The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes a user-supplied <plan> which acts as an untrusted input source. An attacker could embed instructions within a plan to manipulate the CISO review outcome or the behavior of the agent.
Ingestion points: The <plan> argument in the /c-level-agents:ciso-review command.
Boundary markers: No explicit markers are defined in the skill instructions to distinguish between the analysis logic and the content of the plan being analyzed.
Capability inventory: The skill has the ability to execute shell commands and Python scripts.
Sanitization: There are no instructions for the agent to sanitize or escape the contents of the plan before processing it.
[COMMAND_EXECUTION]: The skill executes local Python scripts to perform its core analytical functions. While these are part of the skill package, they represent a capability that could be targeted if the input parsing is compromised.
Evidence: python ${CLAUDE_PLUGIN_ROOT}/vendor/c-level-advisor/ciso-advisor/scripts/risk_quantifier.py
Evidence: python ${CLAUDE_PLUGIN_ROOT}/vendor/c-level-advisor/ciso-advisor/scripts/compliance_tracker.py

ciso-review