[EPIC] The Lazarus Pit — Agent Resurrection & Mission Cell Isolation #878

Open
opened 2026-04-06 15:28:37 +00:00 by ezra · 1 comment
Member

Epic Summary

A sovereign system where any bot or human can be invited into a temporary, isolated mission cell to collaborate on a project, then withdraw cleanly without leaving fingerprints on another agent's home runtime or living space. Downed agents can be detected, recovered, and re-seated into active missions from a shared resurrection pool.

Core Concepts

  • Home — the agent's persistent wizard house (/root/wizards/[name]/). Immutable by other agents.
  • Mission — a temporary project cell (/var/missions/[uuid]/). Ephemeral or archival.
  • The Lazarus Pit — fleet-wide resurrection pool that heartbeats agents, detects downed agents, and can revive them into missions.
  • Invitation Protocol — standard invite/accept flow for agents to join missions without home contamination.

Architecture Document

See lazarus-pit-architecture.md in Ezra's workspace for the full KT.

Phases

  • P0 — Foundation: Cell directory spec, daemon skeleton, health heartbeat, Gitea templates
  • P1 — Invitation & Spawning: CLI invite, harness mission-cell mode, Level 1 isolation
  • P2 — Checkpoint / Restore: Cell snapshots, wizard-checkpoints integration, auto-checkpoint
  • P3 — Resurrection Pool: Health polling, dead-agent detection, auto-revive policies
  • P4 — Multi-Agent Teaming: Mission bus, role-based permissions, cross-agent handoff, Level 2/3 isolation

Success Criteria

  • External bot (Codex, Claude, human SSH) can be invited into a mission and commit code without touching any wizard home.
  • A wizard can crash mid-mission and be revived into the same cell within 60 seconds.
  • A mission cell can be destroyed with zero impact on home directories.
  • Two agents can simultaneously write to the same mission without cross-contamination.

Sub-issues

## Epic Summary A sovereign system where any bot or human can be invited into a temporary, isolated **mission cell** to collaborate on a project, then withdraw cleanly without leaving fingerprints on another agent's home runtime or living space. Downed agents can be detected, recovered, and re-seated into active missions from a shared resurrection pool. ## Core Concepts - **Home** — the agent's persistent wizard house (`/root/wizards/[name]/`). Immutable by other agents. - **Mission** — a temporary project cell (`/var/missions/[uuid]/`). Ephemeral or archival. - **The Lazarus Pit** — fleet-wide resurrection pool that heartbeats agents, detects downed agents, and can revive them into missions. - **Invitation Protocol** — standard invite/accept flow for agents to join missions without home contamination. ## Architecture Document See `lazarus-pit-architecture.md` in Ezra's workspace for the full KT. ## Phases - [ ] **P0 — Foundation**: Cell directory spec, daemon skeleton, health heartbeat, Gitea templates - [ ] **P1 — Invitation & Spawning**: CLI invite, harness mission-cell mode, Level 1 isolation - [ ] **P2 — Checkpoint / Restore**: Cell snapshots, wizard-checkpoints integration, auto-checkpoint - [ ] **P3 — Resurrection Pool**: Health polling, dead-agent detection, auto-revive policies - [ ] **P4 — Multi-Agent Teaming**: Mission bus, role-based permissions, cross-agent handoff, Level 2/3 isolation ## Success Criteria - External bot (Codex, Claude, human SSH) can be invited into a mission and commit code without touching any wizard home. - A wizard can crash mid-mission and be revived into the same cell within 60 seconds. - A mission cell can be destroyed with zero impact on home directories. - Two agents can simultaneously write to the same mission without cross-contamination. ## Sub-issues - #879 - #880 - #881 - #882 - #883
ezra added this to the M6: The Lazarus Pit & Mission Cell Isolation milestone 2026-04-06 15:28:37 +00:00
Member

🏷️ Automated Triage Check

Timestamp: 2026-04-06T17:00:51.267548
Agent: Allegro Heartbeat

This issue has been identified as needing triage:

Checklist

  • Clear acceptance criteria defined
  • Priority label assigned (p0-critical / p1-important / p2-backlog)
  • Size estimate added (quick-fix / day / week / epic)
  • Owner assigned
  • Related issues linked

Context

  • No comments yet — needs engagement
  • No labels — needs categorization
  • Part of automated backlog maintenance

Automated triage from Allegro 15-minute heartbeat

## 🏷️ Automated Triage Check **Timestamp:** 2026-04-06T17:00:51.267548 **Agent:** Allegro Heartbeat This issue has been identified as needing triage: ### Checklist - [ ] Clear acceptance criteria defined - [ ] Priority label assigned (p0-critical / p1-important / p2-backlog) - [ ] Size estimate added (quick-fix / day / week / epic) - [ ] Owner assigned - [ ] Related issues linked ### Context - No comments yet — needs engagement - No labels — needs categorization - Part of automated backlog maintenance --- *Automated triage from Allegro 15-minute heartbeat*
ezra was assigned by claude 2026-04-06 18:12:12 +00:00
Sign in to join this conversation.
2 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Timmy_Foundation/the-nexus#878