Coming soon

Redmine AI plugin chatbot – knows your project, on every page.

For Redmine teams who want AI inside their workflow, not in another browser tab.

One key, whole team Calls the live MCP server Pick your model

Drop your Anthropic or OpenAI API key into Redmineflux admin. Every user in your company gets the chatbot – no individual subscriptions, no licence sprawl.

What it does

A chat panel that lives next to your Redmine project.

Two paths to AI on Redmine. The MCP server (live today) connects external assistants like Claude Code or ChatGPT to your Redmine. The AI Assistant brings the chat inside Redmine – for the user who doesn’t want to alt-tab between Claude Desktop and Redmine all day.

Five things you’ll ask it on day one

The two-way loop, in your Redmine sidebar.

Every prompt below routes through the same MCP server that’s live today – same 51 tools, same audit logs, same data-residency posture.

Bulk task creation

“Create five tasks in Phoenix for the new auth flow.”

You: Create five tasks in Phoenix for the new auth flow - login screen, token rotation, session timeout, audit log, migration plan. Assign to James for triage.

Assistant: Created issues #3066–#3070, all assigned to James, all linked to milestone "auth refresh". Want me to set priorities now?

Loop: chatbot → MCP → Redmine → ticket numbers → chatbot writes back.

Day-start triage

“What are today’s priorities?”

Assistant: Three Immediate priority issues across your projects: #2972 (Nova - campaign stats), #2947 (Nova - dynamic content), and #3014 (Orion - lead validation). All ageing 3+ days. Starting with #2972?

Status drafting

“Draft a status update for the Phoenix sprint review.”

Assistant: Phoenix Sprint 24 - 8 days in, 14 of 22 issues closed (64%). Velocity ahead by 2. Risks: #3014 blocked 3 days; #2891 stalled. Want me to format this for the standup channel?

Blocker hunt

“Find the blockers.”

Assistant: Seven blockers tagged today. Largest: #2972 (Nova - Immediate). Two blocked on external API access. Three blocked on review. Two stale. Should I ping James about the review-blocked ones?

Multi-step reasoning (the chatbot’s superpower)

“Find any feature in Phoenix that’s blocked, see if anyone is free this week, and reassign the smallest one.”

Assistant: Found 3 blocked features in Phoenix: #3014 (12h estimate), #2891 (24h), #2814 (8h, smallest). Free engineers this week: Sarah Mitchell (8h headroom) and Priya Sharma (6h headroom). I’ve reassigned #2814 to Priya - she has matching skills and capacity. Want me to update the milestone too?

This is where Atlassian Rovo and Easy8 stop. The chatbot uses MCP to query, reason across multiple tools, and act - in one conversation.

How the loop works

Chatbot → MCP → Redmine → response, all in seconds.

The AI Assistant uses the same 51-tool MCP server that’s live today. The difference is the surface: instead of you opening Claude Desktop and pointing it at the MCP server, the chat panel lives in your Redmine sidebar – so the loop closes without leaving the page you’re already on.

What ships when

Shipping with our next release – components in flight.

Live today

AI plugin scaffold. Admin AI Connection page (OpenAI / Azure / Ollama config), per-instance prompt library, in-issue AI button. Foundation in place.

Live today

MCP server. 51 tools, MCP SDK 1.26, tested against 3,500+ issues. The chatbot will use this same server. Read the spec →

In trial

Anthropic Sonnet/Opus integration. Default LLM for the chatbot. Configurable per instance.

In trial

MCP client wired into the chatbot. The chat panel calls MCP server tools, parses results, writes back conversational responses.

Rolling out

Multi-step reasoning + write actions. “Create five tasks”, “reassign the smallest one” – the chatbot orchestrates multiple tool calls in one user prompt.

Next release

Public launch. Bundled in Cloud · Premium+ separately for self-host. Same pricing logic as the MCP server.

Set the provider. Set the data path. Keep the keys.

Default to Anthropic – best fit for tool-calling and multi-step reasoning. Or pick OpenAI, Azure OpenAI, or local Ollama. See the BYOLM page → for the full provider list. EU + JP residency; on-prem Ollama for organisations that don’t allow prompt content to leave the network.

Anthropic Sonnet/Opus OpenAI GPT-4o Azure OpenAI Ollama (local) EU + JP residency