Built for terminal-first agents

Sprint Tickets,
Claude-Native.

One command — `claude mcp add` — and Claude Code reads the sprint, picks the next ticket, ships the PR, and comments back. Native MCP, no skill files.

5-min setup
REST API + /llms.txt
No card to start

Why Claude Code + SprintFlint

Built around how your AI actually works.

Other tools

  • API requires OAuth dances and SDK installs
  • No public agent-readable spec
  • UIs assume a human is clicking
  • Per-seat creep — you pay for the agent's seat too

SprintFlint

  • Public /llms.txt — agents discover the API instantly
  • Bearer-token auth, one header, no OAuth dance
  • REST endpoints designed for agent-first workflows
  • £5/user/mo, free tier covers your first 300 tickets
Step 1 — Token

Generate the token.

# In SprintFlint: Profile → API Token → Generate
# Same token works for MCP and the REST API.
Step 2 — claude mcp add

Add SprintFlint as an MCP server.

One CLI command. Replace YOUR_TOKEN with the token from step 1. No file editing — Claude Code stores the server config for you.

claude mcp add sprintflint \
  --transport http \
  https://mcp.sprintflint.com \
  --header "Authorization: Bearer YOUR_TOKEN"
Step 3 — Run

Hand off the sprint.

"Claude, ship the next SprintFlint ticket."
Reads the backlog. Picks one. Branches. Codes. Tests. Comments. Done — all via MCP, no scripted REST calls.

  • Agent reads sprint backlog
  • Auto-updates status when work begins
  • Posts progress comments as it goes
  • Pairs with Autoplay for full one-click PRs

Your Terminal Already Ships Code.

Now your sprint board can keep up.
300 tickets free. Native MCP, no skill files to maintain.