The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands via the Bash tool to analyze repository history and file churn.
Evidence: The procedure explicitly instructs the agent to run git log --format= --name-only --since=<a year or so> to extract change counts and co-change pairs.
Risk: Although used for its intended purpose of structural analysis, shell access is a powerful capability that could be targeted for command injection if input parameters are improperly handled.
[PROMPT_INJECTION]: The skill operates on the entire contents of a repository, making it susceptible to indirect prompt injection.
Ingestion points: The skill uses Read, Grep, and Glob tools to ingest arbitrary files, including architectural records and code files, from the target repository.
Boundary markers: The instructions do not specify the use of delimiters or warnings to ignore instructions embedded within the files being surveyed.
Capability inventory: The agent has Agent and Task capabilities to spawn sub-agents and Write capabilities to generate durable proposal documents.
Sanitization: There is no mention of sanitizing or filtering the ingested data before it is processed by the main agent or its sub-agents, allowing malicious code comments or documentation to potentially influence the analysis output.

refactorkit