Pro Features
Prume Pro extends the free CLI with a visual Web UI, AI classification, and deep integrations. Install with pip install prume-pro.
Web UI
The Web UI provides a visual interface for managing your change units and groups. See all your changes at once, drag hunks between groups, review diffs, and commit — all from your browser.
$ prume ui
Prume Web UI running at http://localhost:5555
Press Ctrl+C to stop.Full project overview
All change units and groups in one view
Drag-and-drop
Move change units between groups visually
Inline diffs
Review every hunk before committing
Group management
Create, rename, merge, split, set messages and dependencies
One-click commit
Stage a group and commit directly from the UI
Stack visualization
See the dependency graph for your commit stack
Options
--port, -P— Port number (default: 5555)--host, -H— Host address (default: 127.0.0.1)--dev, -d— Development mode with hot reload--path, -p— Path to repo
AI Classification
Layer 4 of Prume's classification engine uses AI to handle change units that deterministic rules can't classify. AI is always optional — it supplements the rule engine rather than replacing it.
$ prume suggest --provider openai
Analyzing 5 unassigned change units...
Suggestions:
src/utils/cache.py:12-28 → auth-refactor (confidence: 0.89)
src/api/middleware.py:5-15 → api-endpoint (confidence: 0.92)
...
$ prume suggest --apply
Applied 5 suggestions.The 4-layer pipeline
AI classification is Layer 4 of a 4-layer pipeline. Deterministic rules handle the bulk first:
src/auth/** → auth-refactor, *.css → styling, tests/** → tests
Diff contains TODO/FIXME → cleanup, adds import from newlib → migration
Custom rules you define once, applied every time
For ambiguous changes the rules can’t catch
How it works
Prompt generation
Prume builds a structured prompt with unassigned change units and existing group definitions
AI inference
The prompt is sent to your configured provider (OpenAI, Anthropic, or local models via Ollama)
Confidence scoring
AI returns grouping suggestions with confidence scores for each assignment
Review & apply
You review and apply — or use --apply for automatic assignment
Key insight: By the time AI runs, the majority of changes are already classified by deterministic rules. AI only handles the remaining ambiguous hunks — making it fast, cheap, and accurate.
Jira Integration
Link groups to Jira issues. Commit messages automatically include issue keys, and issue statuses can update when commits are pushed.
$ prume jira test --verbose
✓ Connected to https://your-org.atlassian.net
✓ Authenticated as developer@example.com
$ prume jira link auth-refactor PROJ-142
Linked group "auth-refactor" to PROJ-142.
$ prume commit auth-refactor
Committed: "PROJ-142: Refactor auth middleware for multi-tenancy"prume jira test
Test connection and credentials
prume jira search <JQL>
Search issues by JQL query
prume jira issue <KEY>
View issue details
prume jira link <GROUP> <KEY>
Link a group to an issue
Configuration
Set Jira credentials in your project config or via environment variables:
[extensions.jira]
url = "https://your-org.atlassian.net"
email = "developer@example.com"
api_token = "..." # or use JIRA_API_TOKEN env varStacked PRs
Create one PR per group with proper dependency ordering. Supports multiple stacking tools.
Graphite
Full CLI integration
SPR
Single PR stacking
Aviator
Merge queue
ghstack
GitHub stacked diffs
Git Town
Workflow automation
$ prume stack-pr status --tool graphite
3 groups ready for submission:
1. auth-refactor (PROJ-142)
2. api-endpoint (PROJ-143) → depends on auth-refactor
3. css-fixes (PROJ-144)
$ prume stack-pr submit --tool graphite --draft
Created 3 draft PRs:
#142: Refactor auth middleware
#143: Add /users endpoint (stacked on #142)
#144: Fix settings page CSSJetBrains Changelists
Import and export Prume groups as JetBrains changelists for developers who prefer IntelliJ, WebStorm, PyCharm, and other JetBrains IDEs.
prume jetbrains status
Show current changelists
prume jetbrains export
Export groups as changelists
prume jetbrains import
Import changelists as groups
PR Descriptions
Generate PR descriptions from a group's changes and metadata. Optionally includes Jira issue context.
$ prume pr describe auth-refactor --format markdown --jira
## PROJ-142: Refactor auth middleware for multi-tenancy
### Changes
- Updated AuthMiddleware to support tenant isolation
- Added TenantContext to request pipeline
- Updated 14 files across 3 modules
### Jira
- Issue: PROJ-142 (In Progress → In Review)
- Epic: PROJ-100 Multi-tenancy
$ prume pr describe auth-refactor --copy
Copied to clipboard.Auto-Assign
Automatically assign unassigned change units based on similarity to existing group contents.
$ prume auto-assign --threshold 0.7
Assigned 3/5 unassigned change units:
src/utils/cache.py:12-28 → auth-refactor (0.85)
src/api/middleware.py:5-15 → api-endpoint (0.91)
tests/test_auth.py:1-20 → auth-refactor (0.78)
Skipped 2 (below threshold).Gamification
Optional streak tracking and achievements to encourage clean commit habits. Visible in the Web UI and CLI.
Commit streaks
Track consecutive days of clean commits
Achievements
Unlock milestones (first 100-file session, first stacked PR, etc.)
Insights
Classification accuracy, average group size, and rule coverage over time
Gamification is opt-in and can be toggled in settings.
Licensing
Prume Pro uses offline-capable license keys. Trial keys work for 14 days with no signup.
$ prume license trial --email you@example.com
Trial license activated (14 days remaining).
$ prume license status
License: Pro Trial
Expires: 2026-03-03
Features: web-ui, ai, jira, stack-pr, jetbrains, gamification
$ prume license set <KEY>
License activated.After your trial, purchase a license at /pricing. Keys are per-seat and support offline validation.
Feature gating
Use prume license features to see which Pro features are available on your current license. All features degrade gracefully — the free CLI continues to work without a license.