Bumps package version to 0.7.0. |
| lib/modules/learning_sources.sh | Adds GitHub-history and lesson-artifact extraction functions used to generate dynamic guides. |
| lib/commands/agent.sh | Writes new dynamic AGENTS.md sections sourced from lessons and GitHub history. |
| lib/commands/learn.sh | Adds missing flow-template cases so templates render with backtick-wrapped labels. |
| changes.md | Documents the 0.7.0 release changes. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
…cation - Pass sample_size through to jq instead of hard-coded .[:8] slice - Update AGENTS.md header to reflect all three sources - Cap best-example bodies at 60 lines to limit exposure Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 26 out of 28 changed files in this pull request and generated 4 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 38 out of 40 changed files in this pull request and generated 4 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
fqjony
changed the title
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 41 out of 43 changed files in this pull request and generated 2 comments.
Comments suppressed due to low confidence (1)
lib/commands/agent.sh:180
- The header comment above
dev_kit_agent_write_agents_mdsays “All content is derived from context.yaml”, but the function now also injects workflow/practices from config catalogs plus GitHub history and lesson artifacts. Update the comment to match the actual inputs so maintainers don’t accidentally break the non-context dependencies.
# Write AGENTS.md — the repo's execution contract for AI agents.
# All content is derived from context.yaml. Agents operate from this file,
# not from filesystem discovery.
dev_kit_agent_write_agents_md() {
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Summary
.rabbit/context.yamlfocused on repo facts, tracing, commands, and gaps instead of mixing in workflow and practice guidanceAGENTS.mdworkflow and engineering guidance directly from config catalogs and learned session patterns, with stronger GitHub-first behavior0.8.0and document the release inchanges.mdVerify
bash tests/suite.sh --only core,learndev.kit repodev.kit agent