feat(kimi-inspect): add kap-server web inspector with dev-only /api/v1/debug RPC surface - #1806
Merged
Merged
Conversation
- new apps/kimi-inspect app: connect screen (server URL + optional bearer token, persisted in localStorage, deep-linkable via ?url=/?token=), workspace/session browser sidebar, per-session chat view, and live Service panels with data and trigger buttons for Session/Agent scopes - built on @moonshot-ai/klient (HTTP for calls, /api/v2/ws for events); Vite dev server proxies /api to a running kap-server - register the workspace in AGENTS.md project map and flake.nix workspacePaths/workspaceNames
🦋 Changeset detectedLatest commit: 86ee696 The changes in this PR will be included in the next version bump. This PR includes changesets to release 1 package
Not sure what this means? Click here to learn what changesets are. Click here if you're a maintainer who wants to add another changeset to this PR |
commit: |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 6e9dd59c15
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
Comment on lines
+188
to
+190
| const entries = useMemo( | ||
| () => [...mapHistory((history.data ?? []) as readonly HistoryMessage[]), ...stream], | ||
| [history.data, stream], |
There was a problem hiding this comment.
Because ChatView remains mounted while App only changes sessionId/agentId, stream and running are not scoped to the selected conversation. If a user switches sessions or agents while an optimistic prompt/streamed turn is present, these entries are still appended to the new history and the old running state can disable Send/Cancel for the wrong agent; reset or key this state on sessionId/agentId.
Useful? React with 👍 / 👎.
| if (fn === undefined) { | ||
| return Promise.reject(new Error(`no such method on proxy: ${method}`)); | ||
| } | ||
| return fn(arg); |
There was a problem hiding this comment.
call always invokes the proxy with exactly one argument. Through HttpChannel this serializes no-arg calls as [null], and a JSON array entered for a multi-parameter dynamic method becomes nested; the v2 dispatcher only spreads the top-level array (packages/kap-server/src/transport/dispatcher.ts:129), so generic buttons for methods such as sessionApprovalService.decide(id, response) pass the wrong arguments, and optional no-arg methods like sessionInteractionService.listPending() get null instead of no filter.
Useful? React with 👍 / 👎.
- kimi-inspect: replace @moonshot-ai/klient with an in-app old-klient-style channel layer (service-bound IChannel, HTTP ProxyChannel, shared /api/v2/ws socket with ref-counted event listens), typed by agent-core-v2 interfaces; /channels descriptors + serviceByName keep every wire protocol loaded 1:1 - kimi-inspect: local server auto-discovery (Vite middleware over the kap-server instance registry + home token), zero-config startup connect, and a header switcher for runtime server switching - kap-server: wire the dormant --debug-endpoints flag to a new whitelist-free /api/v1/debug dispatcher (every scoped service callable), gated to loopback binds; repo dev scripts pass the flag - kimi-inspect: probe the debug surface at connect, falling back to /api/v2 on servers without it - tests: channel + discovery unit tests in kimi-inspect; debug RPC and loopback-gating coverage in the kap-server rpc/debugNonloopback suites
sailist
changed the title
The private dev app never ships, so it should never appear in a changeset. Add it to the changeset config ignore list (next to vis*) and note the rule in the gen-changesets skill.
Merged
7723qqq
pushed a commit
to 7723qqq/kimi-code
that referenced
this pull request
Jul 17, 2026
…1/debug RPC surface (MoonshotAI#1806) * feat(kimi-inspect): add web inspector for kap-server /api/v2 surface - new apps/kimi-inspect app: connect screen (server URL + optional bearer token, persisted in localStorage, deep-linkable via ?url=/?token=), workspace/session browser sidebar, per-session chat view, and live Service panels with data and trigger buttons for Session/Agent scopes - built on @moonshot-ai/klient (HTTP for calls, /api/v2/ws for events); Vite dev server proxies /api to a running kap-server - register the workspace in AGENTS.md project map and flake.nix workspacePaths/workspaceNames * fix(kimi-inspect): align dependency versions with the workspace (sherif) * feat: dev /api/v1/debug RPC surface and kimi-inspect channel rework - kimi-inspect: replace @moonshot-ai/klient with an in-app old-klient-style channel layer (service-bound IChannel, HTTP ProxyChannel, shared /api/v2/ws socket with ref-counted event listens), typed by agent-core-v2 interfaces; /channels descriptors + serviceByName keep every wire protocol loaded 1:1 - kimi-inspect: local server auto-discovery (Vite middleware over the kap-server instance registry + home token), zero-config startup connect, and a header switcher for runtime server switching - kap-server: wire the dormant --debug-endpoints flag to a new whitelist-free /api/v1/debug dispatcher (every scoped service callable), gated to loopback binds; repo dev scripts pass the flag - kimi-inspect: probe the debug surface at connect, falling back to /api/v2 on servers without it - tests: channel + discovery unit tests in kimi-inspect; debug RPC and loopback-gating coverage in the kap-server rpc/debugNonloopback suites * chore(changesets): ignore @moonshot-ai/kimi-inspect The private dev app never ships, so it should never appear in a changeset. Add it to the changeset config ignore list (next to vis*) and note the rule in the gen-changesets skill.
