The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from local files and architectural guidelines while maintaining the capability to modify the file system through code refactoring.
Ingestion points: Reads architectural rules from AGENTS.md and project source files across various directories including src/, domain/, and infrastructure/ (Step 1, Step 2).
Boundary markers: The instructions do not specify any boundary markers or directives for the agent to ignore potentially malicious instructions embedded within the analyzed source code.
Capability inventory: The skill is authorized to perform structural file modifications, including creating new interface files, updating existing implementations, and generating configuration files (Step 5, Step 7).
Sanitization: No sanitization process is described for the content read from external files before it is used to influence the agent's refactoring and code generation logic.
[DYNAMIC_EXECUTION]: The skill generates executable configuration files and references dynamic code evaluation.
Evidence: The agent is instructed to generate a .dependency-cruiser.cjs JavaScript configuration file (Step 7). The security section also explicitly mentions dynamic evaluation of code to trace imports.
[COMMAND_EXECUTION]: The skill executes multiple shell commands to perform linting and dependency analysis.
Evidence: Uses npx madge, npx dependency-cruiser, vendor/bin/deptrac, and lint-imports (Step 2).
[EXTERNAL_DOWNLOADS]: The use of npx for linting tools involves the potential download and execution of packages from the npm registry.
Evidence: Execution of npx madge and npx dependency-cruiser.

agentic-linter