MoonshotAI / MoonshotAI/kimi-cli
ACP integration: Session history is not replayed or available to clients (e.g., Zed, JetBrains, ...)
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 11.4k
- Forks
- 1.3k
- Avg merge
- 9h 47m
- Merged PRs (30d)
- 2
Description
Problem
When using kimi-cli as an ACP agent (e.g. with Zed or JetBrains IDEs), loading or switching a session does not replay or provide past conversation history to the client. This means all ACP-based integrations always start at a blank context, even if older session history exists.
- The ACP spec expects that, when loading a session, the complete conversation history is provided to the client via notifications. This matches typical user expectations in IDEs (history/chat review buttons, navigation, etc.).
- In the Web UI and CLI, history replay is already implemented and works as expected, using
context.jsonlandwire.jsonlfiles for persistence. - With ACP, history seems to be totally absent; e.g., in Zed, every session starts with 0 messages. This makes context continuity and long-running chat impossible.
- There is a closed issue ([#824](https://github.com/MoonshotAI/kimi-cli/issues/
What does it mean for Zed Users?
When inspecting History, nothing shows up
And after user clicks History -> View All
However, when using /session in the Kimi CLI directly, it correctly shows the Sessions History
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start at the ACP load_session implementation and compare it with the existing Web UI and CLI replay paths using context.jsonl and wire.jsonl. Check the ACP specification's load_session expectations and reproduce the missing history through a client such as Zed. Done means an existing session's complete conversation is available after loading or switching.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- api, cli
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 50/100