Editor modeling cockpit

Status: Phases A–F shipped (2026-05-22)
See also: simulation-cockpit.md

Historical phase roadmaps live in-repo under docs/plans/shipped/ (not published on the docs-site).

North star

One guided question: Is it safe to ship this process change?

Open Check change in the editor toolbar to show the Modeling Cockpit.

Three visible inputs

Guided flow

Header (ready/blocked + controls) → Impact summary → Detail tabs → Ship verdict

Default landing: Impact. Ship is the terminal step.

Modeling Cockpit tabs

The cockpit header shows ready/blocked status, an operational summary (hover for proposal/baseline detail), baseline picker, and re-run controls on one row. Blocker chips appear only when checks fail. Empty states include CTAs; Capture test on forecast rows adds YAML regression tests with toast feedback.

Toggle Live lens in the cockpit header to overlay open cases, findings, and SLA breaches on the canvas.

What is live?

Process models in the editor follow three layers — there is no separate “activate” button beyond linking a process and saving the file.

Process twin

A process twin is a platform process record whose currentConfigPath points at this workflow file (for example .kiket/workflows/vendor-onboarding.yaml). The API reads that path from the linked repo when:

• Resolving case transitions and guards
• Running the compliance scanner against operational cases
• Building simulation Cases / History tabs in the Modeling Cockpit

If no twin is linked, you can still edit and save YAML in git — but open cases will not use the model until onboarding or workspace setup creates a monitored process for this path.

The editor shows a config lifecycle strip above the canvas: pending save (amber), live config (green), or saved but unlinked (neutral).

Process context bar

When the Modeling Cockpit is open on a workflow file, the editor shows a context bar with workspace, repo/branch, process name, open case/finding counts, and links to cases and findings.

Workflow editor chrome (shipped 2026-05)

The workflow canvas uses a mode bar and right inspector instead of a bottom Process drawer:

Coverage warnings on the canvas banner link to Evidence & checks. Cockpit Edit actions on Tests and Ingestion open the same inspector tab.

Visual process config (Phase B)

In the workflow canvas toolbar, click Process to open structured panels for:

• Evidence requirements
• Compliance checks
• Remediation mappings
• Regression scenarios

Legacy v1 workflow files show an Upgrade to v2.0 banner that runs migrateWorkflowConfigYaml and replaces the editor buffer.

Live operational lens (Phase C)

simulation-data returns extended processContext:

• Repo name, branch, workflow path
• liveLens — per-state open cases, open findings, SLA breach counts

The cockpit merges forecast impact with live lens metrics on canvas nodes.

Branch compare & CI (Phase D)

• Impact tab / baseline picker: compare current YAML against saved file or main (via GET /repos/:id/file?path=&ref=main)
• Local CLI: pnpm kiket:simulate path/to/workflow.yaml [--baseline base.yaml]
• CI: flagship definition packs run through kiket:simulate after package build

Polish & scale (Phase E)

• Auto-layout: Dagre layout (Layout button or L); fit view (F); auto-layout on open when nodes stack
• Template gallery: New workflow → blank scaffold or flagship definition pack
• Keyboard help: Press ? on /editor for global + editor shortcut sections
• Accessibility: Toolbar role="toolbar" and feature badge ARIA labels

Template file API: GET /templates/catalog/:key/file?path=.kiket/workflows/...

Deep links

Share or bookmark a specific cockpit view with URL query params on the editor:

Example: /editor?file=.kiket/workflows/vendor-onboarding.yaml&simulate=1&tab=cases

Tab changes update the URL in place so refresh and share round-trip.

Shipped UX (Phase F)

• Compact cockpit header — ready/blocked chip, operational summary, baseline picker, re-run, auto-check, and tabs on one row; blocker chips only when present
• Impact default — aggregated status rows drill into detail tabs
• Capture test — from Cases rows; toast + highlight; duplicate detection
• Add test / Edit — opens Evidence & checks → regression tests
• Ingestion — source health, coverage gaps, Edit evidence requirements → Evidence & checks
• Re-run check, auto-check toggle, baseline picker (saved vs main)
• Ship tab — Fix links to detail tabs; Save when ready
• Live lens legend on canvas overlay
• URL state — ?simulate=1&tab=<tab> deep links; inline package process cockpit retired

Walkthrough (30 seconds)

Ship verdict

The Ship tab aggregates YAML schema, graph validity, reachability warnings, replay failures, scenario failures, and new guard blockers vs the saved definition. Each failed check includes a Fix link to the relevant detail tab.

Acceptance criteria

Within 30 seconds of opening the cockpit, a user can answer:

1. What is being checked?
2. Against what baseline and how many live cases?
3. What failed?
4. What is the next action?

API extensions

GET /workflows/:id/simulation-data — extended processContext:

{
  "processContext": {
    "linked": true,
    "processId": "...",
    "workspaceName": "Production Operations",
    "repoId": "...",
    "repoName": "acme/compliance",
    "repoBranch": "main",
    "workflowPath": ".kiket/workflows/production-change.yaml",
    "openCaseCount": 4,
    "openFindingCount": 2,
    "liveLens": {
      "openCasesByState": { "review": 2 },
      "openFindingsByState": { "review": 1 },
      "slaBreachesByState": {}
    }
  }
}

GET /repos/:id/file?path=...&ref=main — read file at branch for compare.

Test plan