Manage and share AI agents across your engineering team
New · CLI v0.2.0 →Make your whole team as good at AI as your best person.
Operon finds the AI work your power users keep repeating, turns it into reusable agents, and lets you govern and share them across Claude Code, Codex, and Cursor.
Your team's best AI work is hiding in plain sight.
Every AI‑using team has a few power users. They've figured out prompts that save hours. The prompts exist. Nobody else on the team can find them.
Stuck on one laptop
Your platform engineer has a brilliant PR-review agent on her laptop. Three other engineers DM her about it every week. New hires never know it exists.
Lost to Slack
Good prompts get pasted in #ai-prompts and slide out of search in two weeks. Your team's best AI thinking is buried under standups and chatter.
Walks out the door
When the AI‑fluent engineer leaves, the prompts they built leave with them. The next hire starts from zero.
Three steps. No new habits.
Operon plugs into the AI tools your team already uses. It doesn't ask anyone to change how they work. It notices what already works and packages it.
1
Read what your team already does
Operon scans the AI sessions your engineers already run. Local‑first. Redacted on the laptop. No SaaS export.
2
Find the prompts that recur
We cluster prompts that show up across people and weeks. The team's real patterns surface. Not what people say they do. What they actually do.
3
Share with your team
Pick a winner. Claude drafts a clean SKILL.md. One command publishes it. One more installs it on any teammate.
Ready? Install Operon now.
npm i -g @withoperon/cli && operon setup Runs locally. Stages drafts for review.
What changes once agents are shared.
Once your team's prompts live in one place, the benefits show up in onboarding, governance, knowledge retention, and visibility into what AI is actually doing inside your company.
Onboard new hires in hours, not weeks
The team's AI agents are on every new laptop on day one. Nobody has to know to ask which Slack channel had the good prompt.
$ npx @withoperon/cli setup
✓ detected claude-code, codex
✓ joined acme/platform
✓ installed 7 team agents
your team's top agents
audit-website 12 uses ★
clone-website 8 uses ★
setup-claude 5 uses
release-notes 3 uses
$ _
See which agents are actually being used
Every agent run is recorded. Who ran it, how long, what it cost. Cut the ones nobody uses. Keep spending on the agents that save real time.
runs
0
active agents
0
spend
$0
top by use
audit-website 0 runs · $0
release-notes 0 runs · $0
clone-website 0 runs · $0
candidate to cut
legacy-importer 0 runs · 90d idle
Keep agent knowledge after people leave
When your AI‑fluent engineer moves on, the prompts they built up stay. The next hire starts where the last one left off.
original arush.m
left 9 mo ago
runs since 3,402
original priya.k
left 18 mo ago
runs since 11,840
Standards by example, not by memo
The agents the team installs become the de‑facto standard. Nobody has to read a wiki to know "how we do AI here."
- pr-review 14 / 16 ★
- release-notes 12 / 16 ★
- audit-website 10 / 16 ★
- api-contract 8 / 16
- incident-postmortem 6 / 16
★ = team standard
Fork agents for each team
The agent that works for platform usually needs tweaks for product. Fork it. Modify. Push back. Same instincts as any code.
Built like engineering infrastructure.
Underneath the friendly install command, every Operon agent is built to the standards you'd want from any piece of dev infrastructure: versioned, observable, governable, and local‑first.
- v0.3.2 current
fix: tighten lighthouse threshold
2d ago · arush
- v0.3.1
feat: add a11y audit step
5d ago · arne
- v0.3.0
feat: nightly cron
2w ago · arush
- v0.2.4 rollback
fix: handle 404 selector
3w ago · arne
Share your team's first agent today.
Free. No card. Two minutes from CLI install to your first mined prompt.
Works with Claude Code · Codex · Cursor