Your agent is deployed and running. This keeps it that way.

The agent watches your systems. This retainer means I watch the agent.

When your database falls over, the agent's job is to tell you why. Mine is to make sure it is still there to tell you, still reading every source correctly, and still worth listening to.

$1,000 a month. Cancel any month, no notice period, no minimum term. For builds I delivered.

Updates. When the core moves, your build gets it - a pull request against your repository or a tagged image, whichever your team prefers. You deploy it through your normal pipeline, or I deploy it if you have given me access.

Security patches. Dependency bumps, CVE fixes, base image rebuilds. Every one goes through the same CI as the original build: linting, tests, Grype scan, SBOM and provenance attestation on the image.

Connector repair. Sources change. An API versions up, a schema drifts, a log format shifts, a key rotates. I fix the connector before it becomes a hole in your coverage. This is the work that actually happens most.

Tuning. Thresholds go stale, hooks get noisy, playbooks stop matching reality. I retune hooks, digests, alerts and prompts against what your stack looks like now. Cutting false positives is most of this, and it is the difference between a tool your team reads and one they mute.

Agent health. Its own logs, crash loops, rate limits, token burn, stuck queues. I watch the agent so nobody on your team has to.

Support. One channel, direct to me, any day of the week. You hear back within five hours - I keep my work setup with me wherever I am. Up to 6 hours of change work included each month - a threshold rewritten, a digest reshaped, a playbook adjusted.

Not incident response. The agent is what reports your incidents, and it stays useful precisely because it sits outside the firefight. Your team handles the incident. I keep the agent accurate.

Not application debugging. Your code, your pipelines, your deploys belong to your team, and they are better placed for it than I am.

Not new sources, hooks or playbooks. Anything added on top of your build is a flat per-unit add-on, same as during the build, quoted before I start.

Not new features or rearchitecting. That is a new scope, and I would rather quote it properly than fold it into a retainer badly.

Not a pager. You hear back from me within five hours, any day of the week. What you do not get is someone sitting in your incident channel - the agent does the watching, and I keep the agent sharp.

Not your model bill. You pay your provider directly at cost, same as always.

Your fork is yours. You own the code and you are free to change it. I support the version I ship; if a change of yours breaks something I will get you back onto a supported build, and if you want it upstreamed, send it over and I will look.

The same read-only credentials the build already uses, and nothing more. If you would rather I hold no standing access at all, that works too: your team runs the deploys and sends me the agent's own logs when something needs looking at. Tell me which you prefer and I work that way.

Monthly, through Contra. Starts the month after handover if you want it. Skip it and the agent keeps running exactly as delivered - it just stops receiving updates from me. Lapse for six months or more and there is a one-off $500 reonboarding fee, because by then I have to load your stack back into my head.

Never required.