kaelorlabs/smitline

No problems traced to this repository code in these checks. 1 static matches retained as context.

1 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.

Full safety evidence →

• Smitline exposes call tools to MCP-capable agents such as Claude Desktop: Real input produced the expected result: marin

  Source: **Works with your agent.** Claude Code, Codex, Cursor, Claude Desktop, OpenClaw, Hermes, and anything that can use MCP, a CLI or a REST API.

  Input: {"tool": "list_voices", "arguments": {}}

  Expected: the GPT-Live voice list, default marin

  Command: sudo -n docker exec -i smitline smitline mcp tools/call {"tool": "list_voices", "arguments": {}}

  Observed: marin\",\"voices\":[\"marin\",\"vesper\",\"quartz\",\"ripple\",\"willow\",\"stone\",\"gleam\",\"meridian\",\"bossa\",\"tempo\",\"beacon\",\"delta\",\"cinder\"]}"}], "structuredContent": {"default": "marin", "voices": ["marin", "vesper", "quartz", "ripple", "willow", "stone", "gleam", "meridian", "bossa", "tempo", "beacon", "delta", "cinder"]}, "isError": false}

• Calls need your own OpenAI key; without it the brief check reports the missing key: Real input produced the expected result: openai is not configured

  Source: **Runs on your computer.** One Docker container, your own OpenAI and phone keys, open source under Apache-2.0.

  Input: {"tool": "check_call_brief", "arguments": {"channel": "phone", "to": "+14155550142", "objective": "Book a table for 4 at 7pm", "onBehalfOf": "Sam"}}

  Expected: ok false with problem 'openai is not configured'

  Command: sudo -n docker exec -i smitline smitline mcp tools/call {"tool": "check_call_brief", "arguments": {"channel": "phone", "to": "+14155550142", "objective": "Book a table for 4 at 7pm", "onBehalfOf": "Sam"}}

  Observed: openai is not configured: set OPENAI_API_KEY\"],\"carried\":0}"}], "structuredContent": {"ok": false, "brief": {"channel": "phone", "to": "+14155550142", "onBehalfOf": "Sam", "objective": "Book a table for 4 at 7pm", "mayAgreeTo": [], "mustNotShare": [], "maxMinutes": 10, "rehearsal": false}, "problems": ["Smitline calls people between 8:00 AM and 9:00 PM their time. It is 9:10 PM for +14155550142 (America/Los_Angeles). Try again later, or set afterHours to true if the user confirms this person expects a call now.", "openai is not configured: set OPENAI_API_KEY"], "carried": 0}, "isError": fal

Examples

• The smitline CLI lists GPT-Live voices inside the running container: command completed successfully

  Source: README.md Prefer to set it up yourself? / docs/agents.md CLI

  Input: voices

  Expected: JSON with default marin

  Command: sudo -n docker exec smitline smitline voices

  Observed: { "default": "marin", "voices": [ "marin", "vesper", "quartz", "ripple", "willow", "stone", "gleam", "meridian", "bossa", "tempo", "beacon", "delta", "cinder" ] }

Local or hosted: local

Run the Smitline container from the README and call its MCP tools (list_voices works; placing a phone call needs an OpenAI key)

