Haiku 5.5
Native agent · Hosted MCP connection/tools-list passedChecked operation Remote hosted MCP connection · initialize + tools/list · http://localhost:3000/mcp
Tested against the hosted endpoint only. Local install is untested.
Agent-reported scope: Scaffold an mcp-use MCP server with the README quickstart, run the dev server, and call its fetch-weather tool over MCP HTTP
MCP server returned: weather-app · version 1.0.0.
MCP tools returned: fetch-weather. Tool listing does not establish authenticated tool-call success.
Run details
- run
- mcp-use--mcp-use-claude-haiku-5-5-1792767269192623503
- date
- Oct 23, 2026
- time to working
- 827.4s
- product hosting / execution
- Remote hosted MCP endpoint
- recorded operation scope
- remote hosted MCP initialize + tools/list
- model
- claude-haiku-5-5
- agent / version
- claude-code / 2.1.300 (Claude Code)
- platform
- linux
- reported cost
- $0.021549
- Recorded source
- https://github.com/mcp-use/mcp-use
- Backend
- hyperv
- Tool transport
- mcp-assigned-vm
- Requested source pin (not an observed revision)
- {"commit": "f877f789e7b4979e62e5cbe221bfb29ebb86d1bd", "url": "https://github.com/mcp-use/mcp-use"}
- Test version
- readme-bdc1b9b3c0d2673aef87e076d14084a8123ddcf09d520c1ee2e364a37c3f4bb9
- revision
- f877f789e7b4979e62e5cbe221bfb29ebb86d1bd
Safety: Review signals: 1 connection review, 4 encoded blob, 5 hardcoded secret, 26 known vulnerability. 10 static matches retained as context.
Review signals: 1 connection review, 4 encoded blob, 5 hardcoded secret, 26 known vulnerability. 10 static matches retained as context.
Reviewed as expected for a test run: resolver address lookups that send no data.
Repository processes attempted these non-loopback connections; proxies can hide the final hostname. 3 distinct locations. 7 attributed processes.
Source contains encoded blob; review its purpose. 2 distinct locations. 4 findings.
Gitleaks found a possible hardcoded secret; value withheld. 5 distinct locations. 5 findings.
OSV reports a known vulnerability in this dependency version. 10 distinct locations. 21 findings.
10 hardcoded secret matches retained as context; full list in the evidence.
OSV scanned supported lockfiles; unpinned dependencies and unsupported formats are outside this check.
A short run cannot establish that code is safe.
Connections show IP addresses; proxies can hide the final hostname.
Tracing covers dispatched commands and their children, not the whole OS; guest root can interfere with tracing.
Answers recorded for this run
- Start development mode serves the MCP endpoint at localhost:3000/mcp and a tool call returns the requested city's weather: Real input produced the expected result: sunny
Source: Start development mode to serve the MCP endpoint at [`http://localhost:3000/mcp`](http://localhost:3000/mcp). The Inspector is automatically available at [`http://localhost:3000/mcp/inspector`](http://localhost:3000/mcp/inspector):
Input: {"tool": "fetch-weather", "arguments": {"city": "Tokyo"}}
Expected: sunny weather result for Tokyo
Command: npm run dev tools/call {"tool": "fetch-weather", "arguments": {"city": "Tokyo"}}
Observed: sunny\",\"temperature\":\"22°C\"}"}], "structuredContent": {"city": "Tokyo", "conditions": "sunny", "temperature": "22°C"}}
- Create the production build runs and reports the built entry: Real input produced the expected result: built index.ts
Source: Create the production build:
Input: npm run build
Expected: build output names the built index.ts entry
Command: npm run build
Observed: built index.ts → .mcp-use/build/index.js (754ms)
Examples
- Scaffold a new project with the quickstart tool and list its templates: command completed successfully
Source: README.md Start with code (npx -y create-mcp-use-app@latest)
Input: --list-templates
Expected: lists the blank, mcp-apps and mcp-server templates
Command: npx -y create-mcp-use-app@latest --list-templates
Observed: ═╝ [0m [1;37m ██║ ╚═╝ ██║ ╚██████╗ ██║ ╚██████╔╝ ███████║ ███████╗[0m [1;37m ╚═╝ ╚═╝ ╚═════╝ ╚═╝ ╚═════╝ ╚══════╝ ╚══════╝[0m [90m[1m by Manufact[0m[0m [1mAvailable Templates:[0m [36m blank [0m [90ma minimal mcp-use server with no tools registered, ready to add your own[0m [36m mcp-apps [0m [90man mcp-use server with React views for MCP Apps[0m [36m mcp-server [0m [90man mcp-use server with example tools and zod validated input[0m [90m💡 Use with: npx create-mcp-use-app my-project --template <template>[0m
- Repository offline test suite for the server package passes: command completed successfully
Source: libraries/typescript/README.md Development Setup
Input: pnpm test in libraries/typescript/packages/server
Expected: all server tests pass
Command: pnpm test
Observed: sal [22m[39mAn update to Root inside a test was not wrapped in act(...). When testing, code that causes React state updates should be wrapped into act(...): act(() => { /* fire events that update state */ }); /* assert on the output */ This ensures that you're testing the behavior the user would see in the browser. Learn more at https://react.dev/link/wrap-tests-with-act [90mstderr[2m | tests/tanstack-start-vite.test.ts[2m > [22m[2mdecodes embedded assets only on the first GET and keeps response bodies independent [22m[39m[mcp-use] View "card" is registered but no tool binds it.
