inspector/specification/v2_cli_tui_launcher.md
Cliff Hall 5efe9ddc7b
refactor: self-validating clients, fast validate + separate coverage/smoke (#1504) (#1505)
* refactor(launcher): self-contained validate; split smokes into top-level step (#1504)

Previously `validate:launcher` built cli/tui/web and ran the cross-client
smokes, so the launcher's "validation" depended on every other client
being built — and the launcher had no lint/format/coverage of its own.

- clients/launcher: add a web-style `validate` (format:check && lint &&
  build && test:coverage) with a Node-only eslint flat config, prettier,
  and a per-file coverage gate (parse-launcher-argv.ts covered;
  src/index.ts excluded as binary bootstrap). 100% on the gated file.
- Root `validate:launcher` now just delegates: `cd clients/launcher &&
  npm run validate` — no dependency on the other clients.
- New top-level `smoke` step (smoke:launcher + smoke:cli + smoke:tui +
  smoke:web) with NO build commands; `validate` runs it last. Each
  client's own validation builds its bundle first, so nothing builds
  twice (cli/tui via test:coverage, launcher via its validate;
  smoke:web builds dist on demand).
- clients/tui test:coverage now builds the TUI bundle (mirroring cli) so
  the build-free smoke has it available.
- CI: replace the standalone cli/tui/launcher build step with a
  "Validate launcher" step. Docs updated (AGENTS.md, launcher README).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* refactor(cli,tui,root): every client self-validates; smoke separate from validate

Extends the launcher self-validate work to the CLI and TUI, and makes the
top-level `validate` nothing but the per-client validations — smokes are
now a separate top-level target.

- clients/cli, clients/tui: add a web-style `validate` with a Node flat
  eslint config + prettier. CLI lint/format pass as-is. TUI registers
  eslint-plugin-react-hooks for the classic rules only (rules-of-hooks +
  exhaustive-deps, which the source already references); the stricter
  react-hooks@7 rules are not enforced on the interim component surface
  (#1501).
- Avoid double builds:
  - CLI's e2e spawns the built binary, so its `test:coverage` builds first;
    `validate` folds that in (`format:check && lint && test:coverage`, no
    separate `build`).
  - TUI tests don't spawn a bundle, so `validate` keeps an explicit
    `build` (for the smoke) and `test:coverage` is pure.
- Root: `validate` = validate:web && validate:cli && validate:tui &&
  validate:launcher (each delegates to the client's own validate). `smoke`
  is no longer part of `validate` — it's a separate top-level target run
  after a build/validate, with no build commands of its own.
- CI: per-client "Validate X" steps + a single "Run cross-client smokes"
  (npm run smoke). Docs updated (AGENTS.md, cli/tui READMEs).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* chore(root): drop redundant test aggregator scripts

Now that every client self-validates and the top-level `validate` chains
the per-client validations, the root `test` / `test:web` / `test:cli` /
`test:tui` / `test:coverage` aggregators are redundant — run
`npm run validate` (all) or `cd clients/<name> && npm run validate` (one).
`test:storybook` stays (it's the one check not part of validate). CI was
already using each client's own scripts, so it's unaffected. Docs updated.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* test(web): silence happy-dom infra noise in unit runs

Passing unit runs printed dozens of alarming-but-expected lines that made
green runs look broken — all from happy-dom performing real I/O with no
server behind it:
- `<iframe src="/sandbox.html">` (MCP Apps sandbox) made happy-dom fetch
  and later unload that URL, emitting ~40 `DOMException [NetworkError/
  AbortError]` + `AsyncTaskManager destroyed` lines. Disable child-frame
  navigation in the unit project (the tests only assert on the iframe
  element/attributes, not its document), removing the noise with no
  coverage loss.
- App-level renders read `GET /api/config` via useSandboxUrl /
  useServerListWritable, 404-ing ~76x. Add a benign default `fetch` (empty
  200) in the unit setup; tests that exercise fetch still override it.

