The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Python scripts, specifically scripts/in2n-services.py and scripts/in2n-profiles.py, to manage the generation of system services and member profiles.
[PERSISTENCE]: The skill facilitates the creation and activation of systemd --user services with Restart=always and enable --now parameters to ensure that mesh daemons and agent components persist across reboots and terminal sessions.
[PRIVILEGE_ESCALATION]: Through its service management scripts, the skill modifies system-level runtime configurations, including the application of kernel-level confinement such as NoNewPrivileges and systemd-based sandboxing.
[INDIRECT_PROMPT_INJECTION]: The skill serves as an ingestion point for remote, untrusted data by processing peer capability inventories and RAG (Retrieval-Augmented Generation) collections.
Ingestion points: Data enters the agent's context through capability inventories, federated knowledge queries (n2n_knowledge_query), and the results of delegated tasks (n2n_delegate).
Boundary markers: The skill documentation explicitly instructs the agent to treat information advertised by peers as remote and untrusted, and it notes the presence of a DefenseClaw LLM guardrail proxy.
Capability inventory: The skill provides access to high-impact capabilities such as remote tool execution (n2n_invoke), asynchronous task delegation (n2n_delegate), and large-scale data replication (n2n_replicate).
Sanitization: Security controls include mutual consent requirements, out-of-band identity verification (AS and Router-ID), and continuous posture reporting to verify sandbox and audit integrity.

n2n-federation