The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several CLI tools and local scripts to perform its tasks.
Executes a local bash script located at ./.claude/skills/release-notes/scripts/find-merged-prs.sh to identify relevant pull requests.
Invokes git log and gh pr list within the utility script to extract repository metadata.
Uses gh pr view PR_NUMBER --comments to retrieve discussion threads for analysis.
[EXTERNAL_DOWNLOADS]: Fetches configuration data from the vendor's official repository.
Uses curl to download install.sh from the i-am-bee/agentstack GitHub repository to identify the latest stable version tag.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection attack surface by processing external content.
Ingestion points: The skill ingests PR comments and descriptions from GitHub via the gh tool in SKILL.md.
Boundary markers: The prompt does not define clear delimiters or use "ignore embedded instructions" warnings when processing the fetched PR text.
Capability inventory: The agent has the capability to execute shell commands, run local scripts, and perform network requests via curl.
Sanitization: No explicit sanitization or filtering of the content retrieved from GitHub is performed before the LLM processes it for summarization.

release-notes