The Agent Skills Directory

[SAFE]: The skill serves its stated purpose of competitive analysis without any malicious behavior. It emphasizes provenance tracking and source hygiene as core components of its functionality.
[COMMAND_EXECUTION]: The skill interacts with an internal system ('professional-brain') using provided scripts via relative paths (../professional-brain/scripts/brain_query.py and brain_write.py). This allows the agent to read from and propose updates to a local knowledge base.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external market data, creating an attack surface for indirect prompt injection. It includes specific mitigations for this surface: 1. Ingestion points: External competitor data, pricing information, and reviews gathered from the web (SKILL.md). 2. Boundary markers: Mandatory source tags and provenance markers (e.g., [external], [docs], [G2]) are used to distinguish external claims from trusted internal data (SKILL.md, references/worked-example.md). 3. Capability inventory: The skill can read and write to the local knowledge base using brain_query.py and brain_write.py (SKILL.md). 4. Sanitization: The skill requires all claims to be validated against sources or flagged as assumptions in an 'assumption register' (SKILL.md, references/worked-example.md).

competitive-analysis