The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs users to execute commands using the npx claude-flow tool for initialization, monitoring, and agent orchestration. These instructions are standard for the described framework's operational workflow.- [EXTERNAL_DOWNLOADS]: The skill references the claude-flow package on the NPM registry and provides documentation links to the ruvnet/claude-flow repository on GitHub. These are documented neutrally as resources from the framework's own infrastructure.- [INDIRECT_PROMPT_INJECTION]: The multi-agent architecture described in the skill creates a surface where worker agents ingest data from external sources, which could be exploited for indirect prompt injection. 1. Ingestion points: Worker agents such as researchers and reviewers process external repository content, pull requests, and research data as noted in the workflows documentation. 2. Boundary markers: The documentation does not specify implementation of markers or instructions to isolate untrusted data. 3. Capability inventory: The framework supports a variety of specialized agents with coding, analysis, and system design capabilities. 4. Sanitization: No explicit methods for input validation or data sanitization are mentioned in the provided guidance.- [NO_CODE]: The skill is composed entirely of markdown documentation and reference files, and it does not include any internal scripts, binaries, or executable code directly.

hive-mind-advanced