Help us improve
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
By mistakeknot
Token-efficient code reconnaissance for LLMs. Autonomous skills save 48-85% tokens via diff-context, semantic search, structural patterns, and symbol analysis. Includes MCP server for direct tool integration.
npx claudepluginhub mistakeknot/interagency-marketplace --plugin tldr-swintonSemantic code search - find code by meaning, not just text patterns
Get token-efficient context for recent changes (diff-first workflow)
Get LLM-ready context for a specific function or class
Structural code search using ast-grep tree-sitter patterns
Show the tldr-swinton quick reference guide for agents
```
All internal imports MUST use relative imports:
**FunctionInfo** (`modules/core/ast_extractor.py`):
The MCP server is the primary agent interface. 24 tools organized by category.
| Command | Description |
Use when starting any coding task: fix bugs, debug, implement features, refactor, write tests, review code, migrate, port, or explore a codebase. Also use when resuming a previous session, onboarding to a repo, or before reading code files in a new conversation. Provides diff-focused context that saves 48-73% tokens.
Use when asked to understand a codebase's architecture, explore an unfamiliar project, onboard to a new repo, identify which modules exist, find entry points, or get a bird's-eye view before diving into code. Also use when the user says 'what does this project do' or 'show me the structure'. Provides structural overview without reading individual files.
Sync interbench eval coverage with tldrs capabilities. Run when tldrs gains new formats, flags, or commands. Reads the tldrs manifest as ground truth and generates minimal targeted edits to 4 interbench files.
Use when auditing a codebase for semantic duplication - functions that do the same thing but have different names or implementations. Especially useful for LLM-generated codebases where new functions are often created rather than reusing existing ones.
Admin access level
Server config contains admin-level keywords
Uses power tools
Uses Bash, Write, or Edit tools
Share bugs, ideas, or general feedback.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
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.
Open-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.
Graph-first code intelligence for AI agents. SurrealDB knowledge graph + 52 MCP tools replace Read/Grep/Glob with deterministic graph traversal. 80–95% fewer tokens on code context. Rust-native, fully local.
Structural codebase indexing for efficient navigation. Reduces token consumption by 60-80% through targeted line-range reads instead of full file scans.
CodeAlive context engine for semantic code search and AI-powered codebase Q&A. Enables AI coding agents to understand entire codebases beyond just open files — search across all indexed repositories, trace cross-service dependencies, discover usage patterns, and get synthesized answers to architectural questions. Includes a lightweight code exploration subagent, authentication hooks, and multiple search modes (fast lexical, semantic, and deep cross-cutting). Works standalone or alongside the CodeAlive MCP server for direct tool access via the Model Context Protocol.
Precise local semantic code search via MCP. Indexes your codebase with Go AST parsing, embeds with Ollama or LM Studio, and exposes vector search to Claude through an MCP server — no cloud, no npm.
[DEPRECATED — use intervoice] Analyze your writing style and adapt Claude's output to sound like you. Replaced by intervoice, which reads one global multi-register profile instead of per-project glob-routed files.
Recursive AGENTS.md generator with integrated Oracle critique, CLAUDE.md harmonization, incremental updates, diff previews, and smart monorepo scoping. Cross-AI compatible.
Self-improving agent rig: codifies product and engineering discipline into composable workflows from brainstorm to ship. Compounds knowledge, generates domain agents, monitors its own docs, and surfaces conservative update drift. Orchestrates Claude, Codex, and Oracle through 6 agents, 52 commands, 19 skills, 0 MCP servers. Factory substrate: CXDB turn DAG, scenario bank with satisfaction scoring, evidence pipeline, agent capability policies. Companions: interspect, interphase, interline, interflux, interpath, interwatch, interslack, interform, intercraft, interdev, interpeer, intertest.
Multi-agent review and research with scored triage, domain detection, content slicing, intermediate finding sharing, and knowledge injection. 17 agents (12 review + 5 research), 7 commands, 1 skill (unified flux-drive with review/research modes), 2 MCP servers (exa, openrouter-dispatch). Companion plugin for Clavain.
