The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data (session context) to generate feedback reports, which introduces an indirect prompt injection surface.
Ingestion points: The workflow reads the session transcript and user feedback signals to draft the report (SKILL.md, Workflow step 3).
Boundary markers: The skill instructs the agent to summarize interactions instead of copying transcript chunks and enforces a strict markdown template (SKILL.md, references/report-template.md).
Capability inventory: The skill executes shell commands via the GitHub CLI (gh) to create issues and check authentication status (references/submission.md).
Sanitization: The skill provides explicit instructions to exclude secrets, tokens, and private URLs. It also implements a mandatory "human-in-the-loop" approval step where the user must review the exact publication artifact before submission (SKILL.md).
[COMMAND_EXECUTION]: The skill utilizes the gh CLI tool to perform its primary function of creating issues.
Evidence: The references/submission.md file contains shell script snippets for creating issues (gh issue create) and checking status (gh auth status).
The instructions include specific guidance to avoid insecure shell patterns like eval or direct interpolation of the markdown body into the command line, favoring safer data-passing methods.
[DATA_EXFILTRATION]: The skill transmits information (session summaries and user feedback) to a GitHub repository (folio-org/folio-eureka-ai-dev).
This transmission is a functional requirement of the skill and targets a well-known service (GitHub) at a repository owned by the skill's author.

skill-feedback