The Agent Skills Directory

[SAFE]: The skill performs legitimate versioning and package management tasks using established tools like Changesets and Git.
[DATA_EXPOSURE]: The scripts/profile_store.py utility manages local user profile data. It implements strong security controls, including atomic writes with restricted file permissions (0o600) and an explicit blocklist (token, secret, password, credential, cookie, api_key, apikey) to prevent the accidental or intentional storage of sensitive credentials in durable records.
[INDIRECT_PROMPT_INJECTION]: The skill maintains an attack surface for indirect prompt injection due to its core function of processing project data. Evidence: 1. Ingestion points: The skill reads local package configurations and Git diffs from the workspace. 2. Boundary markers: The skill specifies a workflow that separates read-only analysis and preview stages from actual execution (SKILL.md). 3. Capability inventory: The skill utilizes file write operations and local CLI execution for Git and package managers. 4. Sanitization: The profile script performs path validation and case-insensitive filtering of sensitive keys to protect the user's environment.

lov-version-management