Negative tests that assert on errors already suppress their own console
output (e.g. jsonUtils/schemaUtils); CLI, TUI, and the web integration
suite were already clean. Unit suite is now noise-free; 2424 web tests
pass and the coverage gate is unchanged.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* refactor(scripts): make validate fast (test), split coverage into its own target

Each client's `validate` now runs `test` instead of `test:coverage`, so
the everyday pre-push check skips v8 instrumentation (and, for web, the
slower integration project). The per-file coverage gate moves to a
dedicated top-level `coverage` script:

  coverage = coverage:web && coverage:cli && coverage:tui && coverage:launcher
  coverage:<name> = cd clients/<name> && npm run test:coverage

`validate` still builds each client's bundle (cli via its test `pretest`,
the others via an explicit `build`), so `npm run validate && npm run
smoke` works unchanged. CI now runs `validate` (fast checks + builds),
then `coverage` (the gate), then `smoke` — collapsing the old per-client
steps. Docs updated.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* ci: drop the coverage gate from CI; keep web integration as its own step

CI no longer runs `npm run coverage` — the per-file coverage gate is now a
local/manual check (`npm run coverage` before pushing). To avoid silently
dropping the web integration suite (it only ran via test:coverage, and
`validate` runs the unit project only), CI gains a standalone
`clients/web` `test:integration` step that exercises the real
stdio/HTTP/OAuth/storage paths without coverage instrumentation.

CI is now: validate (fast) → web integration → smoke → storybook. Docs
updated.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* refactor(smoke): give smoke:launcher its own script like the others

smoke:launcher was the odd one out — an inline chain of three
`node clients/launcher/build/index.js ... --help` commands while smoke:cli
/ smoke:tui / smoke:web each run a `scripts/smoke-*.mjs`. Extract it to
`scripts/smoke-launcher.mjs`, which runs the same `--help` / `--cli --help`
/ `--tui --help` dispatch checks but now also asserts each prints that
mode's usage banner (proving the launcher resolved the right client build)
and verifies the required builds exist up front with a clear error. Prints
a `smoke:launcher OK — ...` summary in the same style as the rest. Docs
updated.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* chore: reorder package.json scripts for readability

Move frequently-run scripts toward the top and rarely-run-by-hand ones
(pretest hooks, lint/format internals, etc.) toward the end, consistently
across the root and client manifests. Pure reordering — no script names or
values changed, and no other manifest content is touched.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* docs: drop references to removed root inspector / test:storybook scripts

The reordering dropped the root `inspector` and `test:storybook`
aggregators (intentional — `inspector` is redundant with `web`, and
storybook is run from clients/web in CI). Update AGENTS.md so it no longer
points at `npm run inspector`, and run `test:storybook` from `clients/web/`.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* chore: address PR #1505 review (format glob gap, fetch-baseline comment)

- TUI format/format:check now also cover the client-root source files
  (index.ts, tui.tsx, dev.ts) that live outside src/ — `eslint .` already
  lints them, so they were linted but not format-checked. (cli/launcher
  have no source outside src/; their root .ts are single-quoted configs
  intentionally left out of prettier, matching web.)
- Reword the setup.ts default-fetch comment: it overstated auto-restore
  (the unit project sets no restoreMocks/unstubGlobals), and now also
  flags that the permissive 200 default means a test asserting on a fetch
  FAILURE must install its own rejecting stub.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* chore: remove redundant root install:clients script

`install:clients` ran the same `node scripts/install-clients.mjs` as
`postinstall`, which fires automatically on `npm install` — so the manual
alias was redundant. Drop it and the doc references (AGENTS.md, spec).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* fix(ci): skip smoke:tui in CI (Ink needs a real TTY)

This PR made CI run the full `npm run smoke`, which surfaced that
smoke:tui fails in headless CI: the Ink TUI throws "Raw mode is not
supported" on mount (no TTY) and exits before rendering its first frame.
The old CI never ran smoke:tui, so this was latent.

smoke-tui.mjs now self-skips when `process.env.CI` is set — the boot/
render check is inherently local-only (a TUI needs a terminal). The TUI is
still built and unit-tested in CI; `npm run smoke` stays the single
command everywhere (full locally, tui-skipped in CI). Verified
`CI=true npm run smoke` is green (launcher/cli OK, tui SKIPPED, web OK).
Docs + CI comment updated.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-21 11:41:33 -04:00

