This page shows the raw state of the evidence for the Python 0.3.2 isolation candidate. It updates as tests run. Nothing here is a certification.
Containment is not yet proven. The nine Linux Docker containment tests passed on a real Ubuntu host, but we ran them ourselves. That is builder evidence, not an independent review, which is still pending.
1. Test results (recorded 22 October 2026)
82
Discovered
73
Passed
9
Skipped (Docker-gated)
0
Failed
Historical builder-run local results, not independent. Version 0.3.2 corrects the two qualification-suite defects found in 0.3.1. The completed Linux builder run is recorded below; independent reproduction and qualification of the intended deployment remain required.
Approved-domain and IP transport checksLocal functional evidenceFourteen deterministic tests cover domain/IP refusal, private or mixed DNS answers, CIDRs, TLS socket pinning, redirects, proxies and fixed-provider restrictions. Responses are synthetic; these tests do not prove real-container containment.
Linux worker boundary9/9 Docker probes passed on a builder-run Linux host (10 Nov 2026)The nine Docker-gated probes, skipped locally, passed on a disposable Ubuntu 24.04.5 host (Docker 29.1.3, cgroup v2, runc with AppArmor and seccomp) using test pack v2. Results archive SHA-256 6d21d0661f605a8cb1a65cf3d6ca4cfdcc45bb5cc1d83579819236f13f2b3294. Builder-executed evidence only: not independently reviewed, not production qualification.
Independent findings and retestNo external report receivedNo signed independent conclusion or remediation retest is available. Current observations are builder evidence, not reviewer findings or certification.
Provider budgets and permitted-channel leakageDeployment qualification pendingNo billable model run or provider-usage reconciliation was performed. Approved providers can receive permitted message content; network approval does not establish data-loss prevention.
2. Linux Docker host details
Linux distribution / runtimeUbuntu 24.04.5 · Docker 29.1.3 · cgroup v2
Run started (UTC)2026-11-10T09:53:08Z
Result9 passed · 0 failed · 0 skipped
Who ran itEncased builder, on a disposable DigitalOcean server (deleted after the run)
These fields come from the unedited log of that run. A reviewer should repeat the run on their own host rather than rely on ours.
3. Live check: verify the package yourself
Your browser downloads the published 0.3.2 package and calculates its SHA-256 fingerprint, then compares it with the fingerprint we publish. This proves you get the exact package under review. It does not test containment.
Approved reviewers sign in with their invited email and record the package version and fingerprint tested, their own log reference, findings and conclusion. Each submission is dated and cannot be edited afterwards. Only the review status appears publicly; findings stay private to the reviewer and Encased admins.
Run the test pack on your own Linux Docker host.
Sign in at the reviewer page with your approved email.
Enter results, findings and your conclusion, then submit.
Direct link: https://encased.ai/security-audit/reviewer-sign-off
Access is limited to the approved reviewer account; this link does not grant access by itself.
4. Run the reference Docker tests on your VM
These steps run the reference deployment, not a Smokescreen or gVisor pairing. You need a Linux Docker VM, a registry you control and actual image manifest digests. If reviewed images are not available, stop and ask your reviewer: a self-selected image is not independently approved.
1. Install Docker on a fresh Ubuntu 24.04 VM
Use a disposable VM with no customer data or credentials. Docker group access is root-equivalent. After this block, log out of SSH and reconnect before continuing. These commands require Bash.
This block asks for a reviewed base reference and your registry image tag. Obtain the reviewed base manifest digest from Lucas or your technician. A tag is not a digest. Authenticate to your registry outside the logs using its approved login method; never paste passwords into these commands. Building/pushing a worker does not make it independently reviewed.
