The Agent Skills Directory

[COMMAND_EXECUTION]: The skill workflow requires the agent to execute shell commands (e.g., opentaint test rule reachability) using several variable parameters such as <model-dir>, <rules-dir>, <results-dir>, and <full-id>. These variables are sourced from caller inputs or environment defaults. Direct interpolation of these parameters into shell strings without explicit sanitization or validation instructions creates a risk of command injection if an attacker provides malicious input values.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by instructing the agent to read and process external data from files such as dropped-external-methods.yaml and SARIF reports (debug-ifds-fact-reachability.sarif). Malicious content within these files could attempt to influence the agent's diagnostic logic.
Ingestion points: The skill reads dropped-external-methods.yaml and debug-ifds-fact-reachability.sarif to diagnose taint propagation gaps (SKILL.md).
Boundary markers: No explicit delimiters or warnings to ignore embedded instructions are provided in the workflow.
Capability inventory: The skill uses the opentaint CLI tool for dataflow analysis and reachability testing (SKILL.md).
Sanitization: The instructions do not specify any sanitization, escaping, or validation steps for the content of the external files being processed.

debug-rule