Automated Analysis != Safety. This is a Developer Transparency Tool.
Inspector

Cloudron community app transparency reports

Inspect a Cloudron community app `CloudronVersions.json` feed. Repository analysis is derived from the package metadata in that feed, and the report stays intentionally conservative: it surfaces what the package appears to do, but it does not certify safety.

What this checks
CHECKS EXPLAINED
Checks Explained
Metadata links and missing fields
Reads the Cloudron feed metadata, checks whether important links are valid, and notes when required fields are missing.
Dockerfile command patterns
Scans Docker build steps for risky commands such as destructive deletes, world-writable permissions, or remote scripts piped into a shell.
Runtime shell network calls
Looks through startup shell scripts for commands like curl, wget, nc, or ssh that may contact remote systems during runtime.
Privilege escalation and credential leakage
Flags startup-script patterns that touch privileged system paths, invoke sudo, or combine secrets with outbound commands.
Feed vs repo version drift
Compares the version published in CloudronVersions.json with the version found in the package repository when that repository can be identified reliably.
  • Metadata links and missing fields
  • Dockerfile command patterns
  • Runtime shell network calls
  • Privilege escalation and credential leakage
  • Feed vs repo version drift
Refresh
Example inputs
https://communityapps.appx.uk/cloudron-notediscovery/CloudronVersions.json
https://raw.githubusercontent.com/BrutalBirdie/cloudron-tachyon-app/refs/heads/master/CloudronVersions.json
https://communityapps.appx.uk/cloudron-ocular/CloudronVersions.json
Generated Report

vLLM

High-throughput OpenAI-compatible LLM inference server

Score
57/100
57 reached43 still requires deeper inspection
Automated checks can score at most 70/100. The remaining 30 points require deep inspection.
This tool is for informational purposes only. Use at your own risk.
Safety Flags
1
Fail
2
Unsure
7
Pass
Master Info
packageUrl
Not available
Repository
Not available
Packager
OrcVole
minBox
9.1.0
Version
0.3.0
Tags
aillminferenceopenai
Detailed Checklist
packageUrl link is present and well formed when required (7/7)
pass
No `packageUrl` declared. This is valid for packages targeting minBoxVersion below 10.
Website link is present and well formed (7/7)
pass
https://github.com/OrcVole/vllm-cloudron
Metadata and README links avoid insecure HTTP (7/7)
pass
No HTTP links found.
Metadata and README links avoid URL shorteners (7/7)
pass
No shortener links found.
Cloudron feed version matches the repository manifest (7/7)
pass
Feed and repo both report 0.3.0.
Repository content was fetched for static analysis (7/7)
pass
https://github.com/OrcVole/vllm-cloudron @ 8977ad53bd519ada2d2c71e10c23a28e4816356e
Dockerfile scan did not detect clearly suspicious commands (4/7)
unsure
Dockerfile:16 System package installation found. Review whether all packages are justified.
Runtime shell scripts avoid suspicious outbound network calls (4/7)
unsure
test/secret-scan.sh:227 Network-capable command found in a runtime shell script. test/secret-scan.sh:228 Network-capable command found in a runtime shell script. test/secret-scan.sh:229 Network-capable command found in a runtime shell script. test/secret-scan.sh:232 Network-capable command found in a runtime shell script. test/secret-scan.sh:243 Network-capable command found in a runtime shell script. test/secret-scan.sh:248 Network-capable command found in a runtime shell script. test/smoke.sh:37 Network-capable command found in a runtime shell script. test/smoke.sh:45 Network-capable command found in a runtime shell script. test/smoke.sh:51 Network-capable command found in a runtime shell script. test/smoke.sh:63 Network-capable command found in a runtime shell script. test/smoke.sh:65 Network-capable command found in a runtime shell script. test/smoke.sh:71 Network-capable command found in a runtime shell script.
