Cross-language-safe code-intelligence MCP for AI agents -- 13 languages, family-gated call graph (CodeGraph 745 vs TSA 6 cross-language mis-wires, v1.21.0 same-session -- ~124x by count, ~390x by rate; re-verifying vs current develop). Run miswire-audit on your repo. 8 facade tools, TOON output, 100% local. Python.

MOVED to github.com/professorpalmer/marionette -- this repo is archived. Marionette: a desktop AI coding harness over Puppetmaster durable state.

Persistent semantic memory for AI agents — 100% local on Apple Silicon (MLX) or Linux/Ubuntu (CPU). Markdown source of truth, sqlite-vec + BM25 hybrid search, a codegraph-backed knowledge graph, MCP server + CLI. No cloud, no keys.

Local-first memory and workflow plugin for AI coding agents: Codex, Claude Code, agent preferences, project lessons, and delivery rules.

MCP server that gives Claude Code the rationale behind a symbol before editing - synthesized from git history, PRs, and call-graph context.

AEGIS 2.0: multi-agent code repository analysis system with CodeGraph, source-grounded RAG, and agent handoff for the Volcano Cup intelligent agent competition

Local-first MCP-style code memory with file search, Python symbols, import graph, and repo context packs.

Local code graph for AI coding agents. Indexes your repo into Kuzu + SQLite FTS, exposes 30+ MCP tools to Claude Code, Cursor, Codex, and Gemini. Federates across sibling repos. Tool calls cost zero model tokens.

CTX, TVL, RLM, CodeGraph, durable agent memory and universal cross-agent handoff for lower context/token usage