Linear Labels: AI vs Human (for Agents and Team)
Purpose: Enable filtering in Linear to see “all tasks that can be assigned to AI.” Agents and humans use these labels consistently when creating tickets.
References: .cursor/rules/linear-mcp-ticket-standards.mdc, standards/04-prompts/tickets/linear-ticket-generation-from-transcript.md
Label scheme
| Label | Meaning | When to use |
|---|---|---|
| ai-assignable | AI (Cursor, Codex, cloud agent) can implement this. Human reviews before close. | Implementation tickets with clear acceptance criteria, code changes, docs, tests. |
| human-only | Human must do this; do not assign to AI. | Security review, stakeholder sign-off, subjective UX decisions, sensitive client communication, policy decisions. |
Usage
- Filter for AI work:
label:ai-assignablein Linear — shows all tasks that can be assigned to AI - Optional:
human-reviewfor tickets that need explicit human sign-off before closing (can apply to ai-assignable tickets: AI implements, human reviews) - Deprecate:
cursor-generatedas the “AI did it” signal — useai-assignablefor “AI can do it” at creation time
Setup
Create ai-assignable and human-only labels in Linear (team or org level) if they don’t exist:
- Linear → Settings → Labels (or team Labels)
- Create label: ai-assignable (suggested color: blue or green)
- Create label: human-only (suggested color: amber or red)
Agent instructions
When creating tickets via Linear MCP (save_issue):
- Add
labels: ["ai-assignable"]for implementation work an agent can do - Add
labels: ["human-only"]for work that must be done by a human - Leave
delegateunset for human-only; optionally set for agent-assignable work when delegating to an agent