The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using git and the GitHub CLI (gh) to retrieve project history and pull request information (SKIL L.md Step 2 and Step 1).
[EXTERNAL_DOWNLOADS]: Interface s with external APIs for Slite (api.slite.com), Notion (api.notion.com), and Confluence to publish documentation. The se are well-known services used according to the skill's primary function.
[DATA_EXFILTRATION]: Collect s project-sensitive information—including specification s, task status, git diff s, and pull request review s—and transmit s this data to external documentation platforms (Slite, Notion, Confluence) as part of the document publication process (SKIL L.md Step 7).
[PROMPT_INJECTION]: The skill ingests untrusted content from multiple sources (project file s, com m it message s, PR comment s) and synthesize s the m into a summary document. * Ingestion point s: Read s SPEC.md, task s.md, verify-report.md, .planning/seed.md, .planning/worklog.md, and pull request data via gh pr view. * Boundar y marker s: None observe d in the instruction s. * Capabilit y inventor y: Execut e s shell commands (git, gh), writ e s to the local file system, and perform s network request s. * Sanitization: The instruction s do not specif y sanitization or escaping of the ingested external data before it is processe d or publishe d.
[CREDENTIALS_UNSAFE]: The documentation reference s the use of $SLITE_API_KEY, $NO TION_TOKEN, and basic authentication for API interaction s, which are handle d as environment variable s (SKIL L.md Step 7).

gsdl-close-loop