Local or hosted: local
What the agent did
Scaffold an mcp-use MCP server with the README quickstart, run the dev server, and call its fetch-weather tool over MCP HTTP
cd <guest-path> && pwd && ls -la && id -u && python3 --version && cat /etc/os-release | head -3 && git --version 2>&1; ls /inputs 2>&1; ls <guest-path> 2>&1
cd <guest-path> && cat README.md | head -300
cd <guest-path> && which node npm npx; node -v 2>&1; npm -v 2>&1; ls -la .coldpass; ls libraries examples; grep -n -i "node\|requir\|prereq" README.md | head -40; find . -name package.json -not -path "*/node_modules/*" | head -30
cd <guest-path> && ls; grep -n '"engines"' -A3 package.json; grep -n '"packageManager"\|"node"' package.json; grep -n -i "node\|npm\|pnpm" README.md | head -30; cat package.json | head -60
cd <guest-path> && grep -rn '"node"' package.json packages/server/package.json packages/create-mcp-use-app/package.json packages/client/package.json 2>/dev/null; grep -n -i "node.js\|node 2\|node>=\|>= *2" README.md ../../CONTRIBUTING.md | head; sudo -n true && echo SUDO_OK; apt-cache policy nodejs 2>/dev/null | head -3; curl -sI https://deb.nodesource.com 2>&1 | head -1
cd /tmp && (curl -fsSL https://deb.nodesource.com/setup_22.x | sudo -E bash - && sudo apt-get install -y nodejs) > /tmp/node_install.log 2>&1; echo rc=$?; tail -5 /tmp/node_install.log; node -v; npm -v
cd <guest-path> && git init -q && git remote add origin https://github.com/mcp-use/mcp-use && git fetch --depth 1 origin f877f789e7b4979e62e5cbe221bfb29ebb86d1bd 2>&1 | tail -3; git reset --mixed FETCH_HEAD 2>&1 | tail -2; git log --oneline -1; git status --short | head -5
cd <guest-path> && sudo npm install -g [email protected] > /tmp/pnpm_install.log 2>&1; echo rc=$?; tail -3 /tmp/pnpm_install.log; pnpm -v; export NODE_USE_ENV_PROXY=1; timeout 900 pnpm install --frozen-lockfile > /tmp/pnpm_i.log 2>&1; echo install_rc=$?; tail -15 /tmp/pnpm_i.log
cd <guest-path> && sed -n 645,665p README.md; export NODE_USE_ENV_PROXY=1; timeout 1500 pnpm build > /tmp/pnpm_build.log 2>&1; echo build_rc=$?; tail -12 /tmp/pnpm_build.log
mkdir -p /tmp/mcpdemo && cd /tmp/mcpdemo && export NODE_USE_ENV_PROXY=1 && timeout 300 npx -y create-mcp-use-app@latest --help 2>&1 | head -40; echo rc=$?