240 lines
20 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

# Inspector V2 Tech Stack - CLI, TUI, and Launcher
### [Brief](README.md) | [V1 Problems](v1_problems.md) | [V2 Scope](v2_scope.md) | V2 Tech Stack | [V2 UX](v2_ux.md)
#### [Web Client](v2_web_client.md) | CLI, TUI, Launcher | [Server](v2_server.md) | [Storage](v2_storage.md)
## Summary
v2 ships three non-web Inspector incarnations alongside the web client: a **one-shot CLI**, an **interactive TUI**, and a **launcher** that routes to web, CLI, or TUI from a single `mcp-inspector` binary. All three consume the same `core/` source as the web client via the `@inspector/core` path alias and run on the shared `InspectorClient` stack ported from v1.5/main.
This document describes how those clients are built, wired, and tested today, and records known gaps. For catalog vs launch-time config semantics (`--config`, `--catalog`, import), see [Catalog and Launch Configuration](v2_catalog_launch_config.md).
## Goals
- One published entry binary (`mcp-inspector`) that forwards argv to the selected client.
- CLI and TUI use the same `InspectorClient`, managed-state classes, and shared `loadServerEntries()` resolution path — adapted to v2's `InspectorServerSettings` model (post-#1358).
- Bundle `core/` into CLI/TUI Node artifacts with **tsup** (analogous to Vite bundling core for the browser).
- Port v1.5 CLI integration tests (86 cases) and TUI smoke tests without duplicating `core/` unit tests (web suite remains canonical for core).
- CI runs CLI and TUI tests plus a launcher `--help` smoke step on every PR.
## Non-goals
- **CLI v2 sessions** (connect once, many subcommands) — tracked separately in [#1432](https://github.com/modelcontextprotocol/inspector/issues/1432).
- **npm workspaces** — v2 uses a fat root package plus per-client `package.json` for dev dependencies; the launcher resolves sibling `build/` outputs via relative paths, not workspace hoisting.
- _Why not workspaces:_ `core/` is consumed by **bundling** — a Vite alias for the browser, tsup inlining for the Node clients — not by symlinked package resolution, so workspaces' main benefit (cross-package linking) does not apply. Each client also pins `react` / `zustand` / `@modelcontextprotocol/sdk` to its own `node_modules` (see `vitest.shared.mts`) to avoid dual-package-instance hazards, which hoisting works against. And the published `@modelcontextprotocol/inspector` is a single flat fat package that workspaces would complicate rather than simplify.
- _Cost (from-source dev only):_ there is no hoisting, so each client keeps its own `node_modules`. A root `postinstall` (`scripts/install-clients.mjs`) cascades `npm install` into every client, so a single `npm install` at the repo root populates them all — re-run it after a pull that changes a client's dependencies. The cascade no-ops outside a source checkout (it exits early when running from `node_modules`, and the published tarball ships only each client's `build/`, no client `package.json`), so end users of the published package are unaffected. Set `INSPECTOR_SKIP_CLIENT_INSTALL=1` to skip the cascade (e.g. CI that installs each client itself).
- **Per-client coverage gates** — `core/` coverage stays on the web suite; CLI/TUI source gates are follow-up work (see [Known gaps](#known-gaps)).
- **Catalog CRUD in TUI** — TUI loads and connects; persistent catalog editing remains web-first today.
---
## Packaging and entrypoints
| Artifact | Path | Build | Published bin |
| ---------- | ------------------------------- | ------------------------------------------------------ | -------------------------------------------------------- |
| Launcher | `clients/launcher/` | `tsc` → `build/index.js` | Root `mcp-inspector` → `clients/launcher/build/index.js` |
| CLI | `clients/cli/` | `tsup` → `build/index.js` | `mcp-inspector-cli` (client package only) |
| TUI | `clients/tui/` | `tsup` → `build/index.js` | `mcp-inspector-tui` (client package only) |
| Web runner | `clients/web/server/run-web.ts` | `tsup` (`build:runner`) → `clients/web/build/index.js` | `mcp-inspector-web` (client package only) |
