The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script using uv run to show the current state of the project. While this uses the platform-provided $CLAUDE_PROJECT_DIR variable, executing scripts based on the file system is a significant capability.
[PROMPT_INJECTION]: The skill processes findings from findings.json and a 'brain' storage to discover attack chains. This represents an Indirect Prompt Injection surface where malicious content within the findings could influence the correlation agent.
Ingestion points: Reads data from findings.json and internal 'brain' storage (SKILL.md).
Boundary markers: None identified. The instructions do not define clear delimiters or instruct the agent to ignore instructions embedded within the findings data.
Capability inventory: The skill has the ability to execute shell commands via uv run and update internal project state (SKILL.md).
Sanitization: No sanitization or validation of the findings data is performed before it is passed to the LLM for analysis.

correlate