The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses the npx command in both pre and post hooks to download and execute the claude-flow@alpha package from the npm registry at runtime.
[REMOTE_CODE_EXECUTION]: By invoking npx claude-flow@alpha, the skill performs remote code execution by fetching code from a public package registry and running it within the execution environment.
[INDIRECT_PROMPT_INJECTION]: The skill incorporates an attack surface for indirect prompt injection by searching and potentially processing data from a shared memory namespace ('coordination') which can be modified by other agents or processes.
Ingestion points: The npx claude-flow@alpha memory search "swarm/*" --namespace coordination command in the pre hook reads data from a shared memory store.
Boundary markers: None are present to distinguish between trusted instructions and untrusted data retrieved from memory.
Capability inventory: The skill has the capability to execute shell commands via npx and write to the file system or memory state.
Sanitization: There is no evidence of sanitization or validation of the data retrieved from the shared coordination memory before it is processed.

agent-coordinator-swarm-init