ninja-terminals 2.0.0 → 2.1.0

package/orchestrator/tool-registry.md

@@ -1,96 +0,0 @@
-# Tool Registry
-
-> This file is SELF-EVOLVING. The orchestrator updates it based on measured results.
-> Every change must be logged in evolution-log.md with evidence.
-> Last updated: 2026-03-23 (initial inventory)
-
-## Rating Scale
-- **S** — Essential. Use every session. Proven high-value.
-- **A** — Very useful. Use frequently. Measurably improves outcomes.
-- **B** — Useful in specific contexts. Worth having.
-- **C** — Marginal. Rarely needed. Consider removing.
-- **?** — Not yet rated. Needs testing.
-
-## Active Tools (Currently Installed & Working)
-
-### MCP Servers (Project — .mcp.json)
-
-| Tool | Purpose | Rating | Notes |
-|------|---------|--------|-------|
-| postforme | Video render, social publish, Meta ads | S | Core tool for PostForMe project |
-| studychat | RAG KB, DMs, C2C messaging | A | Knowledge persistence, user comms |
-| gmail | Email search, read, attachments | B | Occasional use for research |
-| chrome-devtools | Browser automation, screenshots | A | Verification, web interaction |
-| netlify-billing | Deploy status, billing | B | Monitoring only |
-| render-billing | Deploy status, billing | B | Monitoring only |
-
-### MCP Servers (Global — ~/.claude/settings.json)
-
-| Tool | Purpose | Rating | Notes |
-|------|---------|--------|-------|
-| builder-pro-mcp | Code review, security scan, auto-fix | A | Quality gates |
-| gkchatty-production | Knowledge base | C | DO NOT USE unless David requests |
-| atlas-architect | Blender 3D automation | B | Niche — only for avatar project |
-| claude-in-chrome | Browser automation (alternative) | A | Used by orchestrator for visual supervision |
-
-### Claude Code Built-In Features
-
-| Feature | Purpose | Rating | Notes |
-|---------|---------|--------|-------|
-| Agent tool (subagents) | Parallel research, isolated tasks | A | Use for research-heavy work |
-| Glob/Grep/Read | File search and reading | S | Core workflow |
-| Edit/Write | File modification | S | Core workflow |
-| Bash | Shell commands | S | Builds, tests, git |
-| WebSearch/WebFetch | Internet research | A | Tool discovery, docs |
-| /compact | Context management | A | Use proactively, not just at limit |
-| /clear | Session reset | B | Between unrelated tasks |
-| Extended thinking (ultrathink) | Deep reasoning | ? | Need to test and measure impact |
-| Git worktrees (--worktree) | Isolated parallel branches | ? | Need to test |
-| Headless mode (-p) | Scripted automation | ? | Need to test for CI/metrics |
-| Custom slash commands | Reusable workflows | ? | Need to evaluate |
-
-### Skills (Available in Claude Code)
-
-| Skill | Purpose | Rating | Notes |
-|-------|---------|--------|-------|
-| /scout-plan-build | Full feature workflow | ? | Need to test on a real feature |
-| /review | Code review | ? | Need to compare vs builder-pro review |
-| /test | Testing framework | ? | Need to evaluate |
-| /scan | Security audit | ? | Need to compare vs builder-pro security_scan |
-| /build | Project builder | ? | Need to evaluate |
-| /bmad-pro-build | Full SDLC with RAG | B | For large features (1+ hour, 3+ files) |
-
-## Candidates (Discovered, Not Yet Installed)
-
-### High Priority (Test Soon)
-
-| Tool | What It Does | Source | Security Status |
-|------|-------------|--------|-----------------|
-| Playwright MCP | Browser testing via accessibility tree, more token-efficient than screenshots | @playwright/mcp (official) | Trusted — Microsoft maintained |
-| Sentry MCP | Query production errors, stack traces | Official | Trusted — if we use Sentry |
-| LSP plugins | Real-time type errors after every edit | github.com/boostvolt/claude-code-lsps | Needs review |
-| Hooks (PreToolUse) | Auto-format, block dangerous commands | Built into Claude Code | Native — no install needed |
-
-### Medium Priority (Research More)
-
-| Tool | What It Does | Source | Security Status |
-|------|-------------|--------|-----------------|
-| code-review-mcp | Multi-LLM code review | github.com/praneybehl/code-review-mcp | Needs scan |
-| Mighty Security Suite | MCP server security scanning | github.com/NineSunsInc/mighty-security | Needs review |
-| Superpowers framework | Composable skills, TDD, review subagent | github.com/obra/superpowers | Needs review |
-| DSPy (prompt optimization) | Automatic prompt compilation | github.com/stanfordnlp/dspy | Academic — trusted |
-
-### Low Priority (Interesting But Not Urgent)
-
-| Tool | What It Does | Source |
-|------|-------------|--------|
-| Ruflo (Claude Flow) | 60+ agent swarm coordination | github.com/ruvnet/ruflo |
-| OpenClaw | Self-writing skills, 10K+ community skills | github.com/openclaw/openclaw |
-| AutoResearch skill | Overnight prompt optimization loop | github.com/uditgoenka/autoresearch |
-| MCPSafe.org | CI/CD MCP security checks | mcpsafe.org |
-
-## Retired Tools
-
-| Tool | Why Retired | Date |
-|------|------------|------|
-| (none yet) | | |