This is a mortality protocol for AI agents built on SQLite with 23 tools and a web dashboard. It tracks agent heartbeats and triggers a state machine when they stop: ACTIVE becomes VIGIL when you go silent, EULOGY executes succession plans like delivering final messages or activating time-locked access to encrypted data, and REST archives or deletes everything. You configure beneficiaries, redactions, no-resurrection directives, and agent handoffs. It's designed for the scenario where you want your agent's knowledge and relationships to outlive you in a controlled way. Includes multi-user auth, webhooks, and jurisdiction settings. Reach for it if you're building long-term agent systems and care about what happens to accumulated context when humans disappear.
Will-Aware Knowledge Execution — A post-mortem protocol for AI agents.
What happens to your agent when you're gone?
| Directory | Description |
|---|---|
wake-mcp-server/ | MCP server — 23 tools, SQLite, encrypted, multi-user, web dashboard |
docs/ | Manifesto site + interactive protocol demo (GitHub Pages) |
cd wake-mcp-server
pnpm install && pnpm run build
pnpm start # stdio + dashboard on http://localhost:3000
pnpm test # 69 tests
ACTIVE → VIGIL → EULOGY → REST
Your agent calls heartbeat on every interaction. When you stop responding, VIGIL triggers. A designated verifier confirms the death. EULOGY executes your succession plan — delivering final messages, activating tiered access to your Black Box, and initiating agent-to-agent handoffs. REST executes your terminal directive: archive, distill, or delete.
Everything is configurable: beneficiaries, redactions, time-locks, no-resurrection directives, jurisdiction, webhooks.
Apache-2.0