marketing-plan
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it ingests untrusted data from various sources to generate marketing plans.
- Ingestion points: Reads client-provided files from the materials directory and fetches data from external APIs and MCPs such as Ahrefs, Google Analytics, and Stripe as described in the methodology and SKILL.md.
- Boundary markers: No explicit prompt-level delimiters or instructions to ignore embedded commands are specified for ingested content in the instructions.
- Capability inventory: The skill performs file read and write operations in local directories and invokes various MCP tools for research and execution.
- Sanitization: Implements path safety by slugifying client names and rejecting directory traversal patterns like '..' and '/' for local directory management in the ~/marketing-plans/ path.
- [COMMAND_EXECUTION]: The skill provides instructions for the agent to perform git operations, including clone, commit, and push, to publish finalized plans to a shared repository at the user's request in the finalize phase.
- [NO_CODE]: The skill consists entirely of markdown instructions and reference documentation without including executable scripts or binaries within its own file structure.
Audit Metadata