The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests untrusted data from external sources to generate its output.
Ingestion points: Reads content from local files (./docs/plans/*.md, ./docs/prd-*.md), fetches remote GitHub issues via the gh CLI, and consumes Notion page content or in-context user conversations.
Boundary markers: The skill uses specific XML-like tags (e.g., <vertical-slice-rules>, <ticket-template>) to delimit instructions and templates from processed data.
Capability inventory: The skill can execute shell commands via gh (to view/create/edit issues), write to local configuration files (command-centre.toml, treepad manifests), and invoke external agent skills (create-adhoc-linear-issue).
Sanitization: A robust human-in-the-loop (HITL) mechanism is enforced. The agent must present a drafted plan and "Quiz the user" for approval before performing any ticket creation or file modification.
[COMMAND_EXECUTION]: The skill utilizes standard shell commands and the GitHub CLI to perform project management tasks.
Evidence: Employs gh issue view, gh issue create, gh issue edit, git rev-parse, and ls to interact with the repository and issue tracker for its primary functionality.

to-tickets