feat: global setup and feature pipeline improvements#27
Merged
Conversation
Changes setup flow from project-level to global ~/.claude folder. Updates help text and removes project-specific instructions. Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Implements /aiworkers:init command that: - Scans project for architectural patterns, frameworks, conventions - Interactive validation loop with user confirmation - Generates patterns.md and architecture.md in target project - Updates target project CLAUDE.md with imports - Enforces mandatory architectural constraints Co-Authored-By: Claude Sonnet 4 <noreply@anthropic.com>
Remove stale imports/links to the deleted src/rules/jsx-style.md from CLAUDE.md and the rn-component architecture reference. Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Split the monolithic reviewer into spec-reviewer (PRD compliance) and code-reviewer (bugs/security). Add spec-writer to convert grill output into a product PRD. Refocus planner on producing a technical PRD from the product PRD instead of planning from a raw description. Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Add Phase 0.5 (grill) and 0.6 (spec) so requirements are locked into a product PRD before planning. Planner now emits a technical PRD. Replace the single reviewer with parallel spec + code reviewers, and route PRD/plan/review artifacts through .claude/tmp/. Include a new architecture reference doc. Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Replace security-violating bash find commands with Glob tool instructions throughout Phase 1 scanning logic. Co-Authored-By: Claude Sonnet 4 <noreply@anthropic.com>
Replace direct bash grep commands with Grep tool instructions in State Management and Enforcement sections. Co-Authored-By: Claude Sonnet 4 <noreply@anthropic.com>
Replace bash exit code pattern with Grep and Edit tools for checking and appending imports. Co-Authored-By: Claude Sonnet 4 <noreply@anthropic.com>
Use Glob to find first 5 files then Read for import/export pattern detection instead of reading all component files. Co-Authored-By: Claude Sonnet 4 <noreply@anthropic.com>
After 3 failed validations, provide free-text description option via AskUserQuestion for user to specify manually. Co-Authored-By: Claude Sonnet 4 <noreply@anthropic.com>
Expand rule 9 to explicitly state waiting for response before next call and prohibit parallel invocations. Co-Authored-By: Claude Sonnet 4 <noreply@anthropic.com>
Replace release-it + comment-triggered deploy with semantic-release using workflow_dispatch and OIDC trusted publisher (no NPM_TOKEN). Enable npm provenance. Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Default baseDir to os.homedir() so the test suite can sandbox it. Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
…/AIWorkers into feat/global-setup
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Move architecture.md from references/ subdirs to docs/ for better discoverability by humans. Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
References now point to docs/ directory. Remove old files from references/ subdirs. Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Update /feature and /rn-component to clarify that agents should talk to users, not spelunk codebases. Only do shallow checks for trivial facts. Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Clarify that grill-me interviews for product understanding, not code exploration. Only do shallow lookups for trivial facts. Form complete picture through dialogue. Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Merge Phase 0.5 (GRILL) and Phase 0.6 (SPEC) into a single Agent 0 that both grills the user and writes the PRD in real time. Simplify grill-me skill to conversation-only (no code lookups). Reduces orchestrator complexity and improves user experience by combining requirement gathering with spec writing in one conversational flow. Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Emphasize that agents write outputs to files and return only summaries, rather than inlining full content in responses. Simplify agent instructions to reference persona files instead of inlining. Fix typo on technical PRD output path. Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
BYOK Configuration Required! 🔑Your plan requires a Bring Your Own Key (BYOK) configuration to perform code reviews. Please configure your API keys in Settings > BYOK Configuration. |
Contributor
|
🎉 This PR is included in version 0.4.0 🎉 The release is available on: Your semantic-release bot 📦🚀 |
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
2 participants
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.
🚀 Summary
Improves the
/featuremulti-agent pipeline for token efficiency and restructures the global setup to install to~/.claude. Agents now hand off via files instead of returning full content inline, significantly reducing orchestrator context bloat.Type of change
📝 Changes
git diffitself instead of receiving the diff injected inline.claude/tmp/review-<slug>.md; fixer reads it directlyinline its full contentinstruction from all agent spawns~/.claudeinstallation via symlinks (idempotent)updateandunsetupsubcommands🧪 Testing
🎟️ Related Tickets