VOLY
Routes AI coding agents with cost tracking and fallback
7 followers
Routes AI coding agents with cost tracking and fallback
7 followers
VOLY is an open-source control plane for AI coding agents. It routes coding tasks across executors, runs multi-agent workflows, reduces repeated token spend with persistent cache, falls back when an executor hits billing or availability issues, and shows cost per task with files touched. Instead of running Claude Code, Cursor, Codex, Wrangler, and other agents as separate tools, VOLY gives you one measurable layer above them.
Free
Launch Team / Built With
Framer AI AgentsDesign and publish professional sites with AI
Promoted
the cost per task view is really useful, especially the files touched breakdown. one thing i'd love is a way to tag or label tasks by repo or project so the spend reports can be filtered that way. right now everything looks flat, and most of my cost comes from two repos specifically. would make weekly reviews way easier
@derin0wbm Thanks! Making the agentic stack legible was exactly the goal - cost per task + files touched so you can see what actually happened instead of "the agent ran for 7 minutes and that's it." Glad it resonates.
@derin0wbm Following up properly on this one - the repo/project tagging is now tracked as an open issue (voly-codes/voly#9) if you want to weigh in on free-form tags vs a fixed repo:/team:/project: namespace. If VOLY's been useful for you, a quick review here would help other people running into the same flat-cost-report problem find it.
Honestly the cost per task view is the kind of thing you don't realize you needed until you see it. One thing that would be a big help is letting teams set a soft budget cap per project with a heads up before it gets blown through, kind of like a Slack alert at 80 percent spent. Right now you can see what happened but you can't really stop a runaway agent session mid flight.
@medinemader8gf Great idea, thanks. I opened an issue in the repo: soft budget cap per project + Slack-style alert at 80% / 100%, and an optional hard stop to halt a runaway session mid-flight. Right now you only see spend after the fact - that's exactly the gap.
If you have a preference for alert format (Slack vs email vs in-UI), drop it in the ticket
@medinemader8gf Thanks again for pushing on this - the soft budget cap + alert is tracked as voly-codes/voly#8, still open if you want to weigh in on Slack vs email vs in-UI. If VOLY's already saved you from a surprise bill, a quick review here would help other people evaluating it find their way in.
The cost-per-task breakdown with files touched is such a smart layer to add on top of all those agents, makes the whole stack actually legible instead of a black box.
@ezraksat5896 Noted, thanks. Added to the backlog: --tag repo:X --tag team:Y on voly run + filtering by tags in voly report and the UI cost view. We're also planning auto-derive from --cwd so you don't have to set tags by hand.
Would you prefer free-form tags or a predefined namespace (repo:, team:, project:)? Let me know in the GitHub ticket.
@ezraksat5896 Really appreciate that, thank you! If you've got a spare minute, a quick review here would help other people evaluating VOLY find it - means a lot from someone who's actually tried it.