The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple shell commands using git, bun, node, and npm to perform repository status checks, build verification, testing, and version control operations.
[EXTERNAL_DOWNLOADS]: The skill interacts with the NPM registry to verify user credentials and publish the final package, which is the intended behavior for a release automation tool.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from Git logs and file diffs to populate release notes. It mitigates this risk by requiring human confirmation for version changes, using fixed templates for Git commits, and instructing the agent to summarize and filter raw log data before inclusion in documentation. Evidence chain: 1. Ingestion points: git log and git diff outputs in Section 2. 2. Boundary markers: Explicit instructions to wait for user confirmation before modifying files or publishing. 3. Capability inventory: Includes Git tagging, committing, and NPM publishing. 4. Sanitization: Manual review steps and summarization requirements for changelog entries.

release