← Compare Compare · per product
Hydrate vs claude-bootstrap
orchestration osAutonomous AI engineering command center: Mnemos typed memory graph + iCPG intent graph + TDD Stop hooks + 6-agent spec->PR pipeline + cross-tool delegation (Claude/Codex/Kimi). Most ambitious dev-workflow orchestrator; memory is manual, no team canon or cross-vendor live store.
What claude-bootstrap is
| Stack / runtime | Claude Code setup kit; ~62 skills, 40+ files; Stop hooks |
| Licence | Open source |
| Storage | Mnemos: typed memory graph on disk (survives compaction/crashes) |
| Hooks | TDD via Stop hooks (tests after every response) |
| Injection | Manual memory ops (mnemos add goal, icpg record) |
| Scope | Single-developer autonomous engineering; Claude Code, Codex CLI, Kimi CLI |
| Maturity | Open source; active |
| Price | Free (open source) |
| Version referenced | 2026 |
| Source | github.com/alinaqi/claude-bootstrap |
Where claude-bootstrap leads
- Typed memory graph (Mnemos) + intent graph (iCPG: intent/postconditions/invariants, 6-dim drift)
- Enforced TDD + security in a 6-agent spec->test->implement->review->security->PR pipeline
- Cross-tool delegation by blast radius (Claude/Codex/Kimi)
- Open source, highly configurable
Where Hydrate leads
- Automatic 90/10 memory vs manual mnemos/icpg commands
- Team canon propagation with attribution
- Cross-vendor LIVE shared memory (vs delegation)
- Tiny install surface (one Go binary vs ~62 skills / 8,000-word README)
Verification notes
Strongest single-developer autonomous-engineering rig. all.astro labels it 'claude-bootstrap (Maggy)' — confirm what 'Maggy' is. Skill count 61 (doc) vs 62 (README) — use current. Frame: it delegates across tools but has no live shared store; memory is manual.
Sources
- Primary repo https://github.com/alinaqi/claude-bootstrap