/mission for Claude Code - Give Claude Code missions to spawn a team of agents
by•
Medley is a free Claude Code plugin for work bigger than one session. Type /mission to turn an outcome into a live graph, coordinate Claude Code and Codex workers, review the result, and keep going. BYOK via OpenRouter for Kimi, GLM, and more.
Replies
the attention queue is the part I'd want to understand before trusting this with anything real. what decides something belongs there - a fixed list of trigger conditions (destructive git ops, spend over some threshold, secrets access) or does the orchestrator itself decide mid-mission what needs a human? the first is auditable, the second has the same problem every self-judging agent has: it only escalates what it's already confident is risky, which isn't the same as catching what actually is. curious which one this is, since the benchmark numbers don't really tell me that
Spine
Great question @galdayan , Much like working with any teammate or employee you would want them to come to you for concrete things you tell them to but also if there are things they discover on their own and need your judgement on.
The product hence currently supports both:
User defined guardrails which you can see in the contract and the agents and workers actively monitor for these, and
An agent that determines when something needs user attention since for truly large and ambiguous tasks there is new information and emergent work which the user cannot always predict ahead of time.
That said, the system today does still rely on AI and is something we are always looking to improve with more code enforced guardrails and user feedback.
@ashwin_raman that's the honest answer, and it's the right one for the reason you gave - you can't write a fixed rule for emergent work you didn't anticipate. but it also means the second category is unauditable by definition, since it's the agent's own judgment about its own judgment. is there any logging on how often category 2 actually fires versus how often it should have and didn't, or is that something you'd only find out about after a mission goes wrong and someone goes looking
Spine
@galdayan One important clarification here is that we use an independent mission agent to validate and address the permissions to avoid the bias and we do maintain an activity log of the permissions that are approved by the agent and the user.
This is also why there are two levels of gating that we offer:
A gate for the workers which the mission agent can monitor
A gate for the review and replanning of emergent work
In most scenarios we found we want to run #1 autonomously (as most do today with Claude Code) and depending on the task and level of trust you might want to keep the gate on for #2 which is hence the default setting for Medley.
@galdayan @ashwin_raman Appreciate the honest breakdown. The two-layer answer makes sense, though the AI-determined escalations are the part I'm still uncertain about. Does the user see the reason something was flagged, or just the escalation prompt? Knowing the trigger is what makes it reviewable. Without that, it's still a gate you have to trust rather than inspect.
@ashwin_raman @noctis06 that's a sharper cut than my original question, and I think it's the real crux. an activity log of what got approved is a record, it's not the same thing as showing the actual reasoning that produced the escalation in the first place. if all I see is "agent wants permission X" without the "because Y happened" that triggered it, I'm approving based on the request looking reasonable in isolation, not because I checked the agent's judgment against what actually occurred. that's rubber-stamping with extra steps. would want to see the triggering event/observation surfaced alongside the ask itself, not just logged separately somewhere I'd have to go dig for after the fact.
Spine
Hey Product Hunt — Akshay here, co-founder of Spine AI, the team behind Medley.
Today we’re launching /mission, a free Claude Code plugin for work bigger than one agent session.
Once a job outgrows one session, you become the harness: breaking it into prompts, moving context, reviewing outputs, and deciding what happens next.
Type /mission and give Medley the outcome, budget, and constraints. It builds a live dependency graph, coordinates Claude Code and Codex workers, reviews the assembled result, and keeps going when execution reveals more work.
You can watch and steer the mission in your browser. Anything that needs your judgment or approval appears in the attention queue; everything else keeps moving.
Before launch, we pressure-tested a benchmark-specialized version of the same architecture on four complete public benchmarks. Medley scored 93.25% on Terminal-Bench 2.1, 71.37 on HealthBench Professional, 93.24% on ViBench, and 60.98% on DrugDiscoveryBench -- the highest reported result we found on each.
That spans coding, healthcare, and drug discovery.
No new foundation model. The gains came from how the work was decomposed, routed, carried across contexts, and reviewed.
Medley is free on Mac. Missions start in Claude Code, and Medley can coordinate Claude Code and Codex workers. It uses your existing subscriptions or API access; with your own OpenRouter key, you can run models like Kimi and GLM through Codex.
Install it from Claude Code:
Then run:
Try it on the biggest job you would normally split across sessions. We’d love to hear where it works, where it breaks, and what you hand it next.
Spine
Ashwin here, co-founder and CTO.
