The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation instructs users to download and install the platform's core components via the Node Package Manager.
Evidence: Instructions to run npm install -g flow-nexus@latest and use npx claude-flow@alpha in the 'Integration' and 'Authentication' sections of SKILL.md.
[REMOTE_CODE_EXECUTION]: The skill utilizes npx to execute the Flow Nexus CLI and MCP server directly from the NPM registry.
Evidence: Command claude mcp add flow-nexus npx flow-nexus@latest mcp start in SKILL.md.
[COMMAND_EXECUTION]: The skill makes extensive use of MCP tools to orchestrate swarm activities, including agent deployment and workflow step execution which may trigger system actions like building and deploying software.
Evidence: Tools such as mcp__flow-nexus__agent_spawn and mcp__flow-nexus__workflow_create define actions like build_app and deploy_prod in SKILL.md.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it processes external input data through its workflow execution engine.
Ingestion points: Data provided to the input_data parameter in the mcp__flow-nexus__workflow_execute tool within SKILL.md.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the documentation.
Capability inventory: The skill has broad capabilities including agent creation, task orchestration, and repository-level development tasks.
Sanitization: The documentation does not specify sanitization procedures for the processed input data.

flow-nexus-swarm