board

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONNO_CODE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes message board content that can be written by various agents or external sources. Ingestion points: The --read command in SKILL.md fetches message content into the agent context. Boundary markers: Content is delimited by YAML frontmatter, but no explicit instructions are provided to the agent to treat this data as untrusted. Sanitization: There is no evidence of sanitization or filtering of the ingested message text.
  • [COMMAND_EXECUTION]: The skill constructs shell commands by interpolating variables like {text}, {author}, and {channel} into a Python script call. Evidence: python {skill_path}/scripts/board_manager.py --post --channel {channel} --author {author} --message "{text}" in SKILL.md. Risk: If input strings are not correctly escaped or validated, they could terminate the intended command and execute arbitrary shell instructions.
  • [NO_CODE]: The implementation script scripts/board_manager.py referenced in SKILL.md is missing from the provided files, preventing verification of its internal logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 02:34 PM
Security Audit — agent-trust-hub — board