The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is configured to invoke scripts located outside its immediate directory, specifically ../../scripts/dynamic_router.py, ../../scripts/neural_router.py, and ../../scripts/skill_router.py. These scripts are used to manage workflow routing and dependency management based on user requests.
[DYNAMIC_EXECUTION]: The skill uses a dynamic router to "build a bounded DAG" (Directed Acyclic Graph) at runtime when explicit dependency chains are requested. This represents a dynamic assembly of execution logic based on user input.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest "current external research" to support material claims. This creates a potential surface for indirect prompt injection if external sources contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Material claims are ingested from external research and references/skill-spec.json.
Boundary markers: The skill employs an "Evidence discipline" that requires classifying claims as fact, inference, assumption, or unknown, which serves as a logical delimiter for processed data.
Capability inventory: The skill has the capability to execute subprocesses via the mentioned routing scripts and interact with multiple specialist agents (response-strategist, marketing-skeptic).
Sanitization: No technical sanitization or filtering of external research text is specified, though instructions mandate verifying claims against primary evidence.

offer-strategy