Skip to content

Security: Pondsi/ModelSelectorOpenClaw

Security

.github/SECURITY.md

Security Policy

Reporting a vulnerability

Please do not open a public issue for a security problem.

Use GitHub's private vulnerability reporting instead: https://github.com/Pondsi/infinitycontext/security/advisories/new

If that is not available to you, open a minimal issue asking for a private channel without any technical detail, and the maintainer will reach out.

Please include, when you can:

  • the version (version in SKILL.md frontmatter) and host (dsh / OpenClaw / Claude Code / …);
  • the exact command and the JSON result, redacted of secrets;
  • a description of the impact and, if possible, a minimal reproduction.

Scope

In scope: the published package — SKILL.md, README.md, 说明.md, CHANGELOG.md, SPONSORS.md, LICENSE, checksums.txt, scripts/*.py, references/*.md, sponsors/*.

Out of scope for this policy: the optional OpenClaw/Windows integration in the repository's openclaw/ folder. It is not part of the published artifact and carries its own openclaw/checksums.txt; report issues there with the same private channel.

Design boundaries worth knowing

  • The core has no network access, no shell and no subprocesses.
  • The archive is owner-only and fail-closed: if permissions cannot be enforced, archiving aborts and the half-written database is destroyed.
  • Redaction is best-effort, not encryption. Treat the archive as sensitive data.
  • Retention is bounded by default (30 days) and can be disabled entirely with INFINITY_CONTEXT_NO_ARCHIVE=1.

Response

Reports are acknowledged as soon as possible. A fix is released with a new patch version, the ClawHub audit for that version is re-run, and the changelog entry names the issue without exposing the reporter.

There aren't any published security advisories