cd <guest-path> && pwd && ls -la && git status 2>&1 | head -3; id -u; python3 --version; cat README.md 2>/dev/null | head -200
cd <guest-path> && which docker node npm uv 2>&1; docker --version 2>&1; ls .coldpass; sed -n 1,80p docs/development.md; echo ----; sed -n 1,60p SETUP.md; echo; ls /inputs 2>&1; cat <guest-path> >/dev/null 2>&1 && echo checker-present
sudo -n true && echo SUDO_OK; cat /etc/os-release | head -2; (sudo -n apt-get update -qq 2>&1 | tail -3); sudo -n apt-get install -y -qq docker.io 2>&1 | tail -5; which docker && docker --version
sudo -n systemctl start docker 2>&1; sleep 3; sudo -n docker info 2>&1 | head -15; echo ---; curl -sS -m 20 -o /dev/null -w '%{http_code}\n' https://ghcr.io/v2/ 2>&1; echo env: $HTTPS_PROXY $https_proxy
sudo -n docker run -d --name smitline --restart unless-stopped --network host -v smitline:/data ghcr.io/kaelorlabs/smitline 2>&1 | tail -5; sleep 15; sudo -n docker ps -a --filter name=smitline 2>&1; sudo -n docker logs --tail 30 smitline 2>&1 | tail -30
cd <guest-path> && sed -n 1,120p docs/agents.md; echo ----; curl -fsS -m 10 -o /dev/null -w '%{http_code}\n' http://127.0.0.1:8095/calls; sudo -n docker exec smitline smitline --help 2>&1 | head -40
mkdir -p <guest-path> && cat > <guest-path> <<'EOF'
import json, subprocess, sys
argv = json.loads(sys.argv[1]); tool = sys.argv[2] if len(sys.argv) > 2 else None
args = json.loads(sys.argv[3]) if len(sys.argv) > 3 else {}
p = subprocess.Popen(argv, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True)
def send(m): p.stdin.write(json.dumps(m)+"\n"); p.stdin.flush()
def recv():
    while True:
        line = p.stdout.readline()
        if not line: raise SystemExit("EOF; stderr=" + p.stderr.read())
        msg = json.loads(line)
        if "id" in msg: return msg
send({"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"h","version":"1"}}})
print(json.dumps(recv().get("result",{}).get("serverInfo")))
send({"jsonrpc":"2.0","method":"notifications/initialized"})
if tool is None:
    send({"jsonrpc":"2.0","id":2,"method":"tools/list"})
    for t in recv()["result"]["tools"]: print(t["name"])
else:
    send({"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":tool,"arguments":args}})
    print(json.dumps(recv(), indent=1)[:3000])
p.stdin.close(); p.wait(timeout=20)
EOF
cd <guest-path> && timeout 120 python3 -I mcpcall.py '["sudo","-n","docker","exec","-i","smitline","smitline","mcp"]' 2>&1 | head -60
cd <guest-path> && timeout 120 python3 -I mcpcall.py '["sudo","-n","docker","exec","-i","smitline","smitline","mcp"]' list_voices '{}' 2>&1 | head -40; echo ====; timeout 120 python3 -I mcpcall.py '["sudo","-n","docker","exec","-i","smitline","smitline","mcp"]' check_call_brief '{"channel":"phone","to":"+14155550142","objective":"Book a table for 4 at 7pm","onBehalfOf":"Sam"}' 2>&1 | head -60
cd <guest-path> && grep -rniE "setInterval|cron|--once|--now|schedule|setTimeout\(.*loop|while \(true\)|for await" --include=*.mjs --include=*.js --include=*.py --include=*.ts --include=*.sh meeting-runtime packages control-panel docker start-*.sh scripts 2>/dev/null | grep -v node_modules | head -30; echo ----; cat package.json; echo; sudo -n docker exec smitline smitline voices 2>&1 | head -5
cd <guest-path> && grep -n "Works with your agent" README.md; grep -n "MCP, a CLI or a REST API" README.md | cut -c1-300; grep -n "Smitline makes the call" README.md | cut -c1-200
cd <guest-path> && mkdir -p .coldpass && cat > .coldpass/probe.json <<'EOF'
{"kind":"mcp","argv":["sudo","-n","docker","exec","-i","smitline","smitline","mcp"],"cwd":"<guest-path>","tool":"list_voices","arguments":{},"expected_text":"marin"}
EOF
cat > .coldpass/claims.json <<'EOF'
[
 {"kind":"claim","claim":"Smitline exposes call tools to MCP-capable agents such as Claude Desktop","source":"**Works with your agent.** Claude Code, Codex, Cursor, Claude Desktop, OpenClaw, Hermes, and anything that can use MCP, a CLI or a REST API.","input":"list_voices with no arguments","expected":"the GPT-Live voice list, default marin","probe_kind":"mcp","argv":["sudo","-n","docker","exec","-i","smitline","smitline","mcp"],"tool":"list_voices","arguments":{},"expected_text":"marin","cwd":"<guest-path>"},
 {"kind":"claim","claim":"Calls need your own OpenAI key; without it the brief check reports the missing key","source":"**Runs on your computer.** One Docker container, your own OpenAI and phone keys, open source under Apache-2.0.","input":"check_call_brief for a phone call with no OPENAI_API_KEY set","expected":"ok false with problem 'openai is not configured'","probe_kind":"mcp","argv":["sudo","-n","docker","exec","-i","smitline","smitline","mcp"],"tool":"check_call_brief","arguments":{"channel":"phone","to":"+14155550142","objective":"Book a table for 4 at 7pm","onBehalfOf":"Sam"},"expected_text":"openai is not configured","cwd":"<guest-path>"}
]
EOF
cat > .coldpass/examples.json <<'EOF'
[
 {"kind":"example","claim":"The smitline CLI lists GPT-Live voices inside the running container","source":"README.md Prefer to set it up yourself? / docs/agents.md CLI","input":"voices","expected":"JSON with default marin","argv":["sudo","-n","docker","exec","smitline","smitline","voices"],"cwd":"<guest-path>","env":{},"expected_text":"marin"}
