The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from GitHub issue bodies and comments via the gh issue view command. This external content is untrusted and could contain malicious instructions designed to hijack the agent's workflow during root cause analysis or implementation.
Ingestion Points: GitHub issue content fetched during Phase 1 (Understand Issue).
Boundary Markers: The skill documentation references a quarantine protocol to extract facts from untrusted inputs before they are used by execution agents.
Capability Inventory: The skill is granted extensive capabilities including Bash, Write, Edit, and Agent (sub-agent spawning), which increases the potential impact of a successful injection.
Sanitization: The workflow includes a specific step to identify repro facts independently of the raw issue body, though the underlying risk remains present due to the complexity of the data processed.
[PERSISTENCE]: The skill utilizes CronCreate to schedule a background task that runs every 5 minutes to monitor the status of Pull Request CI checks. While functional for the intended task, this enables the agent to create long-running scheduled operations that persist beyond the initial user interaction.
[EXTERNAL_DOWNLOADS]: The script prior_fix_lookup.py references an internal dependency yg-mcp-core and its availability on a private package registry at pypi.yonyon.ai. This registry is owned by the vendor (yonatangross) and is used for hosting proprietary MCP client utilities.

fix-issue