TraceDecay CLI
A local code graph answers agent questions about symbols and callers, so they reread whole files far less often.
- 86 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- A background service and agent hooks run on your machine and read the whole repository.
- The install runs a downloaded script, and the edit tool can change project files.
Install
- Works in
Manual install
curl -fsSL https://raw.githubusercontent.com/ScriptedAlchemy/tracedecay/master/install.sh | bashThe first README install command for Linux and Apple silicon macOS. On Windows the archive comes from the release page.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- A background service and agent hooks run on your machine and read the whole repository.
- The install runs a downloaded script, and the edit tool can change project files.
This is third-party code. Review the repository files before installing.
What it does
TraceDecay builds a semantic graph of a repository on your machine. An agent queries it through the MCP tools `tracedecay_context`, `tracedecay_search`, `tracedecay_callers` and `tracedecay_impact`, getting symbols, call chains, affected functions, docs and source snippets. Without it the agent loops through `grep` and whole-file reads. The toolset also covers test mapping, code health checks and anchored edits. A background service keeps the index fresh and understands git branches and worktrees. A local dashboard shows stored context, sessions and token savings.
Who it is for. Developers with large repositories where an agent burns many tokens just locating the right code.
Good fit when
- An agent in Claude Code or Cursor takes ages to find who calls a function.
- Before a refactor you want to see which tests and files an edit touches.
- A monorepo is so large that file reads eat the context before work starts.
- Several git branches run on one project and the agent must know which one it is in.
Not a fit when
- The repository is small and ordinary file search is enough for the agent.
- Policy forbids background services and installing binaries.
- The project language is not among the compiled parser extensions.
Example request
Using the code graph, show who calls the config parsing function and which tests a change to it would affect.Limitations
The README installer downloads and runs a script, then verifies the archive signature with `gh attestation verify`. Without the GitHub CLI the install stops unless you accept a checksum-only mode. The daemon must be running before `tracedecay init`, otherwise the command refuses to write anything. Right after `init` the index may not be ready, so check `tracedecay status` or add `--wait`. Supported languages are set by the project `Cargo.toml`, and building from source requires Rust nightly.
How to disable. Remove the agent settings written by `tracedecay install` and stop the background service. Delete the project index folder and the binary separately.
FAQ
How do I connect TraceDecay to an agent?
Run `tracedecay install`: it detects supported agents itself. Add `--agent` for one agent, for example `--agent codex`, or `--local` for a project without global changes. Then restart the agent.
Which agents are supported?
The README names Claude Code, Codex, Cursor, Hermes, Kiro, Kimi Code, OpenCode and the Cline family.
What if status shows an unready index after init?
Wait or run `tracedecay init --wait`, which blocks until the first index generation is built. Check state with `tracedecay status`.
Does data go to the cloud?
Not by default: storage is local. Configured remote sources follow an explicit policy rather than being on by default.
How do I check that the install works?
Run `tracedecay doctor`: the command only reads state and reports problems. `tracedecay tool` lists every MCP tool.
Discussion
No questions or reviews yet.
Related
CLI toolsA Rust MCP server and CLI with a call graph and hybrid semantic code search for Claude Code and Codex
A local MCP server with call graphs and symbol search: the agent sees code structure and what an edit affects in advance.
CodeGraph (Rust)
codegraph-rust
A Rust code graph for agents: four MCP tools show what an edit will touch, how the architecture fits together and where the fragile spots are
A local code graph with version history: an MCP server that tells the agent who calls a function and what an edit breaks
Local code graph for agents: an MCP server returns entry points, symbol links and change impact in one call
A Rust binary for agents and the terminal that uses tree-sitter to find a symbol's definition, its calls and file dependencies
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| TraceDecay (this card) | no | 86 | ||
| code-index-mcp | no | not required | 162 | searches and indexes code over MCP; TraceDecay adds a call graph, test mapping and a background service |
| Serena | no | not required | 31,335 | gives agents symbol navigation through language servers; TraceDecay builds its own local tree-sitter graph |
README in short
The repository covers installing the service, connecting to agents, and the `status`, `tool`, `dashboard`, `monitor` and `doctor` commands. The version 2 storage model is described separately.