The Agent Skills Directory

[DATA_EXPOSURE]: The skill instructions and associated checklists (in SKILL.md, references/analysis-process.md, and references/analysis-checklist.md) specifically direct the agent to locate and read sensitive configuration files. This includes pattern matching for .env files, environment variable access via os.environ, and configuration modules like ConfigService. While necessary for reverse-engineering legacy systems, these actions involve handling potential credentials within the agent's context.
[COMMAND_EXECUTION]: The skill is configured with Bash tool access and includes multiple examples of shell command usage (e.g., Glob, Grep) to explore file systems and search content. This capability is appropriate for the task but grants the agent broad access to the local environment.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill is designed to ingest and process content from an untrusted external codebase using Read, Grep, and Glob tools.
Boundary markers: The instructions do not define boundary markers or include explicit directions for the agent to ignore instructions embedded in the analyzed code.
Capability inventory: The skill possesses Bash execution capabilities and filesystem write access to the specs/ directory.
Sanitization: There are no provided instructions for sanitizing or escaping content extracted from source files (such as comments or strings) before it is processed by the agent or included in documentation.

spec-miner