AI-native product management for Claude Code
Manage products, track tasks, capture knowledge, debate decisions, generate reports, and configure Atlassian agents — all from your terminal.
The Forge is a suite of 8 Claude Code plugins backed by a shared Python data layer and a Tauri desktop app for visual dashboards. It brings structured product management into your AI coding workflow — no context switching, no separate tools.
Plugins handle conversation and workflow. A deterministic Python CLI (forge-lib) handles all file operations, validation, and indexing. Forge Shell gives you a desktop GUI to browse everything the plugins create.
| Plugin | What it does | |
|---|---|---|
| Product Forge | Initiatives, epics, stories — full product hierarchy with auto-linked relationships | |
| Tasks Forge | Sequential task tracking with status workflow and priority management | |
| Cognitive Forge | Multi-agent debates and explorations with 5 specialized reasoning agents | |
| Forge Memory | Organizational knowledge with taxonomy — products, modules, teams, integrations | |
| Report Forge | 8 report types generated via multi-agent orchestration | |
| Rovo Forge | Interactive builders for Atlassian Rovo Jira & Confluence agents | |
| Slack Forge | Channel intelligence harvester — surfaces tasks, knowledge, and JIRA activity | |
| Outlook Forge | Outlook intelligence harvester — surfaces tasks, knowledge, and meeting context |
graph TB
subgraph CC ["Claude Code"]
A["LLM Reasoning Layer"]
end
A --> B["Plugin Commands · 80–100 lines each"]
subgraph PL ["Plugins"]
direction LR
subgraph col1 [" "]
C1["Product Forge"]
C2["Tasks Forge"]
C3["Cognitive Forge"]
C4["Forge Memory"]
end
subgraph col2 [" "]
C5["Report Forge"]
C6["Rovo Forge"]
C7["Slack Forge"]
C8["Outlook Forge"]
end
end
B --> PL
subgraph FL ["forge-lib · Python CLI"]
D["forge.py"] --> E1["Validation"] & E2["Templates"] & E3["Indexing"] & E4["Relationships"]
end
PL --> FL
FL --> F["cards/ · tasks/ · sessions/ · memory/ · reports/"]
G["Forge Shell · Tauri Desktop App"] -.->|reads| F
Key design principle: LLM handles conversation, Python handles data. Commands dropped from 250–300 lines (v1) to 80–100 lines (v2) by delegating all file operations, validation, and templating to forge-lib.
A Tauri desktop app that gives you visual dashboards for everything the plugins create.
# 1. Install the Python data layer
cd forge-lib && pip install -r requirements.txt
# 2. Verify
python forge.py --help
# 3. Add marketplace to Claude Code
ln -s /path/to/the-forge ~/.claude/marketplaces/forge
# 4. Launch Forge Shell (optional)
cd forge-shell && npm install && npm run tauri devSee individual plugin READMEs for detailed workflows and command references.
| Resource | Description |
|---|---|
| forge-lib CLI Reference | Full CLI docs, usage patterns, examples |
| Product Forge | Card hierarchy, relationships, workflows |
| Tasks Forge | Task workflow, status transitions |
| Cognitive Forge | Multi-agent reasoning sessions |
| Forge Memory | Knowledge taxonomy and recall |
| Report Forge | 8 report types, multi-agent generation |
| Rovo Forge | Atlassian Rovo agent builders |
| Slack Forge | Slack channel intelligence harvester |
| Outlook Forge | Outlook intelligence harvester |
| Forge Shell | Desktop app build and usage |
Jeremy Brice



