Index and query codebases via a persistent SQLite knowledge graph.
npx skills add https://github.com/kkudumu/feed-the-machine --skill ftm-map
Please help me install this Agent Skill. Skill: ftm-map Source: https://github.com/kkudumu/feed-the-machine/tree/main/ftm-map Command: npx skills add https://github.com/kkudumu/feed-the-machine --skill ftm-map
Claude Code often forgets project context across conversations, making end-to-end automation tedious. ftm-map provides a persistent knowledge graph of your codebase that captures files, symbols, and references to enable coordinated workflows and better decision-making.
Run the bootstrap workflow to index your project, then query for blast radius, dependencies, or context.
High-intent search queries and answers about installing and using this skill.
A persistent code graph stores your codebase structure in a 5-table SQLite schema with FTS5 full-text search, capturing files, symbols, and references. It enables blast radius, dependency chain, and context selection queries to reason about code impact and support better decision-making across conversations.
Query the code graph for blast radius directly after running the bootstrap indexing workflow. The query functionality uses mapped symbol edges and file edges to identify affected components, helping you reason about the structural impact of code modifications.
Yes, ftm-map supports incremental indexing to keep your code map updated with code changes without needing a full rebuild. This maintains an accurate persistent knowledge graph efficiently as your project evolves over time.
Yes, ftm-map uses tree-sitter-language-pack for symbol extraction and sqlite-vec for vector storage within a 5-table SQLite schema. This combination enables structural code indexing and FTS5 full-text search over symbols, requiring Python 3.10 or higher.
Generate INTENT.md and ARCHITECTURE.mmd visual files from the indexed code graph to document project structure and dependencies. These outputs provide visual representations of architecture and dependency chains, complementing the structural query capabilities for impact analysis.
Claude Code loses project context across conversations because it lacks persistent codebase awareness. ftm-map solves this by building a persistent knowledge graph that captures files, symbols, and references, enabling coordinated end-to-end automation workflows.
nnkiennCompute blast radii of AST graph changes for token-efficient code review.
MadAppGangTrace AST call chains to locate root causes of code defects.
MadAppGangAnalyze codebases with tree-sitter AST parsing and PageRank symbol ranking.
IoncradeBuilds and queries knowledge graphs from source code using Python libraries.
