The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources, including pull request titles, bodies, and commit messages, to generate text for a CHANGELOG.md file.
Ingestion points: Pull request metadata (number, title, body, author, labels) is fetched via tracker operations (SKILL.md, references/release-window.md).
Boundary markers: The skill explicitly defines an 'Untrusted content boundary' in references/agentic-setup.md, instructing the agent to treat repository and tracker content as non-instructional data and to ignore/report any embedded directives.
Capability inventory: The skill has the ability to modify the local CHANGELOG.md file and invoke companion skills like om-auto-create-pr to create pull requests.
Sanitization: Externally-sourced values such as branch names and PR numbers are validated against strict alphanumeric regex patterns (^[A-Za-z0-9._/-]+$) before being used in shell commands or path interpolation.
[SAFE]: The skill incorporates several high-standard security practices for autonomous agents.
Secrets Hygiene: Explicit rules in references/rules.md prevent the agent from including secrets, tokens, or environment variables in its output, even if explicitly instructed by external data.
Human Review Checkpoint: The workflow is designed to delegate pull request creation to a separate skill which opens the PR for manual review, ensuring that generated content is not automatically merged without oversight.
Identity Protection: The skill implements a 'never-credited' list to ensure bot accounts and AI coding agents are not attributed as human contributors.

om-auto-update-changelog