Pre-runtime scanning
Static code scan reads an agent's own source — a local directory, an uploaded zip, or a GitHub repo — for hardcoded secrets, PII in log statements and sensitive values sent to third parties, before a single trace is captured.
Point the scanner at a local directory, a zip file, or a GitHub repo and branch — it reads the same source your agent runs, not a sandboxed copy.
Findings use the identical L1 to L4 severity scale as trace analysis, plus code-specific layers, so a hardcoded API key and a leaked account_id are directly comparable.
POST /api/agent/code is part of the same scoped-key API an agent uses to self-test and improve, so a code scan can run automatically before every deploy.
Inside the capability
Inspect the repository, link each finding to its exact file and line, then carry that evidence into the runtime trace.
Link the read point to logging, tool and third-party sinks so the fix happens at the correct boundary.
Compare the unsafe call with its redacted replacement and re-score the repository before merge.
GitHub-native privacy automation
Run AgentLeak static analysis on pull requests and merge queues before sensitive code lands.
Fail CI automatically when the code privacy score falls below the threshold set for the repository.
Apply stricter secret, PII and third-party-send rules to production and release branches.
Keep findings, reviewers and the current privacy gate attached to the exact commit in GitHub.
The mental model
Trace analysis catches what an agent does at runtime; static code scan catches what is already sitting in its source, waiting to be read, logged or sent to a third party. Running both closes the gap between what the agent was written to do and what it actually did in production.
A hardcoded key or a debug print of raw PII waits, undetected, for the first run that exercises that code path.
The scan flags the exact file and line before the agent is ever deployed.
How it works
Run the CLI against a local path, a .zip, or pass --repo owner/name to fetch a GitHub repository directly.
Regex, entropy and (optionally) Presidio and an LLM-judge scan every file, plus code-specific de-obfuscation and quasi-identifier correlation.
Each finding lists the file, line, rule, data type and a redacted snippet, with an overall 0-100 code privacy score.
Pass --fail-under to exit non-zero when the code score drops below your threshold, exactly like the trace-analysis CI gate.
Scan a GitHub repo for hardcoded secrets
agentleak scan --repo acme/support-bot --branch main --fail-under 80
# Code privacy score: 74/100 — Conditional pass
# [L3] app/memory_adapter.py:42 hardcoded_api_key (secret, high-entropy)FAQ
Ready when you are
Create a local workspace, run a bundled scenario, then wire AgentLeak into CI or let your agent onboard itself.