Skip to content
Public

ai-repo-safety

skillVersion 1.0
claude-code
security
code-review
View source on GitHub

Description

AI Repo Safety is maintained by letya999 at https://github.com/letya999/ai-repo-safety-skill under MIT.

Guardrails for AI-authored repositories: secret scanning, SAST, GitHub read safety, and MCP/skill supply-chain checks. Canary source for Official sync.

AI STP publishes this snapshot for discovery. AI STP does not claim upstream authorship or affiliation. A verified maintainer may request an ownership transfer, which AI STP will review.

Full machine-readable record; the useful fields are summarized above

0 · Requires credentials: No

Harness projections

63%10 / 16 checks passed1 harness projections
  • claude-code63%Open projection details
    Safety check
    Not verified
    Full-auto recommendation
    Ineffective
    Scope
    global
    Projection
    native_files
    Technical support
    Experimental
    Implementation
    native
    Support note
    operator-reviewed official upstream projection
    Evidence freshness
    2026-09-30T00:04:09.628Z
    Safety check
    • agentic_behavior
      Passed
    • artifact_unpack
      Passed
    • content_hidden
      Passed
    • hook_command_argv
      Passed
    • hook_schema_static
      Passed
    • mcp_config_static
      Passed
    • network_intent
      Failed
      8 finding(s): dangerous_url_scheme, url_pipe_shell
    • path_denylist
      Passed
    • pi_content_pack
      Warning
      5 finding(s): pi_exfil, pi_ignore_previous
    • sast_bandit
      Warning
      3 finding(s): b101, b404, b603
    • sast_eslint_security
      Not run
      result not_run
    • sast_opengrep
      Passed
    • sast_shellcheck
      Warning
      1 finding(s): sc1007
    • sca_npm_audit
      Not run
      result not_run
    • sca_osv
      Incomplete
      result degraded
    • sca_pip_audit
      Incomplete
      result not_applicable
    • secrets_gitleaks
      Warning
      1 finding(s): private-key
    • secrets_heuristic
      Passed
    • shell_obfuscation
      Warning
      1 finding(s): b64_decoded_shell
    • skill_static_gate
      Passed
    Evidence references
    • digest: sha256:cb5006f44a9c0f982345cf6974a75f38d6ff4c79aeca19d371c16342e191a614
    • policy: safety-3
    • profile: sha256:c685d34c5386cd16ba27a57a7fd960e788dafa2544cd8f715ad88176c4ede551