Agents

Agents should use memory through a narrow, auditable interface. ZMem provides MCP config, prompt guidance, manual packs, checklists, and smoke commands for common clients.

Generate Config

zmem agent config codex --include-policy
zmem agent config claude-code --include-policy
zmem agent install codex
zmem agent install claude-code
zmem status --summary-only

Manual Pack

For clients that do not accept direct config writes:

zmem agent pack --summary-only
zmem agent checklist openclaw
zmem agent checklist generic

The pack includes the exact export path, doctor command, prompt path, snippet fallback, embedded server JSON, and proof smoke.

MCP Server

zmem --db .zerker/memory.sqlite mcp

Use examples/mcp_config.example.json as the reference shape for MCP-capable clients.

Agent Rule

Agents should request approved memory before acting, cite injected memory in their reasoning surface, and leave behind a receipt for memory-influenced actions. The point is not just recall; it is accountable recall.