CI your coding agents can run, read and fix

Self-healing GitHub Actions runners, plus a CLI and MCP server your coding agents can drive.

Set up Latchkey CI for this repo. 1) Install the CLI: npm install -g @latchkeydev/cli. 2) Add the Latchkey MCP server: claude mcp add --transport http latchkey https://latchkey.dev/mcp --header "Authorization: Bearer $LATCHKEY_TOKEN" (create the key in the Latchkey dashboard under Settings, API keys, with Allow running CLI jobs on, and export it as LATCHKEY_TOKEN). 3) Run the test suite on a fresh runner with latchkey run, fix anything that fails, and run it again.

30 days free · No credit card · From $0.0025/min

RUN

Run CI before you push

Agents run any command on a fresh Linux runner and get the real exit code back. One runs-on: line moves your workflows over.

Runner Switch
Beforeruns-on: ubuntu-latest
Afterruns-on: latchkey-medium
~2s on latchkey-small|Fresh runner|1 line change

“Our CI spend dropped to a fraction of what we’d budgeted, setup took minutes, not days, and nobody on the team has to be ‘the CI person.’ We ship faster and spend our time on Busbar instead of the plumbing around it.”

BusbarMatthew · Founder, Busbar
HEAL

Self-healing runners

Transient failures diagnosed and repaired mid-run: over 10,000 a week.

AI Insight
Self-heal, this week47 failures repaired · 18.5 dev-hrs saved
Auto

“One line of YAML and our CI bill dropped by two thirds. Honestly the bigger win is that we stopped losing hours every week to reruns that were never our code’s fault.”

CapyVincent · CEO, Capy
HAND OFF

An MCP server for coding agents

What self-heal can't fix goes to your coding agent, with the logs and failing step attached.

Agent hand-off
  1. CI failed · auth.spec.ts2:14 AM
  2. Self-heal: real failure, handed off2:14 AM
  3. Claude Code reads the failure bundle2:15 AM
  4. PR #412 opened, tests passing2:21 AM
APPROVE

One-click fixes, savings priced in dollars

Cost recommendations priced in dollars, security and reliability findings beside them, waiting for your yes.

AI Insight
Insight, ready to apply$3,280/mo · 14 open findings
5 cost4 security3 reliability2 adoption

Watch It Heal

A failing step, diagnosed and repaired mid-run. No human and no agent needed.

01

Step fails

A dependency install times out and the job goes red.

ci.yml · npm ci
npm ci
npm ERR! code ETIMEDOUT
npm ERR! network request to https://registry.npmjs.org/lodash failed
Step failed (exit 1)
02

Latchkey heals

The runner matches the failure pattern and retries with backoff.

ci.yml · npm ci
# [latchkey-bash-wrapper] step failed (exit 1), diagnosing...
# [latchkey-bash-wrapper] verdict: network/registry-timeout (pattern match)
# [latchkey-bash-wrapper] action: retry with backoff (attempt 1 of 3)
Diagnosing and healingPattern matchRetry with backoff
03

Job goes green

The step passes and the job carries on. Your agent never sees the failure.

ci.yml · npm ci
npm ci
added 1291 packages in 42s
# [latchkey-bash-wrapper] heal successful, continuing job
HealedJob continued

Give your agents a CI they can drive

Change one runs-on: line, connect your agent, and let the runners heal what breaks.

30 days free · No credit card · 10,000+ CI failures healed every week

We use cookies to improve your experience. Privacy Policy