The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze codebase source files and process diagnostic outputs from various linters and auditing tools. This activity creates an indirect prompt injection surface where instructions embedded in the analyzed code or tool outputs could influence the agent's behavior. This surface is inherent to the skill's primary function as a code auditing utility.
Ingestion points: Codebase source files, linter diagnostics, and schema definitions (referenced in SKILL.md and stack-specific reference files).
Boundary markers: The skill does not define specific delimiters or boundary warnings for the ingested data.
Capability inventory: The agent is guided to execute numerous shell-based tools for code analysis, including rg, ruff, oxlint, clippy, and opengrep.
Sanitization: No explicit sanitization or filtering of codebase content or tool outputs is specified before processing.
[COMMAND_EXECUTION]: The skill provides numerous shell command snippets for executing third-party development and security tools. These commands are intended for environment setup, static analysis, and automated codebase auditing (e.g., using rg for pattern matching or ruff for Python linting).
[EXTERNAL_DOWNLOADS]: The skill recommends the use of several external security and development tools, such as Opengrep, zizmor, and Ultracite. All referenced resources originate from well-known and trusted open-source organizations or established technology services.

mechanical-enforcement