The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's documentation in references/frameworks.md instructs the agent to execute specific local Python scripts, including scripts/rice_prioritizer.py, scripts/customer_interview_analyzer.py, and scripts/assumption_mapper.py, using the python3 command to perform calculations and data extraction.
[PROMPT_INJECTION]: The skill establishes an indirect prompt injection surface by instructing the agent to process untrusted external data sources.
Ingestion points: The agent is directed to process "raw transcripts" for sentiment analysis and "CSV files" (e.g., assumptions.csv) for prioritization scoring, as detailed in references/frameworks.md.
Boundary markers: The instructions do not provide explicit delimiters or "ignore" directives to help the agent distinguish between its system instructions and potential injection attempts embedded within the user-provided transcripts or data files.
Capability inventory: The skill utilizes local CLI-based Python scripts to extract pain points, quotes, and metrics from these data sources.
Sanitization: No evidence of input validation, content filtering, or sanitization logic is mentioned in the provided documentation to protect the execution environment from adversarial data.

strategy