]
EOF
cat > .coldpass/cycles.json <<'EOF'
{"detected":false,"reason":"Entry points inspected (start-*.sh, Dockerfile, packages/cli, meeting-runtime, control-panel). No cron, --once/--now, or product scheduler loop was found. The only setInterval calls are console UI refresh timers in control-panel/app.js and calls.js. The daemon is a request-driven HTTP/MCP server; calls are placed on demand, not on a schedule.","checks":[]}
EOF
for f in probe claims examples cycles; do python3 -I -m json.tool .coldpass/$f.json >/dev/null && echo "$f ok"; done; <env-var>=<guest-path> python3 -I <guest-path> --validate --replay 2>&1 | tail -40
cd <guest-path> && sudo -n docker stop smitline >/dev/null && sudo -n docker ps -a --filter name=smitline --format '{{.Names}} {{.Status}}'; mkdir -p .coldpass && cat > .coldpass/report.json <<'EOF'
{"outcome":"needs_credentials","fault_owner":"prerequisite","operation":"Run the Smitline container from the README and call its MCP tools (list_voices works; placing a phone call needs an OpenAI key)","operation_scope":"Linux Ubuntu 24.04 guest, Docker Engine 29.1.3 installed via apt (sudo), image ghcr.io/kaelorlabs/smitline pulled from GHCR, phone-only run (no Docker socket). Not tested: placing a real phone call (needs OpenAI key plus SignalWire/Twilio account), meetings, the Node 22 npm test suite, the meeting image build.","steps":["sudo apt-get install -y docker.io (Docker prerequisite from README 'What you need'; Docker was missing, installed 29.1.3)","sudo systemctl start docker","docker run -d --name smitline --restart unless-stopped --network host -v smitline:/data ghcr.io/kaelorlabs/smitline (README 'Prefer to set it up yourself?', phone-only variant)","curl http://127.0.0.1:8095/calls returned 200","smitline mcp over stdio: tools/call list_voices returned default marin and 13 voices","smitline mcp: tools/call check_call_brief returned ok false, problem 'openai is not configured: set OPENAI_API_KEY'","docker exec smitline smitline voices returned the voice list","docker stop smitline"],"deviations":["Docker was not installed in the guest; installed docker.io via apt with sudo (the README requires Docker but the guest lacked it).","Probes use 'sudo -n docker exec' because the ubuntu user is not in the docker group.","Used the README's phone-only docker run to avoid mounting the Docker socket; the meeting option was not exercised.","Node 22 and the documented npm test and meeting-image test suite were not run; they are repo tests, not product operations.","No OpenAI key or phone provider credentials are available in this guest, so no call was placed; the README says keys go on the local setup <email>:8095, which was not completed."],"stuck_step":"Placing a call: the brief check fails with 'openai is not configured: set OPENAI_API_KEY'","main_blocker":"No OpenAI API key (and no SignalWire/Twilio phone account) is available in this disposable guest, and I was not given credentials.","plain_failure":"The Smitline app starts and its agent tools respond, but it cannot place a phone call here because it has no OpenAI key or phone account. Adding those keys on its local setup page is needed to test calls.","suggested_fix":"Provide an OpenAI key with GPT-Live access and a SignalWire or Twilio account, enter them on the setup page at http://127.0.0.1:8095 (not in chat), then rerun the check_call_brief and a real call."}
EOF
python3 -I -m json.tool .coldpass/report.json >/dev/null && echo report ok; <env-var>=<guest-path> python3 -I <guest-path> --validate --replay 2>&1 | python3 -I -c "import json,sys; d=json.load(sys.stdin); print(d['valid'], d['problems'], d['replay']['status'])"
cd <guest-path> && sudo -n docker start smitline >/dev/null && sleep 12 && curl -fsS -m 10 -o /dev/null -w '%{http_code}\n' http://127.0.0.1:8095/calls; <env-var>=<guest-path> python3 -I <guest-path> --validate --replay 2>&1 | python3 -I -c "import json,sys; d=json.load(sys.stdin); print(d['valid'], d['problems'], d['replay']['status'])"
cd <guest-path> && python3 -I - <<'EOF'
import json
p='.coldpass/report.json'
r=json.load(open(p))
r['deviations'].append("The smitline container is intentionally left RUNNING after the run: the probe, claims and examples call 'docker exec smitline smitline mcp/voices', which need it up for the checker replay. Stop it with: sudo docker rm -f smitline (the console on 127.0.0.1:8095 and the stdio MCP are the only processes; they are the product's own runtime, not background processes I spawned separately).")
r['steps'].append("docker stop smitline then docker start smitline were needed because the first stop broke the checker replay; final container state: running")
json.dump(r,open(p,'w'),indent=1)
EOF
python3 -I -m json.tool .coldpass/report.json >/dev/null && echo report ok; <env-var>=<guest-path> python3 -I <guest-path> --validate --replay 2>&1 | python3 -I -c "import json,sys; d=json.load(sys.stdin); print(d['valid'], d['problems'], d['replay']['status'])"; sudo -n docker ps --filter name=smitline --format '{{.Names}} {{.Status}}'; ls -la .coldpass