The Agent Skills Directory

[DATA_EXFILTRATION]: The skill explicitly instructs the agent to locate and read sensitive configuration and environment files that often contain secrets or credentials.
Evidence: SKILL.md (Example Exploration Patterns) and references/analysis-checklist.md (Config) direct the agent to use Grep and Glob to find files matching **/.env*, os.environ, config[, and settings..
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted codebase files and instructions, making it vulnerable to malicious content embedded in the code being analyzed (e.g., comments, README files, or variable names) that may attempt to redirect the agent's behavior.
Ingestion points: Workspace files accessed through the Read, Grep, and Glob tools as defined in SKILL.md and references/analysis-process.md.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the provided skill files.
Capability inventory: The skill has access to Bash, Read, Glob, and Grep tools, and includes instructions for writing documentation to the local filesystem (specs/ directory).
Sanitization: There is no evidence of input validation, sanitization, or filtering applied to the content read from the codebase before it is processed or used to generate documentation.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform structural discovery and exploration of the filesystem. If an attacker can influence the paths or patterns passed to these tools via indirect prompt injection, it could lead to unauthorized command execution.
Evidence: SKILL.md defines Bash as an allowed tool, and references/analysis-process.md provides bash-based patterns for discovery tasks.

spec-miner