The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill description mentions a command to install an alternative version of the workflow: npx skills add gtrabanco/agentic-workflow#claude. This reference targets the author's own repository and is presented as a manual configuration option for users.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data, creating an attack surface for indirect prompt injection during the summarization process.
Ingestion points: Untrusted data enters the agent context via git commit messages (git log), GitHub pull request metadata (gh pr view), and user-supplied notes provided as command arguments.
Boundary markers: The skill instructions mandate a specific Markdown format for log entries (e.g., ISO-8601 timestamp headers and defined bullet points), which serves to delimit the ingested data within the docs/LOGS.md file.
Capability inventory: The skill possesses the ability to execute git and gh (GitHub CLI) commands and has write access to the local file system to maintain docs/LOGS.md.
Sanitization: The instructions do not include specific requirements for sanitizing, escaping, or filtering the content retrieved from git logs or GitHub metadata before incorporating it into the project documentation.

log-session