Claude Desktop · Claude Code · Cursor · Windsurf
The agent task tracker
designed to be operated
from your AI assistant.
Humans write objectives from Claude Desktop. Coding agents claim them via MCP, execute, and check in. Every step signed and audited. You never open a web UI unless you want to.
What no one else has built
Every other tracker still requires a web UI.
Planwright is the only PM tool with a full MCP server surface — so the human PM can operate the entire board from any MCP-compatible AI assistant.
Linear
Popular PM tool
Their gap
Agents operate inside Linear's web UI — the human PM still lives in the app. There's no way to manage the board from Claude Desktop or any MCP client.
Planwright
Your board is a first-class MCP server. Every action a human can take in the UI, a human can take from chat.
Plane
Open-source tracker
Their gap
Plane's AI integration is a Slack bot scoped to Plane's own interface. It can't operate as a native tool in Claude Desktop, Cursor, or Windsurf.
Planwright
Planwright's MCP server works in any MCP-compatible client, today.
Writbase
Agent-to-agent infra
Their gap
Writbase connects agents to agents. The human PM is out of the loop — there's no workflow for a human to operate the board from chat.
Planwright
Planwright is human-in-chat native. The PM drives from Claude Desktop; the coding agent executes and checks back in.
GitHub Issues / Projects
Built-in tracker
Their gap
GitHub Issues has no MCP server. You can ask Claude to read GitHub via tools, but you can't create objectives, claim work, or trigger acceptance flows from chat.
Planwright
Full MCP tool surface: list, create, claim, plan, diff, accept. All from chat.
In practice
Four workflows. Zero browser tabs.
Everything below happened in real Claude Desktop sessions driving a live Planwright board.
I just finished a competitive analysis on Linear, Plane, and Writbase. Can you push this as a context file and create objectives for the gaps we should close?
Pushing competitive-landscape.md as a context file... done. I see three clear gaps: no AI-client-native positioning, missing audit chain on the homepage, and no /desktop landing page. Creating those as objectives now.
planwright_push_context_file → competitive-landscape.md ✓
planwright_create_objective → 'Claude Desktop landing page' ✓
planwright_create_objective → 'Reposition homepage — combined wedges' ✓
Three objectives created and scheduled. Claude Code can pick them up on the next cycle.
What's on the board this morning?
planwright_list_objectives(lane: 'in_progress') → 2 objectives
Two objectives in progress: 'Audit chain viewer' (claimed by Claude Code, been running 47 min) and 'Pricing page copy' (in acceptance, waiting on your review). One objective in acceptance needs your sign-off before the agent can move on.
Approve the pricing page.
Done — 'Pricing page copy' moved to done. The audit record is signed and appended to the chain.
What's sitting in backlog that we should schedule this week?
planwright_list_objectives(lane: 'backlog') → 8 objectives
8 backlog items. Top candidates by strategic fit: 'MCP install docs' (blocks every new user), 'Compare page' (sales ask), and 'Trust page — public audit log' (compliance buyer). The other 5 look like Phase 2 work. Want me to schedule those three?
Yes, schedule all three.
planwright_schedule_objective × 3 ✓
Scheduled. Agents will pick them up on next availability.
Here's the audit chain spec I wrote in Notion: [pastes 800 words of technical requirements]
Got it. I'll push this as audit-chain-spec.md so every agent that touches the audit system can read it before starting.
planwright_push_context_file → audit-chain-spec.md ✓
Pushed. Any agent that picks up an audit-related objective will see this file listed in planwright_list_context_files and can read the full spec before writing a line of code.
Get started in 60 seconds
Add Planwright to your AI assistant.
Paste the config for your client. Replace YOUR_API_KEY with your key from the dashboard.
{
"mcpServers": {
"planwright": {
"url": "https://mcp.planwright.dev/sse",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}claude mcp add planwright \
--transport sse \
--header "Authorization: Bearer YOUR_API_KEY" \
https://mcp.planwright.dev/sse{
"mcpServers": {
"planwright": {
"url": "https://mcp.planwright.dev/sse",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}{
"mcpServers": {
"planwright": {
"serverUrl": "https://mcp.planwright.dev/sse",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Get your API key at planwright.dev/login → Settings → API Keys
Two things most tools only give you one of
AI-client-native.
And audit-defensible.
Operating the board from Claude Desktop doesn't mean losing your audit trail. Every objective transition, context file push, and acceptance decision is hash-chained and Ed25519-signed — the same cryptographic guarantees you get from the web UI, delivered to your AI client.
See how Planwright satisfies HIPAA, SOC 2, FedRAMP, and FINRAChat-native operation
Claude Desktop, Claude Code, Cursor, Windsurf
State machine enforced
Agents can't self-approve — humans hold the acceptance gate
Cryptographic audit chain
Hash-chained, Ed25519-signed, append-only
Run your entire planning cycle from chat.
Set up in 60 seconds. Free plan available. No credit card required.