The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves processing external market data and user objectives, which is a common vector for indirect prompt injection. However, the skill implements rigorous architectural safeguards.
Ingestion points: Market demand evidence, category context, and commercial performance data are ingested via the diagnostic workflow defined in references/skill-spec.json.
Boundary markers: The 'Operating contract' in SKILL.md and the 'Evidence Policy' in references/skill-spec.json establish strict boundaries by requiring the classification of all inputs as fact, inference, or assumption.
Capability inventory: The skill leverages internal routing scripts and specialized agent roles to handle specific sub-tasks.
Sanitization: Invariants in the skill-spec.json forbid the fabrication of research and mandate the verification of platform and pricing claims.
[COMMAND_EXECUTION]: The skill references local scripts (../../scripts/dynamic_router.py and ../../scripts/neural_router.py) to manage task routing. These are standard architectural dependencies within the agent's defined environment and do not involve untrusted remote sources.

market-diagnosis