bash <<'ENCASED_PREP'
set -euo pipefail
umask 077
ROOT="$HOME/encased-run-$(date -u +%Y%m%dT%H%M%SZ)"
mkdir -p "$ROOT/evidence"
printf '%s\n' "$ROOT" > "$HOME/.encased-run-path"
cd "$ROOT"
docker info >/dev/null
curl --fail --location --show-error https://encased.ai/sdk/encased-0.3.2.tgz -o encased-0.3.2.tgz
printf '%s %s\n' 'fc0461a37ef92309be9e18ffb4f26a7332e415db0de0db490d6454a069c8be3f' 'encased-0.3.2.tgz' | sha256sum -c -
sha256sum encased-0.3.2.tgz > evidence/package.sha256
tar -xzf encased-0.3.2.tgz
# Keep the virtual environment outside the Docker build context.
python3 -m venv "$ROOT/venv"
read -r -p 'Reviewed base repository@sha256 digest: ' BASE </dev/tty
if [[ ! "$BASE" =~ ^[^[:space:]]+@sha256:[a-f0-9]{64}$ ]]; then
echo 'STOP: a complete reviewed base manifest reference is required.'; exit 2
fi
read -r -p 'Worker image tag in a registry you control: ' TAG </dev/tty
if [[ -z "$TAG" || "$TAG" =~ [[:space:]] || "$TAG" != */* || "$TAG" == *@* ]]; then
echo 'STOP: enter a repository/tag in your registry, not a digest.'; exit 2
fi
cd "$ROOT/encased-0.3.2"
docker pull "$BASE" 2>&1 | tee "$ROOT/evidence/base-pull.log"
docker build --progress=plain --build-arg "BASE=$BASE" \
-f deployment/isolation-worker.Dockerfile -t "$TAG" . \
2>&1 | tee "$ROOT/evidence/worker-build.log"
docker push "$TAG" 2>&1 | tee "$ROOT/evidence/worker-push.log"
echo 'Enter the worker repository@sha256 manifest reference reported by your registry.'
echo 'Do not enter the local image ID. Ask your reviewer to review this artifact.'
read -r -p 'Worker repository@sha256 digest: ' WORKER </dev/tty
if [[ ! "$WORKER" =~ ^[^[:space:]]+@sha256:[a-f0-9]{64}$ ]]; then
echo 'STOP: a complete worker manifest reference is required.'; exit 2
fi
docker pull "$WORKER" 2>&1 | tee "$ROOT/evidence/worker-pull.log"
printf '%s\n' "$BASE" > "$ROOT/base-reference.txt"
printf '%s\n' "$WORKER" > "$ROOT/worker-reference.txt"
printf 'Prepared: %s\n' "$ROOT"
ENCASED_PREP
3. Capture host details, run the tests and seal the log
Run only after step 2 succeeds. Enter who ran it and whether the images have been independently reviewed. The wrapper saves the actual test and log-writing exit codes, including failures. The shipped script rebuilds the worker and stops if its image ID differs from your pinned registry image; do not bypass that check. Its internal pull/build output is suppressed, so the earlier preparation logs are retained separately.
bash <<'ENCASED_RUN'
set -euo pipefail
umask 077
ROOT="$(cat "$HOME/.encased-run-path")"
BASE="$(cat "$ROOT/base-reference.txt")"
WORKER="$(cat "$ROOT/worker-reference.txt")"
. "$ROOT/venv/bin/activate"
read -r -p 'Who is running this test? ' OPERATOR </dev/tty
read -r -p 'Image review status / reviewer reference (or Not reviewed): ' REVIEW </dev/tty
{
date -u --iso-8601=seconds
printf 'Operator: %s\nImage review status: %s\n' "$OPERATOR" "$REVIEW"
cat /etc/os-release
uname -srm
lscpu
free -h
python3 --version
python3 -m pip --version
docker version --format '{{json .}}'
docker info --format 'OS={{.OperatingSystem}} Kernel={{.KernelVersion}} Architecture={{.Architecture}} CPUs={{.NCPU}} Memory={{.MemTotal}} CgroupDriver={{.CgroupDriver}} CgroupVersion={{.CgroupVersion}} SecurityOptions={{json .SecurityOptions}}'
printf 'Base: %s\nWorker: %s\n' "$BASE" "$WORKER"
docker image inspect "$BASE" "$WORKER" \
--format 'ID={{.Id}} RepoDigests={{json .RepoDigests}} OS={{.Os}} Architecture={{.Architecture}}'
} > "$ROOT/evidence/host-details.txt" 2>&1
cd "$ROOT/encased-0.3.2"
set +e
ENCASED_DOCKER_TEST_IMAGE="$WORKER" \
bash deployment/run-linux-isolation-tests.sh "$BASE" \
2>&1 | tee "$ROOT/evidence/qualification.log"
STATUS=("${PIPESTATUS[@]}")
set -e
printf 'test_exit_code=%s\nlog_capture_exit_code=%s\n' \
"${STATUS[0]}" "${STATUS[1]}" > "$ROOT/evidence/exit-status.txt"
printf '%s\n' "$BASE" > "$ROOT/evidence/base-reference.txt"
printf '%s\n' "$WORKER" > "$ROOT/evidence/worker-reference.txt"
cd "$ROOT/evidence"
sha256sum ./*.log ./*.txt ./*.sha256 > SHA256SUMS
cd "$ROOT"
tar -czf encased-run-results.tgz evidence
sha256sum encased-run-results.tgz | tee encased-run-results.tgz.sha256
printf 'Archive: %s/encased-run-results.tgz\n' "$ROOT"
cat evidence/exit-status.txt
if [[ "${STATUS[0]}" != 0 || "${STATUS[1]}" != 0 ]]; then
echo 'FAILED or incomplete capture: retain the evidence; do not claim containment.'
