work-output
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process potentially untrusted data from task outputs, including command logs, tool results, and file diffs, which creates a surface for indirect prompt injection.
- Ingestion points: As described in
SKILL.mdandreferences/four-tracks.md, the skill collects commands, logs, diffs, and technical evidence from the agent's task environment. - Boundary markers: The instructions lack specific delimiters or isolation markers for raw untrusted data during the 'distillation' phase, relying on the agent to format the content into Markdown or JSON templates.
- Capability inventory: The agent utilizing this skill typically has access to shell command execution, file system read/write operations, and potentially network access to facilitate the 'Public Track' proposals.
- Sanitization: A detailed 'Publication Gate' is defined in
references/publication-gate.md, which includes a checklist for masking secrets, tokens, IP addresses, and personal information before any data is proposed for release.
Audit Metadata