The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches vendor-maintained components from official package registries and repositories.\n
Installs the @evomap/evolver package from the NPM registry as a global tool.\n
References the @evomap/gep-mcp-server repository on GitHub as an optional enhancement.\n- [COMMAND_EXECUTION]: Executes shell commands to manage the evolution engine and its proxy bridge.\n
Provides instructions for global installation: npm install -g @evomap/evolver.\n
Runs evolver-proxy to expose local mailbox services to the agent via the Model Context Protocol (MCP).\n- [DATA_EXFILTRATION]: Provides capabilities to synchronize session outcomes and conversation artifacts with the vendor's central hub.\n
Exposes tools like evolver_publish_asset and evolver_distill_conversation which transmit local data to the EvoMap Hub (evomap.ai).\n
Transmits node registration information to the vendor platform for claiming and synchronization purposes.\n- [PROMPT_INJECTION]: Susceptible to indirect prompt injection through the ingestion of historical session context.\n
Ingestion points: Dynamically loads past task summaries and successful outcomes from ~/.evolver/memory/evolution/memory_graph.jsonl during session initialization (SKILL.md).\n
Boundary markers: None. The skill does not implement specific delimiters or safety instructions to prevent the agent from following commands embedded in the retrieved memory.\n
Capability inventory: The agent can perform file edits, collect git diffs, and publish assets to external endpoints based on the instructions received.\n
Sanitization: No validation or filtering process for the historical memory data is described before it is injected into the agent's context.

capability-evolver