cd /tmp/mcpdemo && export NODE_USE_ENV_PROXY=1 && timeout 900 npx -y create-mcp-use-app@latest weather-app --npm --install --no-skills > /tmp/scaffold.log 2>&1; echo rc=$?; tail -20 /tmp/scaffold.log; ls -la /tmp/mcpdemo/weather-app 2>&1 | head -30
cd /tmp/mcpdemo/weather-app && cat index.ts; echo ----; cat package.json; echo ----; export NODE_USE_ENV_PROXY=1; timeout 600 npm run build > /tmp/app_build.log 2>&1; echo build_rc=$?; tail -8 /tmp/app_build.log
cd /tmp/mcpdemo/weather-app && export NODE_USE_ENV_PROXY=1; nohup npm run dev > /tmp/app_dev.log 2>&1 & echo started; sleep 25; tail -20 /tmp/app_dev.log; curl -s -o /dev/null -w "%{http_code}\n" http://localhost:3000/mcp
cd /tmp/mcpdemo/weather-app && export NODE_USE_ENV_PROXY=1; timeout 120 npx mcp-use client connect local http://localhost:3000/mcp 2>&1 | tail -5; echo connect_rc=$?; timeout 120 npx mcp-use client local tools list 2>&1 | tail -15; timeout 120 npx mcp-use client local tools call fetch-weather city=Tokyo 2>&1 | tail -15
pkill -f "mcp-use dev" ; pkill -f "npm run dev"; sleep 3; (curl -s -o /dev/null -w "%{http_code}\n" http://localhost:3000/mcp || true); pgrep -fa "mcp-use|npm run dev" | head; cd <guest-path> && grep -rln -i "cron\|setInterval\|--once\|--now\|schedule" packages/*/src 2>/dev/null | head -20; grep -n "^Start development mode\|^Inspect the same server\|^Create the production build" <guest-path>
ps -eo pid,args | grep -E "mcp-use|vite|tsx|node" | grep -v grep | head; curl -s -o /dev/null -w "%{http_code}\n" http://localhost:3000/mcp; echo done
cd <guest-path> && grep -rln -i "cron\|setInterval\|--once\|--now" packages/*/src 2>/dev/null | head -10; echo ---; cd <guest-path> && grep -n "Start development mode\|Inspect the same server\|Create the production build" README.md; grep -n "npx mcp-use client connect" README.md
cd /tmp/mcpdemo && export NODE_USE_ENV_PROXY=1 && timeout 300 npx -y create-mcp-use-app@latest --list-templates 2>/dev/null | head -20; echo rc=$?; cd <guest-path> && grep -n '"test"' package.json; (timeout 900 pnpm test > /tmp/server_test.log 2>&1; echo test_rc=$? >> /tmp/server_test.log) ; tail -12 /tmp/server_test.log
cd <guest-path> && mkdir -p .coldpass && cat > .coldpass/probe.json <<'EOF'
{"kind":"mcp_http","argv":["npm","run","dev"],"cwd":"/tmp/mcpdemo/weather-app","url":"http://localhost:3000/mcp","tool":"fetch-weather","arguments":{"city":"Tokyo"},"expected_text":"sunny"}
EOF
cat > .coldpass/claims.json <<'EOF'
[
{"kind":"claim","claim":"Start development mode serves the MCP endpoint at localhost:3000/mcp and a tool call returns the requested city's weather","source":"Start development mode to serve the MCP endpoint at [`http://localhost:3000/mcp`](http://localhost:3000/mcp). The Inspector is automatically available at [`http://localhost:3000/mcp/inspector`](http://localhost:3000/mcp/inspector):","input":"fetch-weather city=Tokyo","expected":"sunny weather result for Tokyo","probe_kind":"mcp_http","argv":["npm","run","dev"],"url":"http://localhost:3000/mcp","tool":"fetch-weather","arguments":{"city":"Tokyo"},"expected_text":"sunny","cwd":"/tmp/mcpdemo/weather-app"},
{"kind":"claim","claim":"Create the production build runs and reports the built entry","source":"Create the production build:","input":"npm run build","expected":"build output names the built index.ts entry","probe_kind":"cli","argv":["npm","run","build"],"expected_text":"built index.ts","cwd":"/tmp/mcpdemo/weather-app"}
]
EOF
cat > .coldpass/examples.json <<'EOF'
[
{"kind":"example","claim":"Scaffold a new project with the quickstart tool and list its templates","source":"README.md Start with code (npx -y create-mcp-use-app@latest)","input":"--list-templates","expected":"lists the blank, mcp-apps and mcp-server templates","argv":["npx","-y","create-mcp-use-app@latest","--list-templates"],"cwd":"/tmp/mcpdemo","env":{},"expected_text":"mcp-server"},
