The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the entire codebase (source code, configuration files, CI/CD scripts) to determine which components are 'dead' and can be safely removed.
Ingestion points: Reads repository files, package.json, Dockerfile, CI/CD workflows, and environment configuration files.
Boundary markers: No specific delimiters or safety instructions are defined to separate the code being analyzed from the agent's own instructions.
Capability inventory: The skill is capable of deleting files and directories (Execute mode), writing to .safe-code/ and AGENTS.md, and executing arbitrary project-defined scripts for verification.
Sanitization: Relies on confidence scoring and post-deletion verification (linting, testing) to mitigate risks, though these do not prevent the initial injection.
[COMMAND_EXECUTION]: The skill executes shell commands defined within the project, such as npm run lint, npm run test, and various build or compile commands, to verify the integrity of the codebase after pruning. This grants the skill the ability to run code defined in the analyzed project's environment.
[DYNAMIC_EXECUTION]: In the 'Execute' mode, the agent performs file system modifications and executes build/test probes that are dynamically determined by the project's structure and configuration files.

codebase-pruner