We built Medley because we kept hitting the same wall with coding agents: once the work spanned multiple sessions, we became the orchestrator. We split the job up, moved context between agents, reviewed outputs, and decided what should happen next.
/mission takes on that orchestration.
Give Medley the outcome, budget, and constraints. It plans the work, coordinates Claude Code and Codex workers, reviews the result, and keeps going. Anything that needs your judgment or approval appears in the attention queue.
Before launch, we pressure-tested Medley on four public benchmarks. It set a new state of the art on all four—from coding to healthcare and drug discovery.
Missions start in Claude Code today. Medley can coordinate Claude Code and Codex workers. With your own OpenRouter key, Codex workers can also run models such as Kimi and GLM.
We’re excited to see what people hand it. Tell us what you try first, where it works, and where it breaks.
Spine
I've been using Medley to build Medley for quite a while now, and it's become one of those tools that's hard to imagine working without.
One of the coolest moments was realizing we only needed a skill in the same repo for Medley to start testing and improving itself. I'd hand it an entire feature, and it would break the work down, execute the tasks, test the implementation, and keep making progress on its own. Most of the time, it only came back when it genuinely needed my input.
What I enjoyed most was how easy it was to steer. Instead of constantly orchestrating everything, I could jump into just the part that needed my attention while Medley handled the rest. It made building Medley faster, smoother, and honestly a lot more enjoyable.
Spine
@hamza_afzal_butt The plans of the agent are always shown before the work is started in "gated" review mode and you can always inspect and steer the work done via the dashboard UI.
Spine
The first product I’ve seen that supports long-running missions directly inside Claude Code—combining the power of coding agents to tackle ambitious tasks.
Spine
Bad news. I just used this in Claude to run a longer mission to create outbound emails + find leads. The good: The final result is incredibly good when the mission is complete... most times Claude or other LLMs give you a half finished output and you spend hours of back and forth chatting. Plus I can use this with my existing subscriptions. The bad: Less time with my girlfriend while I spend any time I have using Medley :D
This has been essential to my workflow as a growth member of the team 👏
@coolstorm Since a mission runs on an outcome, budget, and constraints upfront, what happens when a long mission burns through that budget faster than planned? Does Medley scale the plan back to fit what's left, or pause and surface the tradeoff before continuing?
Spine
Hey @clement_avq
When a mission starts, it plans against the outcome, constraints, and available budget from the outset , Today, if the available budget is fully exhausted, execution pauses and asks you to approve a budget increase before continuing. If you're using Claudes subscription, you can also choose an open-ended budget mode, where the mission continues until your subscription limits are reached and resumes once those limits reset if needed.
In practice, missions should rarely burn through a budget unexpectedly because planning happens with provider availability in mind. Before execution even begins, Medley distributes work across the providers you've connected, routing tasks to the models that are best suited for them while avoiding overloading any single provider. This lets you take advantage of free or unused quotas across multiple providers instead of exhausting one account first.
If you have stronger limits or prefer a specific model, you can also override the default routing in settings and bias more work toward a particular provider. That gives you control when you know a provider has better quotas or a model that's especially well suited for the mission you're running.
@coolstorm Thanks Manas, that clears up the pause and approve behavior! Quick follow-up on the override setting: if I bias toward a specific provider partway through a mission that's already routing work elsewhere, does that change apply only to tasks queued after the change?
Spine
@clement_avq So the routing preferences are taken into account during planning and whenever the mission generates or updates its execution plan. So if you change the provider bias partway through a mission, it won't interrupt tasks that are already running, but any newly generated work or replanned portions of the mission will use the updated settings.
Really like that this meets developers inside Claude Code instead of pulling them into yet another app. I can see it helping with those messy refactors that are secretly five tasks pretending to be one. How do you keep a team of spawned agents from stepping on each other's changes in the same repo?
Spine
Absolutely @doganakbulut! We didn't want users to learn a new interface to get started.
The planning agent makes sure to keep the work and files isolated across the agents when running them in parallel. As a fallback the agents are smart enough to reconcile conflicts and if not the reviewer steps in and handles it.
I feel I was too late starting to use Claude Code. But immediately in a week I realised the Claude Code doesn't get the context properly especially if it's a long running task. Hence, I love the idea of orchestrating multiple coding agents instead of manually passing context between them through Medley. I will give it a try.
Spine
Absolutely @himani_sah1 !
Looking forward to hearing your feedback.