The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to resolve execution paths and manage analysis artifacts. It searches the user's home directory (~/.claude) to locate its internal workflow scripts if environment variables are not set, and it creates timestamped directories for storing audit results. It also executes a findings assembler script using the uv tool.
[DYNAMIC_EXECUTION]: The skill performs dynamic loading and execution of scripts based on computed paths. It resolves a plugin_root at runtime and uses this to invoke the c-review.js workflow and the assemble_findings.py utility.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external C/C++ source code, creating a surface for indirect prompt injection attacks where malicious code patterns could attempt to influence the behavior of the analyzing agents.
Ingestion points: The repository source code identified by the finding_scope_root parameter in SKILL.md.
Boundary markers: The instructions require the agent to provide a prominent disclaimer stating that no false-positive review was performed and that findings are unadjudicated reviewer opinions (SKILL.md).
Capability inventory: The skill has access to the Bash tool for shell commands, the Workflow tool for executing internal scripts, and the Read tool for file access (SKILL.md).
Sanitization: There are no documented sanitization or filtering procedures for the source code inputs before they are processed by the review agents (SKILL.md).

c-review