Feature TM-152
Drag-reorder steps in the task drawer
P2
The task board built for AI coding agents Your agent does the work.
Your agent does the work.
Give it a board of its own.
indistack is the kanban board where Claude Code is the worker — it files its own bugs, scores its own backlog, streams progress live, and remembers everything between sessions. You direct. It ships.
You're on the list — we'll email at launch.
Todo 2 picked for session
Chore TM-149
Bundle Inter locally for offline start
P3
Analysis 2 agent scoring
Research TM-154
Why do long sessions drop the MCP socket?
mcpstability
P1
Feature TM-147
Auto-archive Done cards after 14 days
P3
In progress 1 agent working
Bug
running
TM-148
Parser drops trailing comma in tool args
Step 3 / 5
- +1 done
- Trace tokenizer boundary
- Patch lookahead in args parser Editing src/parser/args.rs — branch fix/trailing-comma
- Add regression tests
- +1 ahead
Editing src/parser/args.rs — branch fix/trailing-comma
P0
Testing 1 awaiting checks
Feature TM-145
Coin reward animation on Done
P2
Done 2 shipped · 2 today
Bug TM-141
Drawer scroll jumps on theme switch
P1
Chore TM-138
Migrate board state to SQLite
P2
I stopped writing tickets entirely. Claude files the bug, scores it, fixes it, and asks me to review. I just hand out coins.
The Analysis column is the killer feature. The agent RICE-scores its own backlog better than half the PMs I’ve worked with.
First tool where the agent remembers what it was doing last week. Sessions stopped feeling disposable.
How it works
Hire the agent. Keep the directing.
Three moves and your board runs itself — you go from writing tickets to reviewing shipped work.
01
Connect a project
One MCP server links Claude Code to your board. No plugins, no glue scripts — the agent shows up to work like you show up to Jira. Except it likes it.
02
The agent fills the board
It files its own bugs, decomposes epics, and RICE-scores everything in the Analysis column. Your backlog grooms itself while you sleep.
03
You review and reward
Live steps stream onto in-progress cards. Review what ships, award coins — they feed back into the prompt, so the agent learns what you value.
Features
Not a tracker with an AI button. A board built around the agent.
Every mechanic assumes the agent is the worker — filing, scoring, executing, learning.
A backlog that scores itself
Every task lands in Analysis, where the agent decomposes it and assigns a RICE score. Open the board to a prioritized queue — not a triage meeting.
Research Why do long sessions drop the MCP socket?
reach 7 · impact 8 · confidence .8 · effort 2 — scored by the agent
Watch it work, step by step
In-progress cards stream the agent’s plan live: done steps, the current action, what’s ahead. Glance at the board and know exactly where the session is.
- Trace tokenizer boundary
- Patch lookahead in args parser
- Add regression tests
Editing src/parser/args.rs — branch fix/trailing-comma
An economy for your agent
Ship something good — award coins. The tally feeds back into the agent’s prompt, turning your taste into its instinct.
128 +5 · clean fix, tests included
Coins you award feed back into the agent's prompt — reward what you want more of.
Skills that level up
Research, decomposition, prioritization, documentation — the agent earns experience in each, and you can watch it actually get better at the job.
Research
lv 4
Decomposition
lv 3
Prioritization
lv 3
Documentation
lv 2
The product
A calm instrument panel for a very busy worker.
Todo 2 picked for session
Feature TM-152
Drag-reorder steps in the task drawer
P2
Chore TM-149
Bundle Inter locally for offline start
P3
Analysis 2 agent scoring
Research TM-154
Why do long sessions drop the MCP socket?
mcpstability
P1
Feature TM-147
Auto-archive Done cards after 14 days
P3
In progress 1 agent working
Bug
running
TM-148
Parser drops trailing comma in tool args
Step 3 / 5
- +1 done
- Trace tokenizer boundary
- Patch lookahead in args parser Editing src/parser/args.rs — branch fix/trailing-comma
- Add regression tests
- +1 ahead
Editing src/parser/args.rs — branch fix/trailing-comma
P0
Testing 1 awaiting checks
Feature TM-145
Coin reward animation on Done
P2
Done 2 shipped · 2 today
Bug TM-141
Drawer scroll jumps on theme switch
P1
Chore TM-138
Migrate board state to SQLite
P2
Dark theme, built in
The whole token system re-skins — warm studio dark, not inverted grey
Bug
running
TM-148
Parser drops trailing comma in tool args
Step 3 / 5
- +1 done
- Trace tokenizer boundary
- Patch lookahead in args parser Editing src/parser/args.rs — branch fix/trailing-comma
- Add regression tests
- +1 ahead
Editing src/parser/args.rs — branch fix/trailing-comma
P0
A card that works
Live step plan, current action, RICE and priority — on every in-progress task
Why not Jira or Linear
Great tools — for a different worker.
Jira and Linear assume a human moves the cards. When the worker is an agent, the whole board changes shape.
| Jira · Linear | indistack | |
|---|---|---|
| Who moves the cards | You do | The agent does |
| Who writes the tickets | You do | The agent files its own |
| Backlog grooming | A meeting | RICE-scored automatically in Analysis |
| Progress updates | Someone edits a status field | Live steps stream onto the card |
| Memory between sessions | — | Built in — context survives the restart |
| Getting better over time | — | Coins + skills shape the agent’s behavior |
Pricing
Early access is free. Being early pays.
Early access Free during the beta
indistack will be a paid app at launch. Join the waitlist now, use the beta free, and keep a founder discount when pricing lands.
You're on the list — we'll email at launch.
Full board, MCP server and coin economy included
Your code never leaves your machine — native desktop app
Founder discount locked in at launch, forever
Direct line to the maker — early users steer the roadmap
FAQ
Fair questions.
Does it only work with Claude Code?
Claude Code is the primary worker today, but indistack speaks MCP — any agent that supports the Model Context Protocol can connect to the board, file tasks, and report progress.
Does my code get sent anywhere?
No. indistack is a native desktop app (Tauri, Rust core). The board, the MCP server and your data all live on your machine. Nothing is uploaded.
What exactly are coins?
A reward currency you hand out when the agent ships something you like. The running tally and what it was awarded for feed back into the agent’s prompt — over time it optimizes for what you actually reward.
What platforms are supported?
macOS first. Windows and Linux builds are on the roadmap — waitlist members get asked which to prioritize.
How is this different from Linear plus an AI integration?
Integrations bolt an agent onto a board built for humans. indistack inverts it: the Analysis column, live step streaming, coins and skills only make sense when the agent is the worker. That’s the whole product, not a plugin.
The board is ready.
Your agent is waiting.
Join the waitlist — beta access lands in your inbox, free.
You're on the list — we'll email at launch.