fafc8367b8Copilot review round 2 on #2100, plus a collision the review did not find. The default app-origin port was 6276 — which is already RUNNER_OAUTH_CALLBACK_DEFAULT_PORT, the CLI/TUI loopback OAuth callback. That port is fixed precisely so an OAuth app can pre-register http://127.0.0.1:6276/oauth/callback, which makes it the one listener here that cannot move when something takes it first. A running --web would hold it and break a later --cli/--tui OAuth flow at its registered redirect URI, and docs/mcp-app-review.md has those two running side by side. This listener has a documented dynamic fallback, so this listener yields: the default is now 6278 (6277 skipped — v1's retired proxy port, which the migration guide tells people to stop forwarding). A test asserts the two defaults stay distinct. Review comments: - POST /api/app-document now applies hono's bodyLimit BEFORE parsing. The 8MiB check ran after c.req.json() had buffered and parsed the whole request, and bounded only `html` — so bulk in `csp`, or in keys the route never reads, was fully materialized before the 413. - `csp` is validated as an HTTP header value and length-bounded. The controller hands it to res.writeHead() verbatim, where Node throws ERR_INVALID_CHAR synchronously — inside the handler for a later, unrelated request, outside this route's error handling. - The app-origin port now propagates everywhere the other two do: the CLI --print-handoff port-forward command (and its test), the Dockerfile's ENV and EXPOSE, and the remote-review SSH recipe in docs/mcp-app-review.md. Without it an app declaring _meta.ui.domain is unreachable through every documented remote workflow. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0177mPHAdECD18nLLCTwR5rh Signed-off-by: cliffhall <cliff@futurescale.com>
70 lines
3.6 KiB
Docker
70 lines
3.6 KiB
Docker
# syntax=docker/dockerfile:1
|
|
|
|
# ---------------------------------------------------------------------------
|
|
# Build stage — install everything, build all clients, and pack the exact
|
|
# tarball npm would publish. `npm install` runs the root postinstall cascade
|
|
# (per-client installs, since v2 is not a workspace); `npm pack` runs `prepack`
|
|
# (`npm run build`) and emits `modelcontextprotocol-inspector-<version>.tgz`.
|
|
# ---------------------------------------------------------------------------
|
|
FROM node:22-slim AS builder
|
|
WORKDIR /build
|
|
COPY . .
|
|
RUN npm install && npm pack
|
|
|
|
# ---------------------------------------------------------------------------
|
|
# Runtime stage — install just the packed tarball (production deps only),
|
|
# yielding a clean global `mcp-inspector` bin: the same artifact a user gets
|
|
# from `npm i -g @modelcontextprotocol/inspector`. The package's postinstall
|
|
# early-exits under node_modules, and the tarball ships only each client's
|
|
# build/ output, so there is nothing to build here.
|
|
# ---------------------------------------------------------------------------
|
|
FROM node:22-slim AS runner
|
|
ENV NODE_ENV=production
|
|
|
|
COPY --from=builder /build/modelcontextprotocol-inspector-*.tgz /tmp/inspector.tgz
|
|
RUN npm install -g /tmp/inspector.tgz && rm /tmp/inspector.tgz
|
|
|
|
# Serve on all interfaces so the UI is reachable from outside the container, and
|
|
# never try to open a browser from inside it. Binding 0.0.0.0 is refused by
|
|
# default (it exposes the process-spawning backend to the network); a container
|
|
# is the sanctioned exception, so opt in explicitly via DANGEROUSLY_BIND_ALL_INTERFACES.
|
|
ENV HOST=0.0.0.0 \
|
|
DANGEROUSLY_BIND_ALL_INTERFACES=true \
|
|
CLIENT_PORT=6274 \
|
|
MCP_SANDBOX_PORT=6275 \
|
|
MCP_APP_ORIGIN_PORT=6278 \
|
|
MCP_AUTO_OPEN_ENABLED=false
|
|
# 6275 is the MCP Apps sandbox and 6278 the dedicated app origin (#2056) — two
|
|
# further listeners the browser reaches DIRECTLY, so neither works through the
|
|
# 6274 publish alone. Both are only needed for the Apps tab (6278 only for an
|
|
# App declaring `_meta.ui.domain`), so they are EXPOSEd but publishing them is
|
|
# optional — see the Docker section of the root README.
|
|
EXPOSE 6274 6275 6278
|
|
|
|
# Run as the non-root `node` user the base image ships. The inspector resolves
|
|
# its runtime-state dir (default catalog, OAuth token storage) from `HOME`
|
|
# (core/storage/store-io.ts), so set it explicitly to the node user's writable
|
|
# home and work from there.
|
|
ENV HOME=/home/node
|
|
|
|
# Create the runtime-state dir up front, owned by `node`. Docker seeds a named
|
|
# volume's ownership from the image's directory at the mount point — and when
|
|
# that directory does not exist it creates it `root:root`, which the non-root
|
|
# `node` user then cannot write. Without this, `-v inspector-data:/home/node/.mcp-inspector`
|
|
# (the recipe for keeping your saved servers across `docker run --rm`) makes
|
|
# every "add server" fail with `EACCES ... open '/home/node/.mcp-inspector/mcp.json.tmp-*'`.
|
|
RUN mkdir -p /home/node/.mcp-inspector && chown -R node:node /home/node/.mcp-inspector
|
|
|
|
USER node
|
|
WORKDIR /home/node
|
|
|
|
# Report readiness by probing the served SPA (`/` needs no auth). Uses Node's
|
|
# global fetch — no curl/wget in the slim image. Assumes the default `--web`
|
|
# mode; running `--cli`/`--tui` has no web server, so add `--no-healthcheck` to
|
|
# `docker run` for those.
|
|
HEALTHCHECK --interval=30s --timeout=5s --start-period=20s --retries=3 \
|
|
CMD node -e "fetch('http://127.0.0.1:'+(process.env.CLIENT_PORT||6274)+'/').then(r=>process.exit(r.ok?0:1)).catch(()=>process.exit(1))"
|
|
|
|
# Default to the web UI; override the args to run --cli / --tui.
|
|
ENTRYPOINT ["mcp-inspector"]
|
|
CMD ["--web"]
|