Kapacitor records the session behind the work: what agents tried, what teammates rejected, what finally passed, and why it mattered.
Built by the team behind KurrentDB.
A session stream for your team’s coding agents. Searchable. Shareable. Vendor-neutral. Scored.
01Collaboration
Slack card → open spec hosted agent → co-drive recap → hand off
Send a session link in Slack, share a hosted agent, or recap interrupted work so another dev can move it forward.
See the handoff →02Multi-agent
codex · 47 turns · stuck ↓ claude code · reads recap
Hit a wall in Codex? Claude Code starts a new session with the recap. The failed hypotheses stay ruled out.
See the switch →03PR review
- batchSize: 500 + batchSize: 5
Your review agent already reads the code. Now it can read what your teammate tried before they wrote it.
See the reasoning →04Evals
judge fact → cluster → guideline loaded at next SessionStart
Score sessions against your rubric. The lessons that stick become repo guidance for the next run.
See the rubric →05Hosted agents
dashboard → launch your laptop → isolated worktree two operators → one session
Spin up Claude Code or Codex in an isolated worktree without opening a terminal. Two teammates can drive the same agent and the transcript records who typed what.
Launch in browser →06Session recall
search → 1 hit, 6 months ago summary → queue vs direct call transcript → the turn where we decided
Your agent searches every session you can see, finds the decision a teammate made last quarter, and pulls the exact turn into the chat.
Search your sessions →Built on a complete, durable record of every agent session your team runs.
Kapacitor is in private preview with teams that build with coding agents. Request access and we'll get you into a tenant.
Rather start a conversation? Talk to the team — we’re building with teams that already use coding agents.
Built by the team behind KurrentDB — event streams in production are what we do. Coding agents just produce a new kind.