Runtime shell scripts avoid privilege escalation patterns (0/7)
fail
start.sh:21 Privileged command or direct system-path modification detected in startup script. test/secret-scan.sh:156 Privileged command or direct system-path modification detected in startup script. test/secret-scan.sh:227 Privileged command or direct system-path modification detected in startup script. test/secret-scan.sh:228 Privileged command or direct system-path modification detected in startup script. test/secret-scan.sh:229 Privileged command or direct system-path modification detected in startup script. test/secret-scan.sh:232 Privileged command or direct system-path modification detected in startup script. test/secret-scan.sh:257 Privileged command or direct system-path modification detected in startup script. test/smoke.sh:31 Privileged command or direct system-path modification detected in startup script.
Runtime shell scripts avoid credential leakage patterns (7/7)
pass
No credential leakage patterns found in scanned shell scripts.
Raw Findings
Flagged lines are highlighted directly in the code snippet.
21 findings
Dockerfile:16
low
System package installation found. Review whether all packages are justified.
14ARG VLLM_VERSION
15
16RUN apt-get update && apt-get install -y --no-install-recommends python3.12-venv \
17 && rm -rf /var/lib/apt/lists/*
18
start.sh:21
high
Privileged command or direct system-path modification detected in startup script.
19# filesystem backup never carries multi-GB weights; survives updates; empty after a restore
20# or clone, at which point first boot re-downloads the model.
21MODELS_DIR="/var/lib/vllm"
22HF_DIR="${MODELS_DIR}/hf" # HF_HOME: hub cache, token state, downloaded weights
23VLLM_CACHE="${MODELS_DIR}/cache" # VLLM_CACHE_ROOT: compiled-artifact cache
test/secret-scan.sh:156
high
Privileged command or direct system-path modification detected in startup script.
154 # it does not ship is simply absent. Only then are the container-side hits for these exact paths
155 # dropped, and only for these exact paths, in the same spirit as the pinned SSH keys below.
156 RUNTIME_PATHS=(/etc/hosts /etc/resolv.conf /etc/hostname)
157 RUNFLAGS=(--network=none) # no DNS/hosts wiring either engine can avoid
158 [[ "$CRI" == *podman* ]] && RUNFLAGS+=(--no-hosts) # podman only; docker always injects
test/secret-scan.sh:227
high
Privileged command or direct system-path modification detected in startup script.
225 # hashes whenever the base image digest changes.
226 declare -A PINNED_SSH=(
227 [/etc/ssh/ssh_host_ecdsa_key]=677458f83d985da3fd7cdd208e90e4eac09da5be205425a5f96a6242dc985c33
228 [/etc/ssh/ssh_host_ed25519_key]=0c575ce8d9ba487b05cc473fad4b0650fb950181028e6ac19796f86f56f22a7a
229 [/etc/ssh/ssh_host_rsa_key]=ae0ea8087e90baf138d277ca52b6cf47b5010adc0e5bd84236713eee1b85de85
test/secret-scan.sh:227
medium
Network-capable command found in a runtime shell script.
225 # hashes whenever the base image digest changes.
226 declare -A PINNED_SSH=(
227 [/etc/ssh/ssh_host_ecdsa_key]=677458f83d985da3fd7cdd208e90e4eac09da5be205425a5f96a6242dc985c33
228 [/etc/ssh/ssh_host_ed25519_key]=0c575ce8d9ba487b05cc473fad4b0650fb950181028e6ac19796f86f56f22a7a
229 [/etc/ssh/ssh_host_rsa_key]=ae0ea8087e90baf138d277ca52b6cf47b5010adc0e5bd84236713eee1b85de85
test/secret-scan.sh:228
high
Privileged command or direct system-path modification detected in startup script.
226 declare -A PINNED_SSH=(
227 [/etc/ssh/ssh_host_ecdsa_key]=677458f83d985da3fd7cdd208e90e4eac09da5be205425a5f96a6242dc985c33
228 [/etc/ssh/ssh_host_ed25519_key]=0c575ce8d9ba487b05cc473fad4b0650fb950181028e6ac19796f86f56f22a7a
229 [/etc/ssh/ssh_host_rsa_key]=ae0ea8087e90baf138d277ca52b6cf47b5010adc0e5bd84236713eee1b85de85
230 )
test/secret-scan.sh:228
medium
Network-capable command found in a runtime shell script.