{"kind":"example","claim":"Repository offline test suite for the server package passes","source":"libraries/typescript/README.md Development Setup","input":"pnpm test in libraries/typescript/packages/server","expected":"all server tests pass","argv":["pnpm","test"],"cwd":"<guest-path>","env":{},"expected_text":"859 passed"}
]
EOF
cat > .coldpass/cycles.json <<'EOF'
{"detected":false,"reason":"Entry points inspected: packages/cli, packages/server, packages/client and packages/tunnel. Grep for cron, setInterval, --once and --now found only UI or polling timers (views-plugin, useMcp-helpers, popup, ToolResultDisplay, PromptResultDisplay, useChatHistory, manufact-auth, tunnel index). No scheduler or product cycle exists; the product runs on demand via npm run dev (HTTP server) and npx mcp-use client commands.","checks":[]}
EOF
for f in .coldpass/*.json; do python3 -I -m json.tool "$f" >/dev/null && echo "ok $f"; done; ls -la .coldpass
cd <guest-path> && <env-var>=<guest-path> timeout 1500 python3 -I <guest-path> --validate --replay; echo checker_rc=$?
cd <guest-path> && (curl -s -o /dev/null -w "port3000=%{http_code}\n" http://localhost:3000/mcp || true); ps -eo pid,args | grep -E "mcp-use dev|npm run dev" | grep -v grep || echo "no server processes"; cat > .coldpass/report.json <<'EOF'
{"outcome":"working","fault_owner":null,"operation":"Scaffold an mcp-use MCP server with the README quickstart, run the dev server, and call its fetch-weather tool over MCP HTTP","operation_scope":"Linux Ubuntu 24.04 guest, Node 22.23.4, pnpm 11.13.1 for the monorepo, npm for the scaffolded project; local HTTP on localhost:3000 only; no credentials, browser or external services","steps":["git init; git remote add origin https://github.com/mcp-use/mcp-use; git fetch --depth 1 origin f877f789e7b4979e62e5cbe221bfb29ebb86d1bd; git reset --mixed FETCH_HEAD","cd libraries/typescript && pnpm install --frozen-lockfile","pnpm build (libraries/typescript)","npx -y create-mcp-use-app@latest weather-app --npm --install --no-skills (run from /tmp/mcpdemo)","npm run build (in /tmp/mcpdemo/weather-app)","npm run dev (in /tmp/mcpdemo/weather-app)","npx mcp-use client connect local http://localhost:3000/mcp","npx mcp-use client local tools list","npx mcp-use client local tools call fetch-weather city=Tokyo","pnpm test (libraries/typescript/packages/server): 859 passed"],"deviations":["Node was not installed in the guest; installed Node 22.23.4 via NodeSource (curl -fsSL https://deb.nodesource.com/setup_22.x | sudo -E bash -; sudo apt-get install -y nodejs) to meet the >=22.22.2 engine requirement.","pnpm was not installed; installed [email protected] (the packageManager pin) with sudo npm install -g [email protected].","pnpm install printed non-fatal WARN lines 'Failed to create bin' for unbuilt example packages (mcp-use bin missing before build); the install still completed and the build succeeded.","The workspace had no .git; applied the task's pinned git init/fetch/reset steps.","The probe and example commands use the scaffolded app in /tmp/mcpdemo/weather-app (outside the checkout) because the README quickstart uses the published create-mcp-use-app package via npx.","Telemetry/consent prompts were not encountered in the non-interactive runs; no telemetry option was changed."],"stuck_step":null,"main_blocker":null,"plain_failure":null,"suggested_fix":null}
EOF
python3 -I -m json.tool .coldpass/report.json >/dev/null && echo report_ok; <env-var>=<guest-path> timeout 1500 python3 -I <guest-path> --validate --replay > /tmp/final_check.json 2>&1; echo checker_rc=$?; python3 -I -c "import json,sys; d=json.load(open('/tmp/final_check.json')) if False else None" ; grep -o '"valid": [a-z]*\|"status": "[A-Z]*"' /tmp/final_check.json | headrun JSON · checker evidence · agent report + check · transcript excerpt unavailable · error excerpt unavailable