We build and maintain developer tools in the open — free to use, free to fork, and better with every contribution.
A native C++ engine that turns codebases into knowledge graphs for persistent memory. Scan a project, extract its structure into a deterministic graph, and query it through a local daemon or MCP server — so coding agents navigate your code instead of grepping it.
Example placeholder — a VS Code extension surfacing CGraph queries inline while you read code.
Give it a diff and it answers two questions: which tests does the change actually reach, and what is its transitive blast radius? Deterministic selection from a CGraph code graph, fail-open by design — replay-verified across TypeScript, Python, Go, C/C++, and Java, with Rust and Kotlin advisory, including on CGraph itself, which dogfoods it on every PR. Selections can even cross repos through a seam spec, so a schema change on one side surfaces the other service’s call sites. Ships as a CLI, a GitHub Action that builds its own graph and comments blast radius on every PR, and an MCP server whose blastline_check tool lets a coding agent verify it really did update every call site.
Example placeholder — each new public repo becomes a card here with its description, language, and stars.
A call graph resolves obj.method() only when something proves what obj is. The five tiers of evidence, and why Java call sites supply it while Rust ones do not.
git show --name-only prints nothing for a merge commit. That one fact turned our dependency oracle into a machine that cleared every miss on zero evidence.
CGraph builds a graph of itself on every PR and lets Blastline pick the tests. Eleven PRs in: six real subsets, five "run everything" — all five from one file.