226 declare -A PINNED_SSH=(
227 [/etc/ssh/ssh_host_ecdsa_key]=677458f83d985da3fd7cdd208e90e4eac09da5be205425a5f96a6242dc985c33
228 [/etc/ssh/ssh_host_ed25519_key]=0c575ce8d9ba487b05cc473fad4b0650fb950181028e6ac19796f86f56f22a7a
229 [/etc/ssh/ssh_host_rsa_key]=ae0ea8087e90baf138d277ca52b6cf47b5010adc0e5bd84236713eee1b85de85
230 )
test/secret-scan.sh:229
high
Privileged command or direct system-path modification detected in startup script.
227 [/etc/ssh/ssh_host_ecdsa_key]=677458f83d985da3fd7cdd208e90e4eac09da5be205425a5f96a6242dc985c33
228 [/etc/ssh/ssh_host_ed25519_key]=0c575ce8d9ba487b05cc473fad4b0650fb950181028e6ac19796f86f56f22a7a
229 [/etc/ssh/ssh_host_rsa_key]=ae0ea8087e90baf138d277ca52b6cf47b5010adc0e5bd84236713eee1b85de85
230 )
231 ssh_listing="$("$CRI" run --rm --user 0 "${RUNFLAGS[@]}" --entrypoint /bin/bash "$IMAGE" \
test/secret-scan.sh:229
medium
Network-capable command found in a runtime shell script.
227 [/etc/ssh/ssh_host_ecdsa_key]=677458f83d985da3fd7cdd208e90e4eac09da5be205425a5f96a6242dc985c33
228 [/etc/ssh/ssh_host_ed25519_key]=0c575ce8d9ba487b05cc473fad4b0650fb950181028e6ac19796f86f56f22a7a
229 [/etc/ssh/ssh_host_rsa_key]=ae0ea8087e90baf138d277ca52b6cf47b5010adc0e5bd84236713eee1b85de85
230 )
231 ssh_listing="$("$CRI" run --rm --user 0 "${RUNFLAGS[@]}" --entrypoint /bin/bash "$IMAGE" \
test/secret-scan.sh:232
high
Privileged command or direct system-path modification detected in startup script.
230 )
231 ssh_listing="$("$CRI" run --rm --user 0 "${RUNFLAGS[@]}" --entrypoint /bin/bash "$IMAGE" \
232 -c 'for f in /etc/ssh/ssh_host_*_key; do [ -e "$f" ] && sha256sum "$f"; done' 2>/dev/null)"
233 found=0; pinned_ok=0
234 while IFS= read -r line; do
test/secret-scan.sh:232
medium
Network-capable command found in a runtime shell script.
230 )
231 ssh_listing="$("$CRI" run --rm --user 0 "${RUNFLAGS[@]}" --entrypoint /bin/bash "$IMAGE" \
232 -c 'for f in /etc/ssh/ssh_host_*_key; do [ -e "$f" ] && sha256sum "$f"; done' 2>/dev/null)"
233 found=0; pinned_ok=0
234 while IFS= read -r line; do
test/secret-scan.sh:243
medium
Network-capable command found in a runtime shell script.
241 shp="$(printf '%s\n' "$shp" | grep -vF "$f:" || true)" # drop ONLY this verified exact path
242 else
243 emit ssh-key "$f sha256=$h is NOT a pinned base host key (new, changed or extra: treat as a leak)"
244 fi
245 done <<< "$ssh_listing"
test/secret-scan.sh:248
medium
Network-capable command found in a runtime shell script.
246 echo " host keys: $found found, $pinned_ok pinned-ok, ${#PINNED_SSH[@]} expected"
247 [[ "$found" -eq "${#PINNED_SSH[@]}" && "$pinned_ok" -eq "${#PINNED_SSH[@]}" ]] \
248 || emit ssh-key "host key count mismatch: $found found, $pinned_ok pinned-ok, ${#PINNED_SSH[@]} expected"
249
250 # DISTRO-GENERATED PLACEHOLDER KEYS. Some Debian packages mint a self-signed key in their
test/secret-scan.sh:257
high
Privileged command or direct system-path modification detected in startup script.
