CI workflows
Runink TIDE’s agents run as CI jobs on your Git host. This page lists each one, what starts it, and what it does. Schedules are in UTC. A comment tag goes in an issue or pull-request comment.
Agents
| Agent | Starts on | Purpose |
|---|---|---|
| Code review | a pull request opened or updated; a comment; a periodic sweep of every open pull request | sovereign code review (reviewer + inline fixer) |
| Fixer | a @tide_fix comment, or a manual run for one pull request | draft a fix from review findings |
| Resolver | every three hours, or a manual run | resolve findings the fixer could not |
| Triage | an issue opened | categorize and label new issues |
| Self-heal | a failed run of a build, deploy or apply workflow | diagnose failing CI |
| Dependency risk | daily at 05:23, or a playbook step | proactive dependency-vulnerability scan |
| Compliance | daily at 01:13, or a playbook step | verify framework alignment |
| Curator | daily at 06:17, a tide @curate comment, or a playbook step | the curator |
| Task deployer | a @tide_deploy comment | execute a task issue on the sovereign model |
| Data governance | daily at 07:11, a @tide_datagov comment, or a playbook step | assess the connected data sources |
| Independent assessor | daily at 08:29, a @tide_judgement comment, or a playbook step | independently judge submitted findings |
| Rules reconciler | daily at 09:37, a @tide_recon comment, or a playbook step | reconcile the governance rule book against the engine |
| Users | an add-users issue | tide @add_users: grant or revoke access on any target |
| Forger | four times a day, or a manual run | build the next forge issue |
| Inference bench | a manual run | time to first token and decode tokens per second, the number every agent budget is written against |
| Tag help | a comment with an unrecognised agent tag | answer with the tags that work |
On-demand commands
| Comment | What it does |
|---|---|
tide @build | a quick build and vet check |
@tide_ship | build image(s) and roll what runs them |
Rolling a workload onto new images
After a delivery roll, workloads are pinned to an image digest. A bare kubectl rollout restart restarts onto the same digest. To move a workload onto new images, run the delivery workflow for that image. See Console Pages.