exit 1
fi
echo 'Supplied probes passed on this setup; independent review is still required.'
ENCASED_RUN
Send the evidence: copy the printed archive from your VM to your computer, upload encased-run-results.tgz here in the Encased builder conversation and paste its final SHA-256 line. Keep the original logs unchanged and private. Review for unexpected sensitive content before sharing; if redaction is needed, retain the original privately and identify a separate redacted copy. Do not include API keys, registry login files or customer data.
The website does not run tests on your VM or automatically email evidence. Results and host details are published only after the actual run is checked; passing this suite alone is not independent review or certification.
Using your own private container registry?
Follow the complete registry workflow for scoped login, building and pushing the worker, pulling by manifest digest, recording VM evidence and a separate synthetic agent trial.
Keep the tools your team already uses for outbound traffic or workload isolation. Encased adds registered-action policies and human review; those tools address different parts of the boundary.
Proposed setups, not verified integrations. Neither pairing has been qualified with Encased 0.3.2. Familiarity with a third-party tool does not establish containment for the combined system. No partnership or endorsement is implied.
Encased: may it run?
The host broker freezes a registered action, obtains a workspace decision and dispatches only permitted effects.
Smokescreen: where can it connect?
An HTTP CONNECT proxy constrains outbound destinations. Separate infrastructure rules must prevent direct bypass.
gVisor: what isolates the code?
An application-kernel sandbox reduces worker exposure to the host. It does not provide business-action approval.
Baseline first: run the reference Docker steps above on a reviewed disposable host and retain the package fingerprint, image manifest digests and untouched logs. Record each modified pairing as a separate configuration; the reference test results do not qualify it.
Define destinations. Inventory registered effects, fixed HTTPS hostnames and explicit approved global IPs/CIDRs. Document Encased decision traffic as a separate trusted control path.
Restrict the proxy. Deploy a pinned Smokescreen release in a trusted segment; configure destination ACLs and limit its callers. Never give the worker internet access or a general proxy tunnel.
Review the transport change. A qualified engineer must implement proxy-aware dispatch. The supplied Encased transports bypass ambient proxies: setting HTTPS_PROXY alone does not connect them to Smokescreen. Preserve exact hostname/address approval, CONNECT destination binding, TLS identity and redirect refusal.
Close direct routes. Enforce infrastructure rules so broker effects cannot bypass the proxy. Account separately for DNS and control traffic. Proxy ACLs alone cannot block direct internet access.
Test failure and bypass. Exercise approved and refused destinations, private/mixed DNS answers, direct-route attempts and proxy outage. Held or denied actions must not dispatch; failed checks must fail closed.
Retain for review: proxy version, ACLs, infrastructure rules, connection observations and correlated broker proposal IDs. Keep original evidence private; identify redacted sharing copies separately.
Choose the layer your team needs; stacking both is not required. Neither detects prompt injection reliably or makes every allowed action safe. Obtain version-specific independent findings, remediate and retest before a production release decision. Third-party operating costs are separate.