The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes multiple Python scripts that utilize subprocess.call to execute other scripts within the local directory.
Evidence: scripts/convert_to_csv.py, scripts/parse_excel.py, and scripts/batch_convert_templates.py use sys.executable to run convert_formats.py or parse_formats.py.
These scripts chain local execution to handle file parsing and format conversion (e.g., Markdown to CSV/JSON).
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it is specifically designed to ingest and analyze untrusted external data such as user requirements and API documentation.
Ingestion points: prompts/requirements-analysis.md and prompts/api-testing.md explicitly instruct the agent to read and analyze external docs, stories, and design notes.
Boundary markers: The prompts do not implement delimiters or specific instructions to disregard potentially malicious commands embedded within the analyzed documents.
Capability inventory: The agent has access to local scripts in the scripts/ folder that can read from and write to the local filesystem.
Sanitization: No evidence of input sanitization or validation of the ingested data was found in the prompt instructions.

sprint-testing-workflow