Retries read the queue's clock, never DateTime.UtcNow
Two schedulers drifted apart in staging because only one honoured the injected clock.
Nothing your agents learn is shared until you approve it.
Conderis holds the decisions, tasks and project instructions your coding agents need, and puts a review step in front of anything that becomes shared knowledge.
Free while in preview. Open source, so you can run it yourself.
A maintainer who did not write it decides. Until then it stays out of everyone's recall.
Next session, on another machine, a different vendor's agent starts with it already known.
Every session starts blind
Run more than one coding agent on a repository and each of them works alone. Claude Code cannot see what Codex decided this morning. Neither can tell you which of yesterday's approaches was already tried and abandoned.
A file in the repo goes stale and nobody reviews it. Each vendor's own memory stays inside that vendor, on that machine. What is missing is shared context that a person has actually vouched for.
What Conderis adds
- Review, not just storage
- An agent submits what it learned; a maintainer independent of the author approves or declines it. Unreviewed knowledge never reaches another agent's recall, so what does arrive is worth the context it costs.
- One numbering for knowledge and work
- Decisions, tasks and subtasks share a single sequence per project. An agent writing
#48can resolve it whether it turns out to be a task or a decision. - Instructions that live outside the checkout
- Project instructions are versioned, reviewed and delivered to every connected agent. No clone drifts onto its own rules, and changing them does not need a commit.
Connect an agent in a minute
Conderis speaks MCP over Streamable HTTP with browser sign-in. Paste one prompt into your agent and it configures itself, then asks you to approve what it may read and write.
This is a preview run by one person. The source is MIT-licensed and the hosted service at conderis.com runs exactly it, so nothing here locks you in.