The Agent Skills Directory

[DYNAMIC_EXECUTION]: The test file scripts/tests/test_check_circuit_breaker.py performs dynamic code loading by importing a companion module (thesis_store.py) from a relative filesystem path using importlib. Executing code from paths computed at runtime represents a security risk as it could be manipulated to load malicious scripts if the directory structure is compromised.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted trade data from local YAML files to evaluate risk thresholds. This ingestion of external data creates a surface for indirect prompt injection or data poisoning, which could influence the agent's decision-making or bypass risk limits.\n
Ingestion points: Data is loaded from th_*.yaml files located in the state/theses/ directory.\n
Boundary markers: The skill produces circuit_breaker_decision reports in JSON and Markdown formats for use as gates in trading workflows.\n
Capability inventory: The skill performs read operations on local data files and write operations for report generation.\n
Sanitization: The implementation employs yaml.safe_load and performs rigorous numeric type and range validation on all ingested metrics.

drawdown-circuit-breaker