The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several standard industry tools for static analysis, including knip, depcheck, vulture, and go mod tidy. These tools are invoked to detect unreachable code and unused dependencies within the user's local repository. The execution is gated by a triage phase where the user must approve findings before any modifications occur.
[DATA_EXFILTRATION]: No network activity or data exfiltration patterns were observed. The skill's operations are confined to the local filesystem for analysis and cleanup.
[PROMPT_INJECTION]: The skill instructions do not contain any patterns intended to bypass AI safety guardrails or override system prompts. It maintains a strictly technical and functional focus on code cleanup.
[CREDENTIALS_UNSAFE]: There are no hardcoded secrets, nor are there instructions to access sensitive environment variables or credential stores like AWS or SSH keys. It correctly references its own local configuration files (e.g., .absolute.config.json) for project-specific preferences.
[REMOTE_CODE_EXECUTION]: The skill does not download or execute scripts from remote or untrusted servers. It relies on the project's own scripts and standard package managers for its operations.
[SAFE]: The skill incorporates a robust 'health engine' that enforces security and stability best practices: it requires starting from a 'green' (passing) state, mandates human confirmation (TRIAGE gate), performs changes in small, independently verifiable batches (waves), and explicitly forbids automatic commits or safety-check bypasses (e.g., @ts-ignore).

absolute-prune