Linear Integration: Assign issues to Claude Code to trigger cloud agent sessions

Agreed this would be super helpful. Have been doing this with Cursor background agents and codex integration to linear but would be awesome if it worked with claude code web too so it works like those in the background and then opens a PR

I tried this with linea GitHub issues integration but blocked triggering claude 😢

I tried this with linea GitHub issues integration but blocked triggering claude 😢

+1

+1

+1

+1

+1

+1

+1

+1

please sir, please

+1

+1

+1

+1

+1

+1

+1

+1

+1

+1

+1

+1

A hook-based approach can connect Claude Code to Linear via their API:

PROMPT=$(cat | jq -r '.userPrompt // empty' 2>/dev/null)
ISSUE=$(echo "$PROMPT" | grep -oE '[A-Z]+-[0-9]+' | head -1)
[ -z "$ISSUE" ] && exit 0
LINEAR_KEY=$(cat ~/.linear-token 2>/dev/null)
[ -z "$LINEAR_KEY" ] && exit 0
DETAILS=$(curl -s -X POST https://api.linear.app/graphql \
    -H "Authorization: $LINEAR_KEY" \
    -H 'Content-Type: application/json' \
    -d "{\"query\": \"{ issueSearch(query: \\\"$ISSUE\\\") { nodes { title description state { name } assignee { name } } } }\"}" 2>/dev/null | jq -r '.data.issueSearch.nodes[0] | "Title: \(.title)\nStatus: \(.state.name)\nDescription: \(.description[0:300])"' 2>/dev/null)
[ -z "$DETAILS" ] && exit 0
jq -n --arg d "Linear issue $ISSUE:
$DETAILS" '{"hookSpecificOutput":{"hookEventName":"UserPromptSubmit","additionalContext":$d}}'
exit 0

This auto-fetches Linear issue context when you mention an issue ID in your prompt.

+1

+1

+1

In that case, wouldn't cyrus be more suitable? https://www.atcyrus.com/

If you're using Claude Desktop, you can add "custom link" to open the issue in Claude Desktop. Not quite the same but better than copying manually.

<img width="858" height="280" alt="Image" src="https://github.com/user-attachments/assets/b720ea0b-4ac1-48bc-8d75-f2fb3c07b606" />

Hey all — while this is being considered, I put together a small bridge that covers most of what's described here: https://github.com/ian-klopper/claude-linear-agent

What it does: assign a Linear issue to the agent user → bridge fires a Claude Code Routine with the issue context → Claude works in a cloud session off origin/main → (if you set up the Linear MCP) Claude posts progress and a final summary back on the Linear issue. You can watch the session live via a link it drops in the Linear sidebar.

What it doesn't do (yet):

• No "Claude is done" signal — Routines is fire-and-forget, so the bridge can't update Linear status on completion. You rely on Claude commenting back via MCP.
• Replies in the Linear thread start a fresh Claude session (Routines has no "continue session" API). The Routine prompt tells Claude to re-read prior Linear comments via MCP, but long back-and-forths get expensive.

Upfront: I'm an amateur dev. This works for me but definitely has rough edges — issues/PRs very welcome.

Hope it helps someone here.

+1

+1

+1

This would be a valuable addition. As competing tools like Codex continue to expand their integrations, including native Linear agent assignment, the gap is becoming a real factor in tooling decisions. First-party support for assigning Linear issues to Claude Code would meaningfully reduce that pressure and keep Claude Code competitive in workflows where Linear is the source of truth.

+1

+1

+1