{"slug":"stetkeep","title":"stetkeep","description":"stetkeep stops Claude from refactoring code that's intentionally the way it is. Two pillars:\n\n1) XML-structured behavior protocols (BRAIN/CRAFT/PERF, ~1.8K tokens each) encode routing, refactor discipline, and measurement-first performance rules using the XML tagging pattern Anthropic's prompting guide recommends for reliable parsing.\n\n2) A 16-entry false-positive catalog of patterns Claude commonly mis-flags as anti-patterns: 1000-line constants files (not God Files, just data tables), V8-optimized .map().map().map() chains, below-fold <img> tags, useMemo that fixes reference-equality bugs. No other plugin in the directory ships this.\n\nEnforcement is mechanical, not prompted: permission deny-lists, PreToolUse hooks (safety-net.sh / .ps1), tool-scoped subagents that cannot Write new files, and path-scoped rules that auto-load CRAFT/PERF on src/** access. Honest 5-layer breakdown in ARCHITECTURE.md §1.\n\nSupply chain hardened: zero runtime dependencies, OIDC Trusted Publisher (no long-lived npm token), Sigstore SLSA v1 provenance on every tarball, GitHub Actions pinned by commit SHA.\n\nCoexists with existing CLAUDE.md and memory/ without overwriting. Distributed via npm (npx stetkeep install) and the self-hosted Claude Code plugin marketplace. MIT license.","developer":"CJ Kim","category":null,"reported_installs":null,"active":true,"first_seen_at":"2026-10-03T22:13:33.942Z","last_seen_at":"2026-11-07T00:08:29.147Z","last_changed_at":"2026-10-03T22:13:33.942Z","platform":"claude","sources":["Community"],"works_with":["Claude Code"],"marketplace_url":null,"repository_url":"https://github.com/chanjoongx/stetkeep.git","web_data":{},"github_data":{},"catalog_sources":{"community":[{"name":"stetkeep","source":{"sha":"434e7b337aff22e31964981053c89205e2330088","url":"https://github.com/chanjoongx/stetkeep.git","source":"url"},"homepage":"https://stetkeep.com","description":"stetkeep stops Claude from refactoring code that's intentionally the way it is. Two pillars:\n\n1) XML-structured behavior protocols (BRAIN/CRAFT/PERF, ~1.8K tokens each) encode routing, refactor discipline, and measurement-first performance rules using the XML tagging pattern Anthropic's prompting guide recommends for reliable parsing.\n\n2) A 16-entry false-positive catalog of patterns Claude commonly mis-flags as anti-patterns: 1000-line constants files (not God Files, just data tables), V8-optimized .map().map().map() chains, below-fold <img> tags, useMemo that fixes reference-equality bugs. No other plugin in the directory ships this.\n\nEnforcement is mechanical, not prompted: permission deny-lists, PreToolUse hooks (safety-net.sh / .ps1), tool-scoped subagents that cannot Write new files, and path-scoped rules that auto-load CRAFT/PERF on src/** access. Honest 5-layer breakdown in ARCHITECTURE.md §1.\n\nSupply chain hardened: zero runtime dependencies, OIDC Trusted Publisher (no long-lived npm token), Sigstore SLSA v1 provenance on every tarball, GitHub Actions pinned by commit SHA.\n\nCoexists with existing CLAUDE.md and memory/ without overwriting. Distributed via npm (npx stetkeep install) and the self-hosted Claude Code plugin marketplace. MIT license."}]},"manifest":{"url":"https://raw.githubusercontent.com/chanjoongx/stetkeep/434e7b337aff22e31964981053c89205e2330088/.claude-plugin/plugin.json","payload":{"name":"stetkeep","agents":["./agents/brain-router.md","./agents/craft-specialist.md","./agents/perf-specialist.md"],"author":{"url":"https://github.com/chanjoongx","name":"CJ Kim","email":"cj@chanjoongx.com"},"license":"MIT","version":"0.4.6","commands":["./commands/brain-scan.md","./commands/craft-audit.md","./commands/perf-audit.md"],"homepage":"https://stetkeep.com","keywords":["claude-code","xml-protocols","false-positive-catalog","safety-net","hooks","subagents","code-refactoring","performance","ai-safety"],"repository":"https://github.com/chanjoongx/stetkeep","description":"XML protocol framework + false-positive catalog for Claude Code. Mechanical guardrails against AI refactoring mistakes via hooks, tool-scoped subagents, and path-scoped rules."},"observed_at":"2026-11-06T10:49:39Z"}}