7723qqq
pushed a commit
to 7723qqq/kimi-code
that referenced
this pull request
Jul 17, 2026
…1/debug RPC surface (MoonshotAI#1806) * feat(kimi-inspect): add web inspector for kap-server /api/v2 surface - new apps/kimi-inspect app: connect screen (server URL + optional bearer token, persisted in localStorage, deep-linkable via ?url=/?token=), workspace/session browser sidebar, per-session chat view, and live Service panels with data and trigger buttons for Session/Agent scopes - built on @moonshot-ai/klient (HTTP for calls, /api/v2/ws for events); Vite dev server proxies /api to a running kap-server - register the workspace in AGENTS.md project map and flake.nix workspacePaths/workspaceNames * fix(kimi-inspect): align dependency versions with the workspace (sherif) * feat: dev /api/v1/debug RPC surface and kimi-inspect channel rework - kimi-inspect: replace @moonshot-ai/klient with an in-app old-klient-style channel layer (service-bound IChannel, HTTP ProxyChannel, shared /api/v2/ws socket with ref-counted event listens), typed by agent-core-v2 interfaces; /channels descriptors + serviceByName keep every wire protocol loaded 1:1 - kimi-inspect: local server auto-discovery (Vite middleware over the kap-server instance registry + home token), zero-config startup connect, and a header switcher for runtime server switching - kap-server: wire the dormant --debug-endpoints flag to a new whitelist-free /api/v1/debug dispatcher (every scoped service callable), gated to loopback binds; repo dev scripts pass the flag - kimi-inspect: probe the debug surface at connect, falling back to /api/v2 on servers without it - tests: channel + discovery unit tests in kimi-inspect; debug RPC and loopback-gating coverage in the kap-server rpc/debugNonloopback suites * chore(changesets): ignore @moonshot-ai/kimi-inspect The private dev app never ships, so it should never appear in a changeset. Add it to the changeset config ignore list (next to vis*) and note the rule in the gen-changesets skill.
ywh114
pushed a commit
to ywh114/kimi-code
that referenced
this pull request
Jul 19, 2026
…1/debug RPC surface (MoonshotAI#1806) * feat(kimi-inspect): add web inspector for kap-server /api/v2 surface - new apps/kimi-inspect app: connect screen (server URL + optional bearer token, persisted in localStorage, deep-linkable via ?url=/?token=), workspace/session browser sidebar, per-session chat view, and live Service panels with data and trigger buttons for Session/Agent scopes - built on @moonshot-ai/klient (HTTP for calls, /api/v2/ws for events); Vite dev server proxies /api to a running kap-server - register the workspace in AGENTS.md project map and flake.nix workspacePaths/workspaceNames * fix(kimi-inspect): align dependency versions with the workspace (sherif) * feat: dev /api/v1/debug RPC surface and kimi-inspect channel rework - kimi-inspect: replace @moonshot-ai/klient with an in-app old-klient-style channel layer (service-bound IChannel, HTTP ProxyChannel, shared /api/v2/ws socket with ref-counted event listens), typed by agent-core-v2 interfaces; /channels descriptors + serviceByName keep every wire protocol loaded 1:1 - kimi-inspect: local server auto-discovery (Vite middleware over the kap-server instance registry + home token), zero-config startup connect, and a header switcher for runtime server switching - kap-server: wire the dormant --debug-endpoints flag to a new whitelist-free /api/v1/debug dispatcher (every scoped service callable), gated to loopback binds; repo dev scripts pass the flag - kimi-inspect: probe the debug surface at connect, falling back to /api/v2 on servers without it - tests: channel + discovery unit tests in kimi-inspect; debug RPC and loopback-gating coverage in the kap-server rpc/debugNonloopback suites * chore(changesets): ignore @moonshot-ai/kimi-inspect The private dev app never ships, so it should never appear in a changeset. Add it to the changeset config ignore list (next to vis*) and note the rule in the gen-changesets skill.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Related Issue
No linked issue (internal dev-tooling PR) — the problem is described below.
