Skills
Navigate
skills/adrianbrowning/agent-skills/conversation-to-html/Gen Agent Trust Hub

conversation-to-html

Pass

Audited by Gen Agent Trust Hub on Aug 30, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (git diffs and conversation history) and interpolates it into a structured HTML template.
  • Ingestion points: SKILL.md instructions specify reading content from the current context, including "Recent code changes / git diff" and "Discussion / decisions".
  • Boundary markers: The skill mentions escaping HTML special characters (<, >, &) which helps prevent HTML injection but does not prevent the agent from being influenced by natural language instructions within the diff or discussion.
  • Capability inventory: The skill uses open to launch a browser and writes to the local file system (/tmp/).
  • Sanitization: Escaping of HTML entities is specified for diff lines, but no further validation of conversation context is mentioned.
  • [COMMAND_EXECUTION]: The skill utilizes shell commands to facilitate its workflow.
  • Evidence: SKILL.md instructs the agent to use date +%s for timestamps and the open command to display the generated file to the user.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 30, 2026, 10:12 PM
Security Audit — agent-trust-hub — conversation-to-html