Skip to content

Security: Faultline-lab/faultline

SECURITY.md

Security policy

Scope

FAULTLINE is a synthetic research workbench. It should not contain production credentials, private documents, personal data, or live provider tokens.

Security-sensitive areas include:

  • dependency and GitHub Action pins;
  • container provenance and non-root execution;
  • trace payload redaction and path handling;
  • command construction in reproduction scripts;
  • evidence integrity and claim-source binding.

Reporting a vulnerability

Please use GitHub’s private vulnerability reporting for this repository:

  1. Open the repository’s Security tab.
  2. Choose Report a vulnerability.
  3. Include the affected revision, reproduction steps, impact, and any proposed containment.

Do not open a public issue for a vulnerability that could expose data, execute untrusted input, or compromise the build.

Supported version

Security fixes target the current main branch and the latest release candidate. Historical day modules are preserved for reproducibility; when a historical artifact is unsafe to execute, the remediation will be documented without silently rewriting the original result.

Build and data guarantees

  • GitHub Actions are pinned to immutable commit SHAs.
  • Python and OS dependencies used by the clean container are exactly pinned.
  • The runtime user is non-root.
  • .git, local environments, caches, and editor state are excluded from the container context.
  • Seeds make experiments reproducible; they are not secrets or security boundaries.
  • Committed traces are synthetic. Contributions must redact real prompts, documents, tokens, request identifiers, and provider metadata before commit.

The pinned build improves reproducibility, not supply-chain certainty by itself. A production release process should additionally produce an SBOM, sign images and provenance, scan dependencies, and define a patch-latency policy.

There aren't any published security advisories