fact-checker
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSMETADATA_POISONING
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from user documents and external web search results while maintaining the capability to modify the local filesystem.
- Ingestion points: The skill scans user-provided files for factual claims (SKILL.md, Step 1) and retrieves data from external web sources (SKILL.md, Step 2).
- Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following malicious instructions potentially hidden within the analyzed data.
- Capability inventory: The skill is instructed to perform web searches and apply corrections back to files (SKILL.md, Step 5).
- Sanitization: A user-approval gate is implemented ("Apply corrections with user approval") to review changes before they are committed.
- [EXTERNAL_DOWNLOADS]: The documentation references external installation methods and source code repositories from an unidentified organization.
- Evidence: README.md instructs the user to install the skill via "ccpm install @daymade-skills/fact-checker" and links to "github.com/daymade/claude-code-skills".
- [METADATA_POISONING]: The skill includes a static file designed to falsely imply a verified safety status.
- Evidence: The file ".security-scan-passed" contains an assertion that a security scan was successful, including a hash and timestamp, which attempts to establish trust through self-reporting.
Audit Metadata