Create Issues
by Voice

The Context Switch That Kills Your Flow

You're debugging a production issue. You find the root cause — a missing null check in the order handler. You need to create a Linear issue before you forget the details.

So you open Linear. Click "New Issue." Pick the team. Type the title. Write the description. Set the priority. Add labels. Click "Create."

By the time you're done, 90 seconds have passed. You've lost the thread of what you were debugging. You switch back to the code. Where was that null check again?

What if you could just speak?

Three seconds. No browser. No clicking. No context switch. You're still in your editor, still looking at the code.

How It Works

Your voice ──▶ SpeechButton STT ──▶ Gemma 4 Local AI ──▶ Linear GraphQL API
  (7ms)         (Parakeet V3,        (structures as         (creates issue)
                 100% offline)         title + desc +
                                       priority, local)

Three components:

1. SpeechButton captures and transcribes your voice locally on Apple Neural Engine
2. Local AI (Gemma 4) reads your prompt file and extracts a structured issue (title, description, priority) from your transcription — entirely on your Mac
3. A Python script in your integrations/ folder sends the JSON to Linear's GraphQL API

The result: a well-formatted Linear issue from your spoken description, in under 4 seconds total. Everything is local except the final Linear API call — no Anthropic API key needed.

Setup

Five steps. Under five minutes.

# ~/.config/speechbutton/config.toml

[global]
model = "parakeet-tdt-0.6b-v3-int8"
language = "en"
auto_punctuation = true

[audio]
vad_enabled = true
vad_silence_threshold = 1.0

# Linear — create issue from voice
[[hotkey]]
key = "RightCommand"
channel = "4"
name = "linear"
transform = "prompts/linear_issue.md"
exec = "LINEAR_API_KEY=lin_api_xxx integrations/send_linear.py"

You are a task formatter for Linear issue tracker.

Convert raw speech into a JSON object. Extract:
- title: concise issue title (under 80 chars)
- description: detailed markdown description
- priority: 1=urgent, 2=high, 3=medium, 4=low

Output ONLY valid JSON:
{"title": "...", "description": "...", "priority": 3}

{
  "title": "Order handler crashes on null shipping address",
  "description": "Missing null check in `process_order` at line 142.\n\nAffects ~5% of international orders where the shipping address field is optional.\n\n**Impact:** Crashes on order processing for international customers.",
  "priority": 2
}

#!/usr/bin/env python3
"""Create a Linear issue from stdin JSON."""
import json, os, sys, urllib.request

API_URL = "https://api.linear.app/graphql"

def graphql(key, query, variables=None):
    payload = {"query": query}
    if variables: payload["variables"] = variables
    req = urllib.request.Request(API_URL, json.dumps(payload).encode(),
        {"Content-Type": "application/json", "Authorization": key})
    return json.loads(urllib.request.urlopen(req, timeout=15).read()).get("data", {})

def main():
    text = sys.stdin.read().strip()
    if not text: sys.exit(0)

    # Strip markdown code fences
    if text.startswith("```"):
        lines = text.split("\n")
        text = "\n".join(lines[1:-1] if lines[-1].startswith("```") else lines[1:])

    try:
        data = json.loads(text)
    except json.JSONDecodeError:
        data = {"title": text[:80], "description": text, "priority": 3}

    key = os.environ.get("LINEAR_API_KEY")
    if not key:
        print("LINEAR_API_KEY not set", file=sys.stderr)
        sys.exit(1)

    # Auto-detect first team
    teams = graphql(key, "query { teams { nodes { id name } } }")
    team_id = teams["teams"]["nodes"][0]["id"]

    # Create issue
    mutation = """mutation($input: IssueCreateInput!) {
        issueCreate(input: $input) { success issue { identifier title url } }
    }"""
    result = graphql(key, mutation, {"input": {
        "teamId": team_id,
        "title": data.get("title", "Untitled"),
        "description": data.get("description", ""),
        "priority": data.get("priority", 3),
    }})
    issue = result["issueCreate"]["issue"]
    print(f"Created {issue['identifier']}: {issue['title'][:50]}")

if __name__ == "__main__":
    main()

chmod +x ~/.config/speechbutton/integrations/send_linear.py

Real Workflows

Advanced: Multiple Teams, Auto-Assign

For teams with multiple Linear teams (Engineering, Design, DevOps), set up a channel per team:

# Engineering bugs
[[hotkey]]
key = "RightCommand"
channel = "4"
name = "linear-eng"
transform = "prompts/linear_issue.md"
exec = "LINEAR_API_KEY=lin_api_eng_xxx integrations/send_linear.py"

# DevOps issues
[[hotkey]]
key = "RightOption"
channel = "5"
name = "linear-devops"
transform = "prompts/linear_issue.md"
exec = "LINEAR_API_KEY=lin_api_devops_xxx integrations/send_linear.py"

Same prompt file, different API keys scoped to each team. RightCommand for engineering bugs, RightOption for infra issues. The right team gets the right issues without you navigating team selectors.

You can also extend the transform to auto-assign based on keywords. Mention "database" and it assigns to your DBA. Mention "auth" and it routes to the security lead. The transform script is just code — make it as smart as you need.

Why Voice Beats Typing for Issue Creation

Linear issues created by typing tend to be either too terse ("fix auth bug") or too slow to write (two minutes of formatting). Voice hits the sweet spot:

The result is a better issue than you'd type manually, created in a fraction of the time. The AI transform handles the formatting you'd skip when rushing and the structure you'd over-think when not.

And because SpeechButton captures in 7ms, you can rapid-fire three issues in sequence without pausing between them. Hold, speak, release. Hold, speak, release. Hold, speak, release. Three issues in your backlog before you'd finish typing the first one.

Privacy

Here's exactly what stays on your Mac and what goes to the cloud:

With the default local AI transform, everything stays on your Mac except the final Linear API call: voice → local STT (Parakeet V3, Apple Neural Engine) → local AI transform (Gemma 4, on your Mac) → Linear API. No audio leaves your machine. No transcription leaves your machine. Only the structured issue data reaches Linear's servers — which is where it needs to go anyway. No Anthropic API key required. If you prefer higher accuracy at the cost of privacy, you can optionally switch to the Claude API transform (Option B) — in that case, transcribed text is sent to Anthropic's servers for structuring.

Get Started