ai-forge-recap

Pass

Audited by Gen Agent Trust Hub on Aug 17, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill analyzes external instruction files (skills, agents, and references) which are untrusted data sources. This creates a potential surface where malicious content in the audited files could influence the agent's recap or lead to incorrect tool invocations. \n- Ingestion points: Reads file paths provided by the environment, targeting 'SKILL.md', '.agent.md', and 'references/' content. \n- Boundary markers: Absent. The instructions do not define delimiters or explicit rules to isolate audited content from the agent's reasoning process. \n- Capability inventory: The skill can trigger updates to frontmatter through 'ai-forge-apply' and to the file body through 'ai-forge-update', creating a cycle where read data can drive write operations. \n- Sanitization: Absent. The skill does not describe any validation or filtering of the ingested content before it is used for comparison and reporting.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 17, 2026, 07:14 AM
Security Audit — agent-trust-hub — ai-forge-recap