255 # Nothing else is exempt: a key at any other path is still a finding, including a second key in
256 # the same directory. Add a path here only after confirming the package does not reference it.
257 PLACEHOLDER_KEYS=(/etc/ssl/private/ssl-cert-snakeoil.key /etc/prosody/certs/localhost.key)
258 for k in "${PLACEHOLDER_KEYS[@]}"; do
259 if printf '%s' "$shp" | grep -qF "$k:"; then
test/smoke.sh:31
high
Privileged command or direct system-path modification detected in startup script.
29echo "==> starting $IMAGE as $NAME on :$PORT"
30$CRI run -d --name "$NAME" -p "127.0.0.1:${PORT}:8000" -v "$VOL":/app/data \
31 -v "${VOL}-models":/var/lib/vllm --memory 8g "$IMAGE" >/dev/null
32
33# 1. Liveness must answer 200 well before the model is ready (ADR 0003). Give nginx 15s to
test/smoke.sh:37
medium
Network-capable command found in a runtime shell script.
35ok=""
36for _ in $(seq 1 15); do
37 code="$(curl -s -o /dev/null -w '%{http_code}' "http://127.0.0.1:${PORT}/health" || true)"
38 [[ "$code" == "200" ]] && { ok=1; break; }
39 sleep 1
test/smoke.sh:45
medium
Network-capable command found in a runtime shell script.
43
44# 2. Landing page is HTML, not blank.
45curl -s "http://127.0.0.1:${PORT}/" | grep -qi "vLLM API server" || fail "landing page missing"
46echo "PASS: landing page served at /"
47
test/smoke.sh:51
medium
Network-capable command found in a runtime shell script.
49echo "==> waiting for /ready (up to ${READY_TIMEOUT}s; first run downloads the model)"
50t=0
51until [[ "$(curl -s -o /dev/null -w '%{http_code}' "http://127.0.0.1:${PORT}/ready" || true)" == "200" ]]; do
52 [[ "$($CRI inspect --format '{{.State.Running}}' "$NAME" 2>/dev/null)" == "true" ]] \
53 || fail "container exited during warmup (vLLM crashed; see logs above)"
test/smoke.sh:63
medium
Network-capable command found in a runtime shell script.
61KEY="$($CRI exec "$NAME" sh -c 'grep ^VLLM_API_KEY /app/data/.secrets/keys.env | cut -d= -f2')"
62[[ -n "$KEY" ]] || fail "could not read generated API key"
63code="$(curl -s -o /dev/null -w '%{http_code}' "http://127.0.0.1:${PORT}/v1/models")"
64[[ "$code" == "401" ]] || fail "/v1/models without key returned $code, expected 401"
65MODEL="$(curl -s -H "Authorization: Bearer $KEY" "http://127.0.0.1:${PORT}/v1/models" | jq -r '.data[0].id')"
test/smoke.sh:65
medium
Network-capable command found in a runtime shell script.
63code="$(curl -s -o /dev/null -w '%{http_code}' "http://127.0.0.1:${PORT}/v1/models")"
64[[ "$code" == "401" ]] || fail "/v1/models without key returned $code, expected 401"
65MODEL="$(curl -s -H "Authorization: Bearer $KEY" "http://127.0.0.1:${PORT}/v1/models" | jq -r '.data[0].id')"
66[[ -n "$MODEL" && "$MODEL" != "null" ]] || fail "/v1/models with key returned no model"
67echo "PASS: auth (401 bare, 200 keyed), served model: $MODEL"
test/smoke.sh:71
medium
Network-capable command found in a runtime shell script.
69# 5. One real streamed completion. Count SSE data lines and check finish.
70echo "==> single streamed completion (the one inference call)"
71RESP="$(curl -sN --max-time 300 "http://127.0.0.1:${PORT}/v1/chat/completions" \
72 -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
73 -d "{\"model\":\"$MODEL\",\"stream\":true,\"max_tokens\":32,
0.0.9