Skip to content

PROJECT_STATUS.md

Latest commit

 

History

History
145 lines (93 loc) · 7.5 KB

PROJECT_STATUS.md

File metadata and controls

145 lines (93 loc) · 7.5 KB

Augmentedcode Configuration - Project Status

Last Updated: 2026-04-02 Overall Status: 🟢 100% Complete - Shared RTK rule path fixed for all tools

Recent Changes

2026-04-02: Shared RTK rule include fixed for Cursor/Codex/Claude ✅

  • Issue: .agents/rules/base.md includes @RTK.md, but no canonical RTK.md existed in .agents/rules/.
  • Fix: Added .agents/rules/RTK.md as the shared, tool-agnostic RTK guidance file so base-rule consumers resolve the include consistently.
  • Impact: Cursor and Codex now inherit RTK guidance from the same canonical include path used by shared base rules.

2026-03-27: Skills library audit against skill-foundry criteria ✅

Full audit of 74 skills in .agents/skills/ across three phases:

Phase 1 (High-impact):

  • Improved descriptions for 4 synced skills (creating-hooks, event-modeling, bdd-with-approvals, refactoring) in discovery index; SKILL.md frontmatter changes need upstreaming to skill-factory
  • Added 21 native skills to .agents/docs/skill-factory-skills.md (new "Native skills" section)
  • Added SKILL.md symlink at .agents/skills/skill-foundry/ root for standard discovery

Phase 2 (Governance):

  • Created catalog-engineering.yaml with 50+ engineering skills (category, pattern, overlap_with, notes)
  • Documented overlap disambiguation for refactoring, testing, and work-splitting skill clusters
  • Standardized reference material for native skills (test-doubles-first, planning-with-files)

Phase 3 (Lifecycle):

  • Assessed retirement candidates (pbt-pragmatic-adoption, creating-hooks, writing-statuslines) -- kept as monitor with 60-day review
  • Evaluated ab-test-designer/trustworthy-experiments merge -- keep separate (synced upstream)

Plan: .cursor/plans/skills_library_audit_affb1665.plan.md

2026-03-27: Product management skills — skill-foundry catalog and docs index ✅

  • Catalog: .agents/skills/skill-foundry/agents/catalog-product-management.yaml — 25 skills from pmprompt/claude-plugin-product-management with category, pattern, overlap_with, and sync provenance (see skills-lock.json).
  • Discovery: .agents/docs/skill-factory-skills.md retitled to a general skills index; added Product management table mirroring each skill’s description frontmatter.

2026-03-26: Stop tracking Claude runtime notification state ✅

  • Issue: .claude/notifications/last_stop_notification remained in the index after .gitignore gained .claude/notifications/; ignored paths do not untrack existing files.
  • Fix: git rm --cached .claude/notifications/last_stop_notification so the file stays local-only and per-session; directory remains ignored.

2026-03-26: Skill-foundry vs repo skills — implementation plan drafted 📋

  • Research: thoughts/shared/research/2026-03-26-skill-foundry-vs-repo-skills.md (governance gap vs minimal SKILL.md contract).
  • Plan: thoughts/shared/plans/2026-03-26-skill-foundry-repo-alignment.md — document two-layer model, add library-catalog.json, validate filesystem vs catalog, keep skill-foundry bundle catalog separate; skill-foundry directory has no top-level SKILL.md (bundle exception encoded in validator spec).

2026-03-19: Base Rule Context Split Completed ✅

Split the rulebook into a universal base plus contextual modules:

  • base.md: Trimmed to universal rules and added contextual loading for Python and Makefile repositories.
  • python-project.md and makefile-project.md: Added canonical contextual rule modules for Python and Makefile projects.
  • Cursor wrappers: Aligned python-dev.mdc, added makefile-dev.mdc, and updated use-base-rules.mdc to point at the canonical modules.

2026-03-18: Self-Improvement Mechanisms Hardened ✅

Closed 6 gaps in recursive self-improvement infrastructure:

  • base.md: Added <!-- version: 1.1 --> header, fixed broken path to ai-feedback-learning-loop.md, added §16 Periodic Self-Audit.
  • ai-feedback-learning-loop.md: Broadened scope from rules-only to cover skills, workflows, and commands. Added PROJECT_STATUS.md refresh to step 6.
  • Root shims: Converted AGENTS.md, GEMINI.md, CLAUDE.md from copies to symlinks → .agents/rules/base.md.

2026-03-07: XP Skills Tool-Awareness Follow-up ✅

Tool-awareness documentation so every coding tool knows how to use skills:

  • base.md: Added §9 Skills (Canonical Location and Use) — .agents/skills/, trigger-based use, skill format.
  • create-skill: Storage Locations row for canonical shared path .agents/skills/skill-name/ and symlink exposure.
  • Research: 2026-03-06-commands-movable-as-skills.md updated — canonical .agents/skills/ and open question "where XP skills should live" resolved.

Reference: Validation report thoughts/shared/research/2026-03-07-VALIDATION-REPORT-xp-skills-tool-awareness.md.

2026-03-06: Config Deduplication via Symlinks ✅

Completed migration from copy-based sync to symlink-based configuration:

  • Eliminated: ~50 duplicated files across ~/.cursor/
  • Consolidated: All skills moved to repository (skills-cursor)
  • Symlinked: ~/.cursor/, ~/.claude/, and root configs now point to repo
  • Repurposed: sync-cursor-config.sh → setup-symlinks.sh

Impact: Single source of truth established. All config edits in any tool propagate to repo automatically.

Files: See implementation plan at thoughts/shared/plans/2026-03-06-config-deduplication-symlinks.md

Executive Summary

Component Status Progress Blocking
Cursor rules baseline ✅ Complete 100% -
Workflow/rule documentation 🟡 In Progress 90% -
Skills catalog ✅ Complete 100% -

Current Readiness: 🟢 Ready - Configuration repo is usable and actively maintained.

✅ Completed Components

Test Doubles Skill (2026-02-19)

  • Added project skill at .cursor/skills/test-doubles-first/.
  • Added SKILL.md with decision tree favoring fake/stub/spy before mock.
  • Added examples.md with minimal Python and TypeScript templates.
  • Added usage.md with trigger prompts and copy-paste templates.
  • Added language-specific quick chooser tables for Python and TypeScript dependency types.
  • Added anti-pattern -> replacement table to speed up test review and refactoring decisions.
  • Added PR review comment templates for fast, consistent code review feedback.
  • Added quick.md single-page fast reference for test-double decisions and review.
  • Reduced overlap between usage.md and quick.md to keep quick reference concise and detailed templates centralized.
  • Added SKILL.md quick navigation section for faster in-file discovery.
  • Reworked examples.md to Jest-first JavaScript/TypeScript patterns (fake, stub, spy, contract-focused mock).

🚧 In Progress

  • Expanding reusable Cursor skills for common engineering workflows.

📋 Next Steps

  1. Upstream improved descriptions for synced skills (creating-hooks, event-modeling, bdd-with-approvals, refactoring) to skill-factory.
  2. Upstream reference file standardization (REFERENCE.md/reference.md -> references/) to skill-factory for synced skills.
  3. Validate the test-doubles skill with real test-writing prompts.

🐛 Known Issues

  • Synced skill description improvements in discovery index may drift from SKILL.md frontmatter on next skill-factory sync.

📝 Notes

  • Skill content is cross-platform and uses forward-slash paths.
  • Skill intentionally encourages behavior/state assertions over interaction-heavy mocking.