The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses various command-line tools to manage the feedback workflow:
git tag for versioning feedback rounds.
gh issue list to retrieve reader notes from GitHub issues.
A custom story CLI for building HTML copies, comparing anchors, and maintaining story metadata (reindex, validate, continuity).
A fallback mechanism to execute a local script at a relative path using node ../story-maintenance/scripts/story.js or bun run story.
[EXTERNAL_DOWNLOADS]: The skill relies on external resources and tools:
Interaction with GitHub's API via the gh CLI to fetch external reader feedback.
Dependencies on a 'Story Skills repository' for the story CLI tool.
[INDIRECT_PROMPT_INJECTION]: The skill has a defined surface for indirect prompt injection as it ingests and processes untrusted text from external readers.
Ingestion points: External reader feedback is collected from GitHub issues and local markdown files (feedback/round-{N}/{reader-kebab}.md).
Boundary markers: Feedback is structured within markdown templates and YAML frontmatter, though no explicit "ignore instructions" delimiters are used for the reader content itself.
Capability inventory: The agent can perform file system writes (creating folders, stub files, and copying GitHub templates), execute shell commands, and trigger other skills based on the synthesized feedback.
Sanitization: The workflow involves quoting/paraphrasing feedback, but lacks specific escaping or validation logic to filter out potential malicious instructions embedded in reader notes.

feedback-triage