Dry Run Mode

Dry run mode lets you test GitWire's behavior without executing any mutations. All AI analysis runs normally, but GitHub API calls that modify state (labels, comments, PRs, branches) are intercepted and logged.

How It Works

Every worker checks isDryRun(repoConfig) before each mutation:

graph TD
    A[Worker processes job] --> B{Dry run?}
    B -->|Yes| C["Log: DRY RUN: would apply labels"]
    B -->|No| D["Execute mutation normally"]

Worker	What's Intercepted
triageWorker	Label application, triage comments
ciHealWorker	Workflow re-run, patch PR creation
issueFixWorker	Entire PR pipeline (branch, commit, PR)
maintainerWorker	Stale warn/close comments, branch deletion
phase2Worker	Merge queue processing
phase4Worker	AI review execution

Enabling Dry Run

Option 1: .gitwire.yml

settings:
  dry_run: true

Option 2: Dashboard UI

Navigate to Config → select a repo → scroll to Settings section → toggle Dry Run Mode.

A yellow banner appears at the top of the config page when dry run is active:

⚠️ Dry Run Mode — No mutations will be executed for this repo.

Option 3: API

# Enable
curl -X PATCH https://gitwire.example.com/api/config/org/repo \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"settings":{"dry_run":true}}'

# Disable
curl -X DELETE https://gitwire.example.com/api/config/org/repo

What Gets Logged

In dry run mode, workers produce detailed logs of what they would do:

INFO  DRY RUN: would apply labels ["bug","priority:high"] to org/repo#42
INFO  DRY RUN: would post triage comment on org/repo#42
INFO  DRY RUN: would create patch PR for org/repo run 12345 (test_flaky)
INFO  DRY RUN: would warn stale issue org/repo#38 (30 days old)
INFO  DRY RUN: would close stale PR org/repo#15 (90 days old)
INFO  DRY RUN: would delete branch org/repo:feature/old-branch

The maintainer worker records skipped actions with "status": "skipped" in the stale management audit log for traceability.

Use Cases

Scenario	How
Trial run	Enable dry run, push some issues/PRs, review logs, disable
Per-repo testing	Enable for one repo while others run normally
Pre-deployment validation	Enable before upgrading, verify behavior, disable
Demo mode	Show what GitWire would do without side effects

Combining with Risk Scoring

Dry run and risk scoring are independent layers:

graph LR
    A[Job received] --> B{Pillar enabled?}
    B -->|No| X[Skip]
    B -->|Yes| C{Dry run?}
    C -->|Yes| D[Log + skip mutation]
    C -->|No| E{Confidence gate?}
    E -->|Below| F[Comment-only]
    E -->|Above| G{Risk score?}
    G -->|High| H[Block]
    G -->|OK| I[Execute]

Dry run is checked first, so it overrides all downstream logic including risk scoring.

→ Policy-as-Code | Risk Scoring | Config API