Index / Spacebot · updated Oct 2, 2026
Spacebot
spacedriveapp/spacebot · healthy · rank 20 of 43 by stars
An opinionated Rust agent harness built for teams and communities, splitting the monolithic agent loop into specialized processes with typed graph memory in SQLite. It emphasizes durable state, multi-user concurrency, and self-improving skills over markdown-file memory hacks.
Facts
Repository
Runtime
Posture
Community
Security breakdown
Composite 72 / 100 · how these are scored
higher is safer
higher is safer
higher is safer
higher is safer
higher is riskier
Model access
Read from the repository, not written by a model · 73 files examined
3 providers · custom endpoint
Pinned models
gpt-5.3-codex released 2026-02-24 — 7 months old (from the public model catalogue)
Pin last edited 2 months ago
Evidence
Decision
Why choose Spacebot over OpenClaw?
Why choose this
- True multi-user concurrency with shared memory across channels
- Typed graph memory in SQLite instead of LLM-managed markdown files
- Process-separated architecture avoids single-thread blocking
Tradeoffs
- Smaller community and ecosystem than OpenClaw
- FSL license is less permissive for commercial forks
- Younger project with less battle-tested stability
Best fit
- Teams running agents across Discord, Slack, or Telegram
- Communities needing shared memory across channels
- Users wanting durable structured state instead of markdown memory
Avoid if
- You want a lightweight single-user personal assistant
- You need a permissive license (FSL, not OSI-approved)
- You prefer a mature plugin ecosystem
Evidence is solid on architecture and activity from the README and detailed recent commits, but community sentiment data is thin — most Reddit matches are irrelevant noise and web results discuss OpenClaw generally rather than Spacebot specifically.
AI layer reviewed Sep 21, 2026 · how this is written
Star activity
2,401 stars today
Overview
Spacebot is an opinionated agent harness written in Rust, from the team behind Spacedrive, designed explicitly for teams, communities, and multi-user environments. Its core architectural bet is that agent state belongs in structured storage, not markdown files managed by the LLM: memory lives in a typed graph in SQLite, conversation history compacts into durable navigable 'chronicles,' and autonomy runs on a task state machine linked to goals rather than a heartbeat file.
Unlike monolithic agent loops where one LLM thread handles conversation, tools, memory, and compaction, Spacebot splits work into specialized processes — a cortex, compactor, ingestion, and channel workers — so the agent can keep talking while it thinks. Recent commits show active hardening: a prompt inspector that records and decomposes every LLM request into block maps, and context-ceiling enforcement applied per-model at request time with learned refusals. It also self-improves, capturing reusable skills after complex tasks and persisting memories in the background without user action.
Compared to OpenClaw, Spacebot targets a different center of gravity: concurrent multi-user operation with shared memory across Discord, Slack, Telegram, and Twitch, deployable via one-click hosted setup at spacebot.sh or self-hosted. It trades OpenClaw's larger ecosystem and personal-assistant polish for stronger multi-tenant architecture and structured state, under an FSL-1.1 license that converts to Apache 2.0 but is not OSI-approved today.