SigMap — the deterministic, verifiable grounding layer for AI code work

Release: v8.9.1·New — squeeze_output MCP tool + squeeze --response: the deterministic squeeze engine is now reachable by agents mid-session — compress a noisy stack trace, CI/build log, or JSON payload before it enters context (keeps the signal, strips the noise, enriches the top frame). The 19th MCP tool; also a named CLI flag. Zero-dependency, offline, deterministic.

Benchmark: sigmap-v8.9-main·88% hit@5 · 97.0% token reduction · 2026-07-06

30-second start ​

Step 1: Generate context for your project

Step 2: Ask for relevant files (query-specific context)

sigmap ask "explain the auth flow"
# Outputs: ranked file list + .context/query-context.md (ready to paste)

Step 3: Copy context to your AI assistant

• Open .context/query-context.md
• Paste the content into Claude, Copilot, ChatGPT, or your IDE's AI chat
• Ask: "Explain the auth flow"

Step 4: Save the AI response

# Copy the AI's answer into a file
echo "Paste AI response here..." > response.txt

Step 5: Validate coverage (optional)

sigmap validate --query "auth login token"
# Check if coverage is high enough to trust the response

Step 6: Judge groundedness

sigmap judge --response response.txt --context .context/query-context.md
# Score: shows if the answer is grounded in your code

That flow gives you: a compact signature map · a focused query context · a coverage sanity check · a groundedness score for the answer.

The workflow ​

SigMap is no longer just "shrink the context file." Every step has a purpose:

• Generate a compact signature map once
• Ask for the files that matter to the current task
• Validate whether coverage is high enough to trust the context
• Judge whether an answer is grounded in the supplied code
• Learn from good and bad results locally, inside the repo

See the full end-to-end walkthrough to watch this in action on a real repo.

Latest saved benchmark snapshot ​

Latest saved benchmark run: 2026-07-06 (v8.9.1).

Where to go next ​

• New to the product: Quick start
• Want the core daily flow: ask, validate, judge, learning
• Using Claude Code or Cursor: MCP setup
• Evaluating the launch claims: Benchmark overview
• 🌍 See the community: where SigMap's stargazers are around the world