Feature request for product/service
API Keys
Describe the request
I’m using the Cloud Agents API (POST /v0/agents) to programmatically dispatch agents.
However, I’ve noticed that the “Long-running” toggle available in the web UI at cursor.com/agents doesn’t have an equivalent API parameter. The long-running mode (planning-first, multi-agent checking) would be ideal for my larger domain builds, but I can only access it by manually pasting prompts into the web UI.
What I’ve observed:
- POST /v0/agents accepts prompt, model, source, target, webhook — no mode or longRunning field
- GET /v0/models returns composer-1.5, claude-4.6-opus-high-thinking, gpt-5.3-codex-high, gpt-5.2-high — none obviously map to “long-running”
- Agents launched via the web UI with long-running toggled on show the same schema in GET /v0/agents/{id} responses — no field distinguishes them from standard agents
- The /v0/agents/{id}/conversation endpoint returns 0 messages for long-running agents that are actively running (works fine for standard agents)
Questions:
- Is there a way to enable long-running mode via the API today that I’m missing? (e.g., a specific model name, an undocumented parameter)
- If not, is API support for long-running agents planned? An additional field like “mode”: “long-running” on the create endpoint would unlock programmatic orchestration of complex builds.