The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses npx to download and execute packages (ruv-swarm, claude-flow) from the NPM registry at runtime. These tools originate from an external source not identified in the trusted vendor list.
[REMOTE_CODE_EXECUTION]: Execution of npx ruv-swarm and npx claude-flow commands allows for the running of arbitrary code downloaded from a remote registry.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from GitHub issue bodies, comments, and labels.
Ingestion points: Data is fetched via gh issue view, gh issue list, and gh project item-list in SKILL.md.
Boundary markers: The skill includes a 'Security Considerations' section warning about these risks, but markers are not consistently applied in the provided command examples.
Capability inventory: The skill has access to Bash, gh CLI, and npx for network and system operations.
Sanitization: The skill advises using heredocs but examples demonstrate interpolation of issue content into shell variables ($ISSUE_BODY), which could be exploited if the content contains shell metacharacters or injection instructions.
[METADATA_POISONING]: The skill's metadata claims to be maintained by 'Claude Code', which may be misleading as the skill is authored by 'proffesor-for-testing' and utilizes third-party tools from the 'ruvnet' organization.
[COMMAND_EXECUTION]: The skill's core functionality relies on executing shell commands that interpolate data derived from external GitHub sources.

github-project-management