Problem
Debugging the v2 (kap-server) engine today means either driving klient's typed facade or hand-crafting
/api/v2calls: there is no interactive way to browse workspaces/sessions, watch a live agent, or inspect and invoke Services. Two gaps make this worse:--debug-endpointsflag has been fully plumbed (startServeroption, CLI flag, loopback gating instart.ts) but mounts zero routes — a dead end for exactly this use case.~/.kimi-code/server.tokenby hand, with no discovery of the instances already running on the machine.What changed
1. New app:
apps/kimi-inspect— web inspector for the/api/v2surfacePrivate dev app (not published), React 19 + Vite + react-query. Workspace/session browser, per-session chat with live deltas, dynamic Service panels (data + trigger buttons) driven by the server's
/channelsintrospection, pending-interaction handling, and a merged core/session/agent WS event log. Registered inpnpm-workspace.yaml,flake.nixworkspace lists, and the rootAGENTS.mdproject map.2. In-app channel layer (old-klient / VS Code
ProxyChannelmodel)The app composes its own transport (
apps/kimi-inspect/src/channel/) instead of deep-importing klient internals: service-boundIChannel(call/listen), an HTTPProxyChannelrouting to the RPC surface,makeProxytyped byagent-core-v2interfaces/decorator ids, and a shared/api/v2/wssocket with ref-counted service-emitterlistens (soonDidXxxmembers work 1:1)./{rpcBase}/channelsdescriptors +serviceByNamematerialize every exposed Service into a typed proxy — all protocols load 1:1. This drops the app's@moonshot-ai/klientdependency.3. Zero-config local development
A dev middleware (
GET /__inspect/servers) scans the local kap-server instance registry (~/.kimi-code/server/instances+ legacylock) and the home token, so the inspector auto-connects on startup (no URL/token typing) and can switch between running servers from a header dropdown at runtime. Manual URL/token and?url=deep links keep working unchanged; statically hosted builds fall back to the manual form.4. kap-server: dev-only
/api/v1/debugRPC surfaceWires the dormant
--debug-endpointsflag to a real surface (src/transport/registerDebugRoutes.ts): the same reflection dispatcher and envelope semantics as/api/v2, but the channel lookup spans the whole scoped DI registry — every scoped Service (App/Session/Agent) is callable and listed via/api/v1/debug/channels. Gated to loopback binds by the pre-existingstart.tscheck and still behind the global bearer-auth hook. The/api/v2whitelist is untouched (klient's production contract is unchanged). The dispatcher's channel lookup was parameterized and route registration factored into a shared helper used by both surfaces. Repo dev scripts (dev:server,dev:kap-server,dev:kap-server:multi) now pass--debug-endpoints; the inspector probes/api/v1/debugat connect and falls back to/api/v2on servers without it.Tests
test/rpc.test.ts(full listing, non-whitelisted call, unknown-service 40001, bearer gating), andtest/debugNonloopback.e2e.test.tspins the three mount states (public-bind 404 / loopback default-off 404 / loopback mounted). Full suite: 669 tests green.src/channel/channel.test.ts) and server-discovery unit tests (vite/serverDiscovery.test.ts).Checklist
gen-changesets:.changeset/debug-rpc-surface.mdpatches@moonshot-ai/kimi-code(kap-server enters the CLI bundle);kimi-inspectis a private dev app and follows the no-changeset precedent of its initial commit.gen-docs, or this PR needs no doc update: internal dev tooling; no user-facing CLI behavior change (the flag is off by default in production).