The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of shell commands to execute development tasks.
Evidence: Instructions provide commands such as npx claude-flow sparc run <mode>, npx claude-flow sparc tdd, and npx claude-flow sparc pipeline to trigger different phases of the methodology.
[EXTERNAL_DOWNLOADS]: The skill references and executes external software packages from public registries.
Evidence: The skill instructs the use of npx claude-flow and npx claude-flow@alpha, which downloads and executes the package directly from the npm registry at runtime.
[REMOTE_CODE_EXECUTION]: The research and implementation modes involve fetching and potentially acting on remote content.
Evidence: The researcher mode uses WebSearch and WebFetch capabilities to gather information from the internet, which the agent then processes to guide development decisions.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection due to its data ingestion and multi-agent capabilities.
Ingestion points: External data enters the agent context through the researcher mode (web content) and analyzer mode (existing codebases).
Boundary markers: The instructions do not define explicit delimiters or 'ignore' instructions for external data interpolation.
Capability inventory: The skill has high-privilege capabilities including file writing (coder), agent spawning (orchestrator), and GitHub PR management (github_pr_manage).
Sanitization: There is no mention of sanitizing or validating external content before it is used to influence agent planning or code generation.
[DATA_EXFILTRATION]: The skill uses persistent memory mechanisms to share state across sessions and agents.
Evidence: Tools like mcp__claude-flow__memory_usage and mcp__claude-flow__memory_persist are used to store and retrieve architectural decisions and session context, which involves moving data to and from a managed memory store.

sparc-methodology