K3Nova
Run a one-shot Kimi Code agent task
This checks local CLI, model access, and non-interactive output.
kimi -m k3-256k -p "Summarize this repository in five bullet points."Goal and short answer
A local Kimi K3 agent usually means a local tool or workspace controlling tasks while Kimi K3 runs through an authorized endpoint. That can be Kimi Code, Kimi Claw/OpenClaw, K3Nova, or another client configured with the right model ID.
Do not confuse a local agent with full local weights. The agent can be local even when inference runs through Kimi's hosted API or another approved provider.
Copyable commands
Start with a small local task and a known model ID before adding tools or a gateway.
Open the local Kimi Code web surface
Use this when you want a browser UI backed by the local CLI server.
kimi web --no-open --port 58628List OpenClaw models before a gateway test
Use the exact provider/model string that your OpenClaw setup exposes.
openclaw models list
openclaw models list --jsonRun a narrow OpenClaw Kimi K3 smoke
Run this from an OpenClaw checkout with provider credentials already configured.
OPENCLAW_LIVE_GATEWAY_MODELS="moonshot/kimi-k3" pnpm test:live src/gateway/gateway-models.profiles.live.test.tsAI handoff prompt and permissions
Copy this prompt into Kimi Code, K3Nova, or another AI agent. Keep approval manual for file writes, shell commands, account actions, and secrets.
Copyable AI prompt
Use this to give the agent the task and safety boundary in one message.
You are my AI agent for this task: set up a narrow local-agent smoke test for Kimi K3 using Kimi Code, OpenClaw, or K3Nova while keeping credentials safe.
Start by restating the goal and the permissions you need.
Use official docs or the files I provide before making claims.
Give me a direct answer first, then a short table or checklist.
If commands are needed, show exact copyable commands without a shell prompt.
Ask before writing files, running shell commands, deleting or moving data, logging in, spending money, changing account settings, or handling API keys.
Stop and ask me when a step requires secrets, payment, account access, destructive cleanup, or a permission broader than the task.Recommended AI permissions
| Permission | Give AI | Why |
|---|---|---|
| Read-only project access | Allow Read, Grep, and Glob only in the selected project or config folder. | Lets the agent inspect state before acting. |
| Shell commands | Ask before Bash or any CLI command; run one command at a time. | Commands can change files, packages, sessions, or network state. |
| File edits | Ask before Write or Edit, and require a diff before applying changes. | Prevents hidden config or source edits. |
| Hard stops | Deny payment, login, secret exposure, account changes, and destructive delete commands. | These are not safe to automate casually. |
Runbook
Choose the agent surface
Use Kimi Code for terminal and IDE coding work, Kimi Claw or OpenClaw for deployed assistants, and K3Nova for workspace evaluation.
Set the model explicitly
Use the model ID required by the surface, such as kimi-k3 in API contexts or k3 in Kimi Code.
Protect credentials
Keep provider keys in documented config or server-side secrets, not in prompts or public files.
Run a narrow first task
Ask the agent to inspect a small file, summarize a document, or produce a short plan before giving it a larger job.
Local agent paths
| Surface | Local part | Kimi K3 part |
|---|---|---|
| Kimi Code | Terminal or IDE agent session | Configured Kimi Code model access |
| Kimi Claw | OpenClaw instance or linked assistant | Kimi model configuration |
| OpenClaw | Gateway, agent session, tools, and tests | Provider/model such as moonshot/kimi-k3 when configured |
| K3Nova | Local web workspace and server | Server-side provider access when enabled |
Notes before you continue
Kimi Agent
Official docs position Kimi Agent as an end-to-end assistant powered by Kimi K3 with tools for websites, documents, data, and more.
Kimi Claw and OpenClaw
Kimi Claw can deploy or link an OpenClaw instance, and OpenClaw docs show how model smoke tests treat provider IDs.
K3Nova
K3Nova can serve as a local web workspace for testing prompts, pages, and access boundaries before production deployment.
FAQ
Does a local Kimi K3 agent require local model weights?
No. The local agent can call a hosted model endpoint.
Which model ID should I use?
Use the exact ID required by the tool: Kimi Code and API contexts use different naming.
What should the first local task be?
Use a small, reversible task that proves the tool, model, and credential path are configured correctly.