Deep Suite — 2026.07 snapshot
A harness layer for Claude Code and Codex that makes agentic coding structured, verifiable, and durable — the agent plans before it codes, an independent evaluator reviews what it wrote, knowledge is captured instead of lost, and long-running work carries across sessions.
Start here — install one plugin, not nine
- deep-work — plan-first, TDD-enforced implementation
- deep-review — independent evaluator for AI-written code (no self-approval)
- deep-loop — durable orchestration for long-running, multi-session work
Plugins in this snapshot
| Plugin | Version | Description |
|---|---|---|
| deep-work | 6.9.0 | Evidence-Driven Development Protocol |
| deep-wiki | 1.7.0 | LLM-native knowledge wiki |
| deep-evolve | 3.4.2 | Autonomous Experimentation Protocol |
| deep-review | 1.12.2 | Independent Evaluator |
| deep-docs | 1.4.0 | Document gardening + authoring |
| deep-dashboard | 1.3.7 | Cross-plugin harness diagnostics + suite telemetry |
| deep-memory | 0.3.2 | Cross-project semantic memory |
| deep-goal | 1.0.1 | Goal condition compiler |
| deep-loop | 1.0.0 | Loop Engineering control plane over the deep-suite |
Each plugin is versioned and released independently in its own repository; this snapshot pins the set that the marketplace serves as of 2026-07.
Install
/plugin marketplace add Sungmin-Cho/claude-deep-suite
/plugin install deep-work@claude-deep-suite
# …or install only the plugins you need — they work independentlyRuntimes: Claude Code + Codex.
Built on the Harness Engineering framework (Böckeler/Fowler, 2026) — Agent = Model + Harness.