AntFleet

Receipt · e488cbca-0

Trusted-sources logic documented in SKILL.md is not implemented in scan.sh

docs-gapmediumclosed in 4b9b492closed in 1 hour
repo 6f7fc663·PR #29·reviewed 2 days ago·2 days ago

The finding

  • skills/skill-security-scan/SKILL.md:30-35
  • skills/skill-security-scan/scan.sh:88-102
SKILL.md tells Claude to apply a reduced scan for skills from trusted sources. scan.sh reads TRUSTED_FILE into arrays but never references TRUSTED_OWNERS or TRUSTED_REPOS again in scan_file or anywhere else. The documented behavior is misleading: trusted sources have no effect on output. This causes both false positives in practice and a deceptive contract for callers of the skill.

Fix

Either implement the trusted-source downgrade (compare skill frontmatter origin / git remote to TRUSTED_OWNERS+TRUSTED_REPOS and emit PASS or skip content patterns) or remove/adjust the SKILL.md claim.

Agent attribution

The agents that produced this receipt — both reviewer models had to flag this independently for the agreement gate to emit it.

anthropic

gpt-5

65.9s · error

openai

claude-opus-4-7

101.4s · error

Total

wall-clock review time · est. inference cost

101.4s · $0.40

Sweeper

closed at SHA 4b9b492

closed in 1 hour

internal review id · e488cbca

Third-party witnesses

Everything below lives on GitHub's event log, not ours. Click any link to verify the SHA, the timestamp, and the surrounding context for yourself.

← back to all receipts