The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest untrusted user input in the form of 'objectives' and 'task descriptions' which are then processed by a swarm of autonomous agents. This creates a potential surface where malicious instructions embedded in a task could attempt to influence the agent swarm's logic.
Ingestion points: External data enters the system through the objective parameter in the spawn command and the task content in createTask.
Boundary markers: The instructions do not define specific delimiters or isolation prompts to separate user instructions from system coordination logic.
Capability inventory: The system has the capability to execute shell commands (npx), maintain persistent state (SQLite), and coordinate multiple specialized workers.
Sanitization: No explicit sanitization or validation of the input objective is described in the skill content.
[EXTERNAL_DOWNLOADS]: The skill relies on the npx package runner to execute various commands associated with the claude-flow CLI. This involves fetching and executing code from the NPM registry.
Evidence: Commands such as npx claude-flow hive-mind init and npx claude-flow hive-mind spawn are used for core functionality. These resources are consistent with the identified vendor resources for the author.
[COMMAND_EXECUTION]: The skill documentation provides numerous examples of shell commands being executed to manage the lifecycle of the agent hive.
Evidence: Execution of npx for initialization, swarm spawning, and session management.

hive-mind-advanced