An explicit tool-feedback loop
Study how a task continues from command and file results, then define the evidence your own workflow needs at each step.
Kimi Code CLI
Study Kimi Code CLI’s tool loop, focused subagents, lifecycle hooks, and editor protocol as references for your own workflows. Use them to clarify task roles, execution checks, and context handoffs around work in Floatboat.
About Kimi Code CLI
Kimi Code CLI is Moonshot AI’s current open-source coding agent for the terminal. Its official description includes reading and editing code, running commands, searching files, and fetching web pages, with the agent choosing subsequent steps from tool feedback. It is an application around the model, not a Skill package.
For a Floatboat user improving a workflow, the useful ideas are role separation, explicit tool feedback, and well-defined points for checks. Kimi Code CLI provides a concrete reference for those mechanisms. ACP also illustrates how an interface can remain separate from the process that executes an agent task.
The maintained repository is MoonshotAI/kimi-code and uses MIT. The earlier Python project, MoonshotAI/kimi-cli, was archived and directs users to the newer CLI. Keep that distinction in mind when reading tutorials: the old project’s architecture, package names, and instructions are not interchangeable with the current release.
Workflow Mechanisms to Learn From
Understand what the project demonstrates, then decide which mechanism addresses a real gap in your own workflow.
Focused task responsibilities
Use subagent contexts as a reference for dividing exploration, planning, and implementation without losing the parent task’s goal.
Checks at lifecycle boundaries
Study hooks and interface protocols when deciding where to validate an action and how context should pass between tools.
Apply the Ideas to Your Workflow
Use these scenarios as design references for work organized in Floatboat. Any software integration needs its own compatibility and permission checks.
A multi-role coding workflow
Separate research, planning, and implementation outputs so each role contributes a result the next stage can inspect.
Checks before consequential steps
Design a validation or review point before a command, file change, or external action proceeds.
A consistent context handoff
Use the interface–runtime separation as a reference when a workflow spans an editor, a workspace, and an execution process.
From Reference to a Workflow Experiment
Start with a task and a small testable change, whether that means better instructions, a new tool connection, or a runtime component.
- 01
Use the maintained reference
Use kimi-code documentation when studying the mechanisms; legacy Python CLI examples can describe a different architecture.
- 02
Define each role and checkpoint
Specify what each delegated task receives, what it returns, and which operation must wait for review.
- 03
Test the context transfer
Check whether each stage gets enough relevant information without carrying unrelated conversation history into every step.
Related Agent Projects
Compare independent agents through their interfaces, extension models, and approach to carrying work forward.
Kimi Code CLI FAQ
Which Kimi repository is current?
Use MoonshotAI/kimi-code for Kimi Code CLI. MoonshotAI/kimi-cli is the archived Python implementation and explicitly directs users to the current project.
Is Kimi Code CLI an independent agent project?
Yes. It includes an interactive application, tools, sessions, and agent execution. It is not just a prompt, a model, or a collection of Skills, even though Skills can extend its workflows.
Can it use a provider other than Kimi?
The README says other compatible providers can be configured. Verify the provider’s protocol and feature support rather than assuming that every model endpoint behaves identically.
What does ACP integration provide?
It allows compatible editors to drive a Kimi Code CLI session over the Agent Client Protocol. The editor supplies its interface while the CLI handles the agent session; the editor must support the integration.
Must I switch products to use these workflow ideas?
No. You can first use Kimi Code CLI as a reference for task structure, tool boundaries, and result review in your existing workflow. Reusing software is a separate choice that requires compatibility, permissions, and license checks; applying the design ideas does not automatically connect the software to Floatboat.
Does MIT cover a Kimi subscription?
No. MIT applies to the project source. Kimi Code account access, Moonshot API use, and any other provider’s service are governed by their own entitlements, limits, and terms.