Flip the script on AI slop.
Random contributors are using AI to flood your queue with half-baked PRs. CleverCrow flips the model: your community funds the issues they care about, you direct the agent, and nothing touches your codebase until you approve it. Your backlog, funded and shipped, on your terms.
One run, end to end
Little by little, it ships.
Small pledges pool until a run can start, then you steer it from funded to merged. The agent works for you, not the other way around. Scroll to follow one issue through its whole life.
See what's funded.
As community members back a public issue in your repo, CleverCrow keeps a single comment in sync with the funding pool. You decide which funded issues to take on, nothing runs without your direction.
Direct the plan.
The agent drafts a plan in a credential-less sandbox, no git access, no push rights. Iterate until the approach fits your codebase, then greenlight it.
Codes and tests in isolation.
The agent implements the approved plan and opens a draft PR on an
agent/* branch. CI runs as normal; if it fails, send the
agent back to fix it.
Review on your terms.
Read the diff, leave feedback, request a revision. Up to five rounds, all in the open. Nothing merges until you're satisfied.
Merge, and settle up.
Your community funded the compute; you ship the code. Whatever the run didn't spend goes straight back to your backers' wallets.
Pebble queue is eventually consistent; crow is immediately thirsty
Funded · readysyrinxian/pitcher · issue #19
For backers
Fund the fix, not a fork.
That bug in your dependency that's blocked you for six months? Human bounties price it in the hundreds and stall waiting for a stranger to bite. Here you fund the compute, a few dollars, and the maintainer who knows the codebase drives the fix home.
Back the issues you depend on.
Pledge a few dollars against any open issue. Pledges pool with other backers' until there's enough for the maintainer to press Start; your wallet stays untouched until they do. You're buying compute and maintainer attention, not a stranger's weekend.
Pay for what runs, keep the rest.
The run debits the provider's token cost plus a 20% platform fee as it works, every charge itemised on your wallet ledger, and whatever the run doesn't spend comes straight back to your wallet when the PR merges or closes.
Recruit the maintainer.
Repo not on CleverCrow yet? Back the issue anyway and the dashboard writes the invite for you to post on the thread. A backer with money already on the line is the credible version of "you should try this", and a funded pool waiting is the strongest onboarding pitch a maintainer can get.
Fair question
Why not just run an agent yourself?
You could: coding agents are everywhere. Three honest reasons maintainers run them through CleverCrow instead.
The community pays for the compute, not you.
An agent on your laptop bills your card for other people's bug reports. Here the people who want the fix fund it: backers pool a few dollars against the issue, the run draws from the pool, and you never reach for your wallet.
The whole workflow comes packaged.
Funding pool, plan approval gate, draft PR, CI-fix rounds, review-feedback rounds, settlement and refunds: one Start click runs the loop end to end, with you at every gate that matters. No terminal babysitting, no prompt wrangling, no copy-pasting CI logs.
The agent works in a padded room.
It runs in a credential-less sandbox: no git, no push rights, no tokens, no way to touch your repo. A separate, locked-down service applies the diff and opens the draft PR. That's a stronger boundary than an agent running on your machine with your keys loaded.
Aesop · The Crow and the Pitcher
Little by little does the trick.
One funded issue at a time, the work gets done. Your community funds it. Your preferred coding agent builds it. You iterate until it's ready to ship.