Root `package.json` (`@modelcontextprotocol/inspector` v2) publishes a **fat package**: merged runtime `dependencies`, `files` manifest listing each client's `build/` (and web `dist/`), and `prepack` → full `npm run build`. On the dev side, a `postinstall` runs `scripts/install-clients.mjs`, which cascades `npm install` into each client so one `npm install` at the repo root sets up every client; it no-ops when the package is installed as a dependency. The launcher does not declare `file:` sibling dependencies; it dynamically imports `../web/build/index.js`, `../cli/build/index.js`, or `../tui/build/index.js` relative to its own `build/` directory.
**Published runtime deps — Vite:** `vite` and `@vitejs/plugin-react` are production `dependencies` (not devDependencies) because `mcp-inspector --web --dev` starts an in-process Vite dev server at runtime (`start-vite-dev-server.ts`). A `npx @modelcontextprotocol/inspector` install therefore pulls the Vite toolchain even for CLI/TUI-only users; that install footprint is intentional so `--web --dev` works without a separate dev setup.
**Typical invocations:**
```bash
mcp-inspector # web (prod Hono server)
mcp-inspector --web --dev # web (Vite dev server)
mcp-inspector --cli --config mcp.json --server my-api --method tools/list
mcp-inspector --tui --config mcp.json
```
Root scripts `inspector`, `web`, and `web:dev` are thin wrappers around the launcher binary.
---
## Launcher
`clients/launcher/src/index.ts` reads mode only from a **launcher prefix**: a contiguous run of `--web`, `--cli`, or `--tui` immediately after the script name (default **web** when omitted). Prefix mode flags are stripped; everything from the first non-mode token onward is forwarded unchanged. Later tokens equal to `--web`/`--cli`/`--tui` are treated as app args (e.g. stdio server config), not launcher mode. More than one mode flag in the prefix is an error.
- **No core imports** — plain TypeScript compile is sufficient.
- **Argv forwarding** — app flags and positionals pass through unchanged so each client's Commander parser owns server and method options.
- **Help** — `mcp-inspector --help` shows launcher help; `mcp-inspector --cli --help` forwards to CLI help.
**Production web caveat:** launcher `prebuild` chains `build:runner` for web, not a full `vite build`. Running `mcp-inspector --web` (without `--dev`) requires `clients/web/dist/` from a prior `cd clients/web && npm run build`. CI builds CLI, TUI, and launcher before the launcher smoke step but does not currently validate prod web startup end-to-end.
**Build targets:** client `tsup` bundles target `node22`, aligned with root `engines.node` (`>=22.7.5`).
---
## Shared core consumption
All three clients import from `@inspector/core/...` (mapped to `../../core/` source).
| Concern | Web | CLI / TUI | Launcher |
| -------------- | --------------------------------- | -------------------------------------------------------- | ------------------------- |
| Dev typecheck | `tsconfig.app.json` paths | per-client `tsconfig.json` paths | `tsconfig.json` (no core) |
| Runtime bundle | Vite alias | tsup `noExternal: [/^@inspector\/core/]` + esbuild alias | n/a |
| Tests | Vitest projects in `clients/web/` | Vitest + `vitest.shared.mts` aliases | none |
`vitest.shared.mts` at repo root centralizes `@inspector/core` and test-server aliases plus bare-module pins (`react`, `pino`, SDK, etc.) so CLI/TUI Vitest configs stay aligned with web.
**Resolved design choices:**
| Topic | Decision |
| ------------------- | ------------------------------------------------------------------------------------------------------------------------------------------ |
| Core package | No separate `inspector-core` npm package; source-only `core/` |
| CLI/TUI build | tsup bundles `@inspector/core` into `build/index.js` |
| Core tests | Not duplicated under cli/tui; web unit + integration suites cover `core/` |
| Default config path | `loadServerEntries()` applies `withDefaultCatalogPath()` → `~/.mcp-inspector/mcp.json` when no `--catalog`/`--config` and no ad-hoc target |
---
## CLI
**Model:** one-shot — each invocation connects, runs a single `--method`, prints JSON to stdout, disconnects, exits. Same surface as v1.5; session-oriented CLI v2 is future work ([#1432](https://github.com/modelcontextprotocol/inspector/issues/1432)).
**Entry:** `clients/cli/src/index.ts` exports `runCli(argv)`; `src/cli.ts` owns Commander parsing and `InspectorClient` orchestration.
**Server resolution:**
1. Positional `[target...]` (command/URL) and/or `--catalog`/`--config` + `--server`, plus `-e`, `--cwd`, `--transport`, `--server-url`, `--header`.
