The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill makes extensive use of npx to execute scripts from the claude-flow@alpha and ruv-swarm packages. These packages originate from the ruvnet organization, which is not a trusted vendor. Using npx with unversioned or alpha-tagged packages allows for the execution of arbitrary remote code at runtime.
[EXTERNAL_DOWNLOADS]: The skill integrates with external tools and services such as the ruvnet/swarm-action@v1 GitHub Action and MCP servers for task orchestration. These dependencies are fetched from external repositories during execution.
[COMMAND_EXECUTION]: Utilizes shell commands and the GitHub CLI (gh) to automate actions like creating issues, editing project boards, and fetching repository data. While these tools are standard, their execution is driven by logic that incorporates external, untrusted input.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from GitHub issues and comments as part of its core workflow.
Ingestion points: The skill reads issue titles, bodies, and comments using gh issue view and gh issue list in SKILL.md.
Boundary markers: There are no explicit instructions or delimiters used to prevent the agent from following commands embedded within the issue content.
Capability inventory: The agent has the ability to modify GitHub issues, manage project cards, and execute npx scripts based on the ingested data as seen in SKILL.md.
Sanitization: No sanitization or validation of the ingested GitHub content is performed before it is used to drive swarm tasks and command arguments.

github-project-management