10k-value-prop-match
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes large volumes of external text from SEC 10-K filings fetched from public URLs. This content is analyzed by the agent to identify and grade strategic initiatives.
- Ingestion points: Fetches Item 1 (Business), Item 1A (Risk Factors), and Item 7 (MD&A) from public SEC EDGAR filings (SKILL.md, Step 2.4 and Step 5).
- Boundary markers: The instructions lack explicit delimiters or warnings for the agent to ignore potentially malicious instructions embedded within the fetched document text.
- Capability inventory: The skill executes platform commands (
clay whoami) and resolves company identities using external APIs (SKILL.md, Step 0 and Step 2). - Sanitization: No explicit sanitization or filtering is performed on the filing text before it is presented to the agent for analysis.
- [COMMAND_EXECUTION]: The skill performs environment verification by executing a shell command via the platform's CLI tool.
- Evidence: Instructions require running
clay whoami; echo "exit_code=$?"to determine the current workspace and configuration (SKILL.md, Step 0). This is a diagnostic use case using a fixed command string.
Audit Metadata