Skip to content
Model cards

Agent model cards

A model card says, in plain words, what one of Runink TIDE’s agents does, what it reads, what it produces, what a person still decides, which model it runs on, and its scope. There is one card for each agent you meet in the console, in a pull request or in an issue. Agents that work as a pair share a card.

Every card has the same eight parts: what it does, what it reads, what it produces, human oversight, model, where it runs and data handling, guardrails, and scope.

Three things true of every card

  • Agents propose; people decide. No TIDE agent merges a pull request, deploys a change or sends anything outside your organisation by itself.
  • Your own inference. Every agent calls the model plane of your own TIDE deployment, on your Server or in your cloud. No third-party AI service is called. See Models & inference.
  • Open base models. Agents run on open base models, listed below.

The base models

TierModelMakerLicenceUsed by
coderQwen3.8-27BQwenApache-2.0Code Reviewer, Fix Drafter, CI Doctor (failed-build diagnosis), the FORGE coding step
generalQwen3.8-27BQwenApache-2.0every other agent that calls a model

One base model, Qwen3.8-27B, serves both tiers, from the same server. Agents that screen their input also use the guard tier, Qwen3Guard-Gen-0.6B (Qwen, Apache-2.0). The Independent Assessor also uses a small ranking model, Qwen3-Reranker-0.6B (Qwen, Apache-2.0), to narrow evidence. The Data Governance Checker uses no model at all.

The model that actually answers each agent, as resolved from what it sends, is in the Model section of the agent’s page on DataEx › Agents.

The cards

CardAgentModelWhat a person decides
Code Reviewerreviewercoderwhich findings matter and what merges
Fix Drafterfixer, resolvercoderwhether to merge the draft correction
CI Doctorself-heal, opsdoctorcoder (self-heal), general (opsdoctor)whether to apply the proposed fix
Issue Triagertriagegeneralthe final labels and who works on the issue
Backlog Hunterhuntergeneralwhat to do about each reported item
Dependency Riskriskgeneralwhich upgrades to schedule
Compliance Evidencecompliancegeneralhow to close a gap; any claim of compliance
Release Curatorcuratorgeneralwhether to merge the curator’s pull request
Independent Assessorjudgegeneral, with a ranking modelwhat to do with each verdict
Rules Reconcilerrecongeneral, for one questionwhether to fix the rule or the engine
Data Governance Checkerdatagovno modelowners, credentials and masking
Task Deployerdeployergeneralwhether to merge the release-notes pull request
Ask TIDEconsole Ask and chatgeneralwhat to do with an answer; every provisioning action
FORGEFORGE planner, forgergeneral (planner), coder (code)every planned step, and every merge

The roster names in the second column are the ones the DataEx › Agents page uses.

Not given a card

  • eval is a test harness for the agents themselves, not an agent you meet. Its trace-scoring step uses the general model.
  • users applies access grants from an add-users issue. It uses no model.
  • session is the tide session coding agent, which you drive yourself. See Coding Sessions. The Fix Drafter’s resolver and FORGE’s forger use it.