pricing-packaging-planner
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external content such as competitor pricing pages and web crawler exports.\n
- Ingestion points: Data enters the context via
scripts/connectors/firecrawl.pyand user-provided pricing documents.\n - Boundary markers: Includes explicit instructions to treat external data as untrusted and to disregard any embedded instructions in the 'Instructions' section.\n
- Capability inventory: The skill proposes data updates to
memory/events/claims.ndjsonvia theregistry-events.pyscript.\n - Sanitization: Employs natural language constraints to prevent the agent from executing instructions found in processed data.\n- [COMMAND_EXECUTION]: The skill utilizes an internal registry script to maintain records.\n
- It invokes
registry-events.pyfor proposing candidates to the claims and launch registries. This represents standard interaction with the skill's supporting infrastructure.\n- [DATA_EXFILTRATION]: The skill accesses external competitor data for research purposes.\n - Network activity is limited to fetching public pricing information using established connectors, with no evidence of sensitive data being sent to unauthorized external destinations.
Audit Metadata