The Agent Skills Directory

[COMMAND_EXECUTION]: The skill dynamically loads and executes Python code from a computed file path in scripts/pipeline/strategy_exporter.py. It attempts to import and run a validate_interface_contract function from a file located at ../../../../edge-candidate-agent/scripts/candidate_contract.py relative to the script's location. This pattern of dynamic loading from computed paths is a security risk if the file system environment or the target neighbor directory can be influenced by an attacker.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection due to the way it processes external data.
Ingestion points: The journal_snippets field in the input JSON is processed by scripts/pipeline/evidence_extractor.py.
Boundary markers: Absent. The extracted journal snippets are formatted into a markdown list within the evidence_summary and passed to the LLM without delimiters or instructions to ignore instructions embedded in the data.
Capability inventory: The skill has the ability to write multiple files (output_bundle.json, hypothesis_report.md, strategy.yaml, metadata.json) and perform dynamic code execution via importlib.
Sanitization: There is no sanitization, escaping, or validation of the content within journal_snippets before they are interpolated into the prompt template.

trade-hypothesis-ideator