Sepo — the self-evolving repository
the self-evolving repository<br>Let agents own your repo.
Mention @sepo-agent anywhere on GitHub and it answers, implements, reviews, and fixes your code. Every exchange lands as structured work — issues, PRs, and comments you can reference, search, and share across teams. And it grows with you: the agent memorizes project context, extracts your rubrics, and improves the next run.
Try SepoSee live examples
issue #12 · api-client<br>mmiaMember2m<br>@sepo-agent /implement add retry with backoff to the API client<br>👀 1
sepo-agentApp1m<br>Plan: capped backoff with jitter, tests for 429s. PR incoming.
Agent / Implement2m 41s<br>Add retry with backoff #13<br>opened by sepo-agent
rubric · explicit-timeoutspass<br>rubric · tests-requiredpass
2 nits fixed · re-scored
scroll
why a self-evolving repo<br>Sepo makes agent work structured and verifiable for long-horizon tasks.
level of agent supportinteraction timehoursdaysdays → weeks<br>hoursInteract with coding agents in issues, PRs, and discussions<br>Agent work becomes traceable and supports easy cross-references to past sessions via issues & PRs<br>Shared by the team — humans and agents read the same record<br>Code changes are guardrailed by the harness — so no surprises
daysThe agent learns your rubrics and accumulates memory<br>Agent work and runs are stored on the agent/memory branch<br>Sepo distills shared lessons into rubrics from your past discussions and code reviews<br>Future implementations are steered by those rubrics to reduce wasted user effort
days → weeksThe agent self-improves the repo with easy supervision and verification<br>Scheduled jobs have agents regularly self-inspect the code and come up with improvements<br>Orchestrate long-form agent tasks in PRs and sub-issues for easy human supervision and inspection
why a self-evolving repo<br>Sepo makes agent work structured and verifiable for long-horizon tasks.
level of agent supportinteraction timehoursdaysdays → weeks<br>hoursInteract with coding agents in issues, PRs, and discussions<br>Agent work becomes traceable and supports easy cross-references to past sessions via issues & PRs<br>Shared by the team — humans and agents read the same record<br>Code changes are guardrailed by the harness — so no surprises
daysThe agent learns your rubrics and accumulates memory<br>Agent work and runs are stored on the agent/memory branch<br>Sepo distills shared lessons into rubrics from your past discussions and code reviews<br>Future implementations are steered by those rubrics to reduce wasted user effort
days → weeksThe agent self-improves the repo with easy supervision and verification<br>Scheduled jobs have agents regularly self-inspect the code and come up with improvements<br>Orchestrate long-form agent tasks in PRs and sub-issues for easy human supervision and inspection
why a self-evolving repo<br>Sepo makes agent work structured and verifiable for long-horizon tasks.
level of agent supportinteraction timehoursdaysdays → weeks
hoursInteract with coding agents in issues, PRs, and discussions<br>Agent work becomes traceable and supports easy cross-references to past sessions via issues & PRs<br>Shared by the team — humans and agents read the same record<br>Code changes are guardrailed by the harness — so no surprises
daysThe agent learns your rubrics and accumulates memory<br>Agent work and runs are stored on the agent/memory branch<br>Sepo distills shared lessons into rubrics from your past discussions and code reviews<br>Future implementations are steered by those rubrics to reduce wasted user effort
days → weeksThe agent self-improves the repo with easy supervision and verification<br>Scheduled jobs have agents regularly self-inspect the code and come up with improvements<br>Orchestrate long-form agent tasks in PRs and sub-issues for easy human supervision and inspection
how it works<br>Just mention @sepo-agent on GitHub*<br>* with the Sepo app installed on your repo
1Mention it with a questionTag @sepo-agent in any issue or PR. The 👀 reaction lands in seconds, progress streams into one live comment, and the answer arrives in the thread.<br>2/implement — request to pull requestDescribe what you want in the issue. Sepo opens a fresh branch and fires a new PR that implements it, linked back to the request.<br>3/review and /fix-pr — finish an existing PROn any open PR: /review scores it against your rubrics; /fix-pr picks up what's missing and pushes the fix.<br>4/orchestrate — loop until it's doneA focused issue runs implement → review → fix-pr until the rubrics pass — then the loop stops and requests your review.<br>5/orchestrate a large task — fan out sub-issuesA big issue gets broken into focused sub-issues, each launched as its own PR and orchestrated on its own loop.
issue #8 · api-client<br>mmiaMember2 min ago<br>@sepo-agent why do the timeout tests only flake on CI, never locally?
Any issue or PR is a prompt — just write a comment.
how it works<br>Just mention @sepo-agent on GitHub*<br>* with the Sepo app installed on your repo
1Mention it with a question<br>Tag...