The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a utility for generating changelogs from git history.
Evidence: roadmap-communicator/scripts/changelog_generator.py executes the git log command via subprocess.run using a list of arguments.
Context: This implementation does not use a shell and is a secure, standard way to provide legitimate development functionality.
[INDIRECT_PROMPT_INJECTION]: Multiple tools in the collection process external data, creating an attack surface where instructions embedded in processed content could influence the agent.
Ingestion points: customer_interview_analyzer.py (transcripts), extract_citations.py (research papers), codebase_analyzer.py (source code), and spec-to-repo (project specifications).
Capability inventory: The skill has the ability to write files to the local filesystem (scaffolders) and execute local git commands, which could be leveraged if the agent obeys instructions embedded in the input data.
Sanitization: No explicit sanitization or boundary markers are present in the script logic to distinguish data from instructions.
[EXTERNAL_DOWNLOADS]: The skill instructions describe fetching competitive intelligence from well-known services.
Evidence: competitive-teardown/SKILL.md provides patterns for querying the itunes.apple.com API.
Context: These are well-known technology platforms and the data collection is intended for research purposes.

product-skills