Your coding agent doesn't know why this feature matters. Now it can.

Give Claude Code, Cursor, or Codex the customer problem behind the task—not just the code around it. Your agent can see where users struggle, why the work was prioritized, and what should improve before it starts building.

Run the Custory CLI Read the MCP docs

Recommended · connect Cursor in two commands

npx @custory/cli@latest login
npx @custory/cli@latest mcp install --client cursor --workspace "Acme"

Context

Customer problem loaded

Scope

Right workspace connected

Ready

Agent connection verified

Use claude or codex instead of cursor to connect another supported client.

Give your agent the context your repo cannot

Your codebase explains how the product works. Custory explains why the change matters to customers, why it deserves attention now, and what better should look like.

Build for a real customer moment

Start with the affected journey step and persona, so the agent understands where the change sits in the experience—not only where the component sits in the repo.

Solve the supported problem

Bring the related friction, insights, and customer evidence into the session so the implementation responds to a known problem instead of a plausible guess.

Aim at the intended outcome

Keep the prioritized opportunity and metric visible, so the agent builds toward the result the team chose rather than optimizing the first solution it finds.

The CLI sets it up. MCP keeps the why with the work.

Create the first customer journey, bring its context into your coding environment, and keep the journey current as improvements ship.

Create the first journey from the terminal

Start from your product website or a private brief. The CLI creates a structured journey with personas and starting context, verifies it, and opens it in Custory.

View @custory/cli on npm

Bring customer context into every session

Connect the workspace once, then let your coding agent pull the relevant journey context when it plans and implements product changes.

Read the MCP server docs

Keep delivery connected to the journey

When the change ships, the agent can update the solution, record the PR, and preserve what the team should measure next.

From prioritized friction to a measured fix

The agent starts with the customer problem, uses that context while working in the repo, and closes the loop when the change is ready.

Example · coding agent session

You: Implement the teammate-invite improvement we prioritized in Custory.

Agent: I found the affected moment in the onboarding journey.
Customers hesitate because member roles are unclear before they send an invite.
The team prioritized clearer role guidance and a better recovery state.
Success means more new workspaces complete their first teammate invite.

Agent: I updated the invite flow, added the missing recovery state,
and covered both paths with tests. PR #418 is ready.

Custory: Solution marked as shipped.
PR #418 and the invite-completion metric are attached to the journey.

One connected loop

The customer why survives the handoff to code

Start with the problem: the agent sees the affected moment and supporting evidence.
Keep the decision: the intended improvement stays visible while the agent works.
End with learning: shipped work and the metric to watch return to the journey.

What your agent writes back after shipping

Custory keeps the decision, implementation, and next learning connected instead of letting the customer context disappear when the pull request merges.

The solution that shipped

Update the solution status so the journey reflects what customers can experience now, not what the team was planning weeks ago.

The implementation trail

Record the pull request and release context so anyone can move from the customer problem to the code that changed it.

The result to watch

Keep the baseline and target beside the journey moment, so the next review asks whether the change worked—not only whether it shipped.

Use Custory inside the agent workflow you already have

Keep coding where you code. Custory adds customer context and durable follow-through without becoming another place where implementation has to happen.

Works with Claude Code, Cursor, and Codex

Connect the coding agent your team already uses instead of moving development into a separate AI workspace.

Scoped to the right workspace

The connection stays tied to the selected Custory workspace and can be revoked when a device or client should no longer have access.

No hidden usage tradeoff

Direct MCP reads and writes do not consume AI messages or automation runs. Your plan’s normal journey, persona, and seat limits still apply.