Token efficiency benchmarking, session analytics, and API-equivalent cost analysis for agent workflows
Share bugs, ideas, or general feedback.
Token-efficient code analysis platform for LLMs and AI agents.
33 commands. 5 analysis layers. 13+ languages. 48–93% token savings.
| For Humans | For AI Agents | For Both |
|---|---|---|
| Explore unfamiliar codebases fast | Structured context that fits in a prompt | Semantic search by meaning, not text |
| Understand call chains and data flow | Token budgets to control cost | Diff-focused context for recent changes |
| Find dead code and architectural layers | Session tracking for multi-turn work | 5-layer analysis: AST → Call Graph → CFG → DFG → PDG |
tldrs isn't just extraction: it's full static analysis (control flow, data flow, program slicing), semantic search (embeddings + BM25 hybrid), and session-aware artifact storage, all designed to produce the smallest context an LLM needs to do its job.
# One-liner (recommended)
curl -fsSL https://raw.githubusercontent.com/mistakeknot/tldr-swinton/main/scripts/install.sh | bash
# Or via uv
uv pip install tldr-swinton
# Or manual
git clone https://github.com/mistakeknot/tldr-swinton
cd tldr-swinton && uv sync --extra semantic-ollama
tldrs structure src/ # Code structure (functions, classes, imports)
tldrs diff-context --project . # Context pack for your recent changes
tldrs index . && tldrs find "authentication logic" # Semantic search
tldrs context main --project . # Call-graph context around a symbol
tldrs --version # 0.7.0
tldrs doctor # Check optional tools (type checkers, linters)
All 33 top-level commands grouped by use case. Each command supports --help for full options.
Get oriented in a codebase or understand recent changes.
| Command | Description |
|---|---|
diff-context | Diff-focused context pack (merge-base → HEAD) |
context | Call-graph context around a symbol (--depth, --budget, --format) |
find | Semantic code search (requires tldrs index first) |
structural | Structural search using ast-grep patterns |
search | Regex pattern search across files |
quickstart | Quick reference guide for AI agents |
Understand file and project organization.
| Command | Description |
|---|---|
structure | Code structure: functions, classes, imports (codemaps) |
extract | Full file analysis as JSON |
tree | File tree |
imports | Parse imports from a source file |
importers | Reverse import lookup: find all files importing a module |
Static analysis beyond surface-level extraction.
| Command | Description |
|---|---|
cfg | Control flow graph for a function |
dfg | Data flow graph for a function |
slice | Program slice: what affects a specific line |
calls | Cross-file call graph |
impact | Reverse call graph: find all callers of a function |
dead | Find unreachable (dead) code |
arch | Detect architectural layers from call patterns |
change-impact | Find tests affected by changed files |
diagnostics | Type checker and linter diagnostics |
distill | Compress context for sub-agent consumption |
hotspots | Most frequently used symbols across sessions |
Embedding-based search: find code by meaning, not keywords.
| Command | Description |
|---|---|
index | Build or update the semantic index (--backend ollama|sentence-transformers|auto) |
find | Query the index (e.g., tldrs find "error handling patterns") |
semantic | Low-level semantic search subcommands |
Daemon, caching, and health checks.
| Command | Description |
|---|---|
daemon | Daemon management (start, stop, status) |
warm | Pre-build call graph cache for faster queries |
prebuild | Precompute context bundle for current HEAD |
doctor | Check and install diagnostic tools |
presets | List flag presets and their expansions |
manifest | Machine-readable JSON of all tldrs capabilities |
Persistent storage for agent workflows.
| Command | Description | Subcommands |
|---|---|---|
vhs | Content-addressed store for tool outputs | 9 (put, get, cat, has, info, rm, ls, stats, gc) |
wb | Agent workbench: capsules, decisions, hypotheses | 18 (capture, show, decide, hypothesis, link, ...) |
bench | Benchmarking harness for agent improvements | 4 (run, list, report, compare) |
/plugin marketplace add mistakeknot/interagency-marketplace
/plugin install tldr-swinton