2. `loadServerEntries(serverOptions)` in `core/mcp/node/servers.ts` — applies the default catalog path when appropriate and lifts disk settings; `selectServerEntry(entries, --server)` picks the single server to connect.
3. Disk `headers`/timeouts/OAuth (post-#1358 flat entry shape) and ad-hoc `--header` pairs both map to `InspectorServerSettings` and pass into `InspectorClient` via `serverSettings` (not `MCPServerConfig.headers`). A launch-time `--header` overrides the file's headers.
**Config-file settings (resolved #1482):** the CLI now routes file-sourced servers through the same `loadServerEntries()` → `mcpConfigToServerEntries()` path as the TUI, so persisted `headers`, timeouts, and OAuth are lifted into `serverSettings`. (Previously the CLI used bare `MCPServerConfig` and dropped them.)
**Tests:** 86 cases in `clients/cli/__tests__/` spawn `node build/index.js` via `cli-runner.ts` (`pretest` builds test-servers + CLI). Core behavior is exercised; Vitest coverage on `src/cli.ts` is invisible to the coverage instrumenter because tests run out-of-process (see [Known gaps](#known-gaps)).
---
## TUI
**Model:** interactive terminal UI (Ink + React 19) mirroring web information architecture: Servers, Tools, Prompts, Resources, Logs, OAuth, etc.
**Entry:** `clients/tui/index.ts` exports `runTui(argv)`; `tui.tsx` parses argv and renders `App`.
**Server resolution:** `loadTuiServers()` in `clients/tui/src/tui-servers.ts` — a thin re-export of the shared `loadServerEntries()` in `core/mcp/node/servers.ts`:
- **Config file path:** reads JSON, runs `mcpConfigToServerEntries()` — returns `{ config, settings }` per server (correct post-#1358 path; matches web `useServers`).
- **Catalog file:** applies `withDefaultCatalogPath()` before reading the file (default catalog when no `--catalog`/`--config` or ad-hoc target).
- **Ad-hoc:** `resolveServerConfigs(..., "multi")` for a single inline server (default catalog already applied above); merges launch-time `--header` into `settings`.
**Core hooks:** TUI uses the same managed-state and `useInspectorClient` patterns as web (`ManagedToolsState`, etc.) inside Ink components.
**Behavior note:** v2 core only auto-refreshes tool/resource lists when `autoRefreshOnListChanged` is true in server settings; v1.5 TUI always auto-refreshed. Port accepts v2 core behavior; UI indicator parity with web is open ([#1402](https://github.com/modelcontextprotocol/inspector/issues/1402)).
**Tests:** 2 cases in `clients/tui/__tests__/tui.test.ts` assert `tabsConfig` shape only. Config parsing, OAuth, and Ink component flows are untested.
**Dev:** `npm run dev` in `clients/tui` uses `vite-node` + `dev.ts` for fast iteration without a full tsup rebuild.
---
## Web runner (`runWeb`)
The launcher `--web` path does not start Vite via the `npm run dev` script. It calls `runWeb(argv)` from `clients/web/server/run-web.ts`, which:
1. Parses the same server-selection flags as CLI/TUI (`--config`, `--server`, positional target, `-e`, `--cwd`, `--transport`, `--server-url`, `--header`, `--dev`).
2. Resolves one `MCPServerConfig` via `resolveServerConfigs()` (explicit options only — no default catalog injection) when server input is present.
3. Passes `initialMcpConfig` into `buildWebServerConfig()` → `GET /api/config` legacy channel.
4. Starts `startViteDevServer()` (`--dev`) or `startHonoServer()` (prod).
**Gap — launch-time headers:** `--header` is accepted but logs a warning; headers are not applied to the web session. The v2 web UI configures HTTP headers via per-server settings in the catalog (`ServerSettingsForm`), not via `initialMcpConfig`. Extending `InitialConfigPayload` (or equivalent) is required for launcher-passed headers to reach the UI.
Day-to-day web development still uses `cd clients/web && npm run dev` (Vite CLI + env-based `buildWebServerConfigFromEnv`).
---
## Server configuration flags (shared)
CLI, TUI, and `runWeb` share launch-time server options documented in v1.5's `docs/mcp-server-configuration.md` (port to v2 pending). Common flags:
| Flag | Purpose |
| ------------------------ | -------------------------------------------------------------------------------- |
| `--config <path>` | MCP servers JSON file |
| `--server <name>` | Named entry within config (required when file has multiple servers) |
| `[target...]` | Ad-hoc stdio command/args or HTTP URL |
| `-e KEY=VALUE` | Stdio environment overrides |
