By mertcanaltin
Compress code files into health-aware intermediate representations (IR) to cut LLM tokens—while proactively detecting security issues, code smells, and codebase decay from git history.
Use at the start of every session — initializes Composto proactive monitoring and Health-Aware IR capabilities
Generate Health-Aware IR for any file — compressed code representation with health signals. Use when you need to understand or share file context with minimal tokens.
Scan the codebase for security issues, debug artifacts, and code smells. Use before starting work or when asked to check code quality.
Analyze codebase health trends — find hotspots, decay signals, and inconsistencies. Use before refactoring or when investigating recurring bugs.
Admin access level
Server config contains admin-level keywords
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
npx claudepluginhub mertcanaltin/composto --plugin compostoOpen-source, local-first Claude Code plugin for token reduction, context compression, and cost optimization using hybrid RAG retrieval (BM25 + vector search), reranking, AST-aware chunking, and compact context packets.
Token optimization for Claude Code. Automatic tool output compression (40-60% reduction), token meter in statusline, auto-compact at 70% context, structured output formats, self-building project wiki, URL ingestion, Karpathy compile pass, and semantic lint.
Governor: always-on compact professional output, telemetry, context slimming, tool-output filtering, prompt guidance, and drift guardrails for Claude Code Max users.
Structural codebase indexing for efficient navigation. Reduces token consumption by 60-80% through targeted line-range reads instead of full file scans.
Fast and token-friendly code reading for AI coding agents. Symbol-aware MCP tools that replace cat/grep with ~85% fewer tokens, sub-millisecond search, and a raw fallback that preserves cat/grep parity byte-for-byte.
Claude Code context optimizer. Builds a local static repo graph, injects ranked file:line candidates through a UserPromptSubmit hook, and adds a SessionStart prewarm brief. No embeddings, no server, no telemetry. Measured live A/B: 40.9% aggregate token reduction across 36 Claude Code calls. v2.10 structural slicing: a PreToolUse hook intercepts whole-file reads and hands Claude the file's outline (symbols + exact line ranges from the graph) so it reads the 40 lines it needs, not 800 that get re-sent every turn. Receipts (/savings) measure both sources — searches avoided and big reads sliced — causally from your transcript.