superpm
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external public sources to ground project management artifacts.
- Ingestion points: The
reference/signal.mdfile usesWebSearchandWebFetchtools to ingest content from Hacker News, GitHub, Polymarket, Jina Reader, and general web search results. - Boundary markers: The skill implements a mechanical 'Critic' verification step in
reference/critic.mdto red-team drafts and verify claims against sourced evidence. - Capability inventory: The skill has the capability to write Markdown and HTML files and perform network requests via
WebSearch/WebFetch. - Sanitization: Data is validated against the framework's 'completeness bar' and 'real decision' context during the independent Critic phase, which serves as a logical filter for malicious or irrelevant input.
- [EXTERNAL_DOWNLOADS]: The skill references and fetches data from well-known technology services.
- Fetches market signals from Hacker News via the Algolia API.
- Retrieves repository metadata and demand signals from the official GitHub API.
- Accesses prediction market data from the Polymarket API.
- Converts web pages to markdown using the Jina Reader public utility (
r.jina.ai).
Audit Metadata