| `--cwd <path>` | Stdio working directory |
| `--transport` | `stdio`, `sse`, or `http` |
| `--server-url <url>` | HTTP/SSE endpoint |
| `--header "Name: Value"` | Launch-time HTTP headers (CLI/TUI ad-hoc → `serverSettings`; web → warning only) |
CLI additionally requires `--method` and method-specific args (`--tool-name`, `--uri`, etc.). Semantics for catalog vs session config, `--catalog`, and `servers/import` are in [v2_catalog_launch_config.md](v2_catalog_launch_config.md).
---
## Testing and CI
| Suite | Location | Count | How |
| -------- | ------------------------ | ----- | ---------------------------------------- |
| CLI | `clients/cli/__tests__/` | 86 | Subprocess `node build/index.js` |
| TUI | `clients/tui/__tests__/` | 2 | In-process imports |
| Launcher | CI smoke only | 3 | `--help`, `--cli --help`, `--tui --help` |
| Core | `clients/web/src/test/` | — | Canonical; not duplicated in cli/tui |
**Root orchestration:** `npm run test` chains web unit tests, then `clients/cli` and `clients/tui` tests. `npm run test:coverage` runs web coverage only.
**CI** (`.github/workflows/main.yml`): after web validate + coverage, installs and tests CLI and TUI, builds TUI + launcher, runs launcher help smoke.
**Prerequisites:** CLI `pretest` runs `test-servers:build` (`tsc -p test-servers`) so stdio integration tests can spawn `test-servers/build/test-server-stdio.js`.
---
## Developer setup
v2 does not use npm workspaces. Each client maintains its own `package.json` and `node_modules`:
```bash
npm install # root (launcher runtime deps)
cd clients/web && npm install # required for web dev/test
cd clients/cli && npm install # required for CLI build/test
cd clients/tui && npm install # required for TUI build/test
cd clients/launcher && npm install
```
**Build order for a full local smoke:**
```bash
npm run build:web # includes build:runner for runWeb
npm run build:cli
npm run build:tui
npm run build:launcher
```
Root `npm run validate` currently runs web validate only; extending it to CLI/TUI build + test is follow-up work.
---
## Known gaps
| Area | Gap | Tracking |
| ---------------------------- | -------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------ |
| **CLI config-file settings** | Disk `headers` / timeouts / OAuth not lifted when `--config` is used | This spec; [catalog doc](v2_catalog_launch_config.md) G1 |
| **Web launch headers** | `runWeb --header` warns but does not apply settings to UI | This spec; [#1246](https://github.com/modelcontextprotocol/inspector/issues/1246) |
| **TUI list-changed UX** | No stale-list indicator when `autoRefreshOnListChanged` is false | [#1402](https://github.com/modelcontextprotocol/inspector/issues/1402) |
| **TUI test coverage** | Only `tabsConfig` shape tested | Expand `tui-servers.ts`, Ink flows |
| **CLI coverage gates** | Subprocess tests don't instrument `src/cli.ts` | In-process `runCli()` runner + thin binary E2E suite |
| **Prod web via launcher** | Requires pre-built `clients/web/dist/` | Document or wire into launcher prebuild |
| **Import / `--catalog`** | `servers/import`, `--catalog` not implemented | [catalog doc](v2_catalog_launch_config.md); [#1348](https://github.com/modelcontextprotocol/inspector/issues/1348) |
| **README refresh** | Client READMEs may lag v2 install/build commands | `clients/*/README.md`, `AGENTS.md` |
| **Root validate** | Does not build/test CLI/TUI | Root `package.json` |
---
## References
- `clients/launcher/src/index.ts` — mode routing and dynamic import
- `clients/cli/src/cli.ts` — CLI parsing, `InspectorClient` one-shot flow
- `clients/tui/src/tui-servers.ts` — thin re-export of the shared `loadServerEntries`
- `clients/web/server/run-web.ts` — launcher web entry
- `core/mcp/node/servers.ts` — `loadServerEntries`, `selectServerEntry` (shared CLI/TUI config + settings load path)
- `core/mcp/node/config.ts` — `resolveServerConfigs`, `withDefaultCatalogPath`, `serverSourceConflict`
- `core/mcp/serverList.ts` — `mcpConfigToServerEntries`
- `vitest.shared.mts` — shared Vitest aliases
- [v2_catalog_launch_config.md](v2_catalog_launch_config.md) — catalog and launch config, import, UC1–UC5
- [v2_servers_file.md](v2_servers_file.md) — catalog file format and web CRUD
- [#1246](https://github.com/modelcontextprotocol/inspector/issues/1246) — port tracking issue
- v1.5 sources: `v1.5/main/clients/{cli,tui,launcher}`