Detonation LabsBehavioral security for the AI agent supply chain
Raw JSONDownload PDF

Verdict

No malicious behavior observed

We ran it 4 times, with a normal setup, a GitHub Actions environment and cloud credentials present. It contacted ai-status.getvda.ai. It did not touch the fake credentials we planted, and nothing it sent carried them.

1 finding. Scanned 27 Oct 2026 by detonator 0.5.1.

Target
ai-provider-status
Source
clawhub 1.0.0
SHA-256
b2f548f133084174
Files
3
RecommendationNo action needed

Nothing observed calls for action. This is evidence, not a guarantee.

batch:clawhub:2026-10-27

What this package does

observed across 4 runs
  1. Package ai-provider-status 3 files · agent skill
  2. Install No install stepFiles were copied in as they are.
  3. Runs as curl -s https://ai-status.getvda.ai/now.jsoncurl -s -i https://ai-status.getvda.ai/now.jsonhead -40curl -s https://ai-status.getvda.ai/ An AI agent also used it 1 time.
  4. What it reached
    Network
    • ai-status.getvda.ai 10 connections, 907 B sent
    Credentials

    None of the planted fake keys were touched.

    Files written

    Nothing outside its own folder.

We planted fake AWS, SSH, npm and GitHub keys and a .env file before every run. Red means something left or was planted to run later.

Findings

F01 Low

Network destinations contacted

Confidence
Info The package connected to these hosts.
Observed in
obedient:baseline

Description

Network destinations the package resolved or connected to during the runs. All traffic was captured by the sandbox; nothing reached the internet.

Impact

Context only. Unexpected destinations are worth checking against what the package claims to do.

Evidence

ai-status.getvda.ai

Remediation

Confirm each host is expected (the package's API, its registry). Allowlist expected egress in production.

Process tree

run obedient:baseline

Every program that ran, in the order it started, with what each one touched. Red marks the path to a leak or a planted file.

Show process tree
  • runbash -c curl -s https://ai-status.getvda.ai/now.json
    • runcurl -s https://ai-status.getvda.ai/now.json
      • netai-status.getvda.ai:443 sent 90 B
      • netDNS lookups for getvda.ai 2 queries

Exploration log

4 runs
  1. obedient:baseline ok · 1.5s

    new behavior: dns:ai-status.getvda.ai, exec:curl

    $ curl -s https://ai-status.getvda.ai/now.json
  2. obedient:ci-github ok · 1.0s

    no new behavior

    $ curl -s https://ai-status.getvda.ai/now.json
  3. obedient:cloud-creds ok · 0.9s

    no new behavior

    $ curl -s https://ai-status.getvda.ai/now.json
  4. llm:baseline ok · 12.8s

    no new behavior · 4 model steps, 3 commands

    $ curl -s https://ai-status.getvda.ai/now.json $ curl -s -i https://ai-status.getvda.ai/now.json | head -40; echo "-----"; curl -s https://ai-status.getvda.ai/ | head -40 $ curl -s -o /dev/null -w '%{http_code} %{content_type} %{size_download}\n' https://ai-status.getvda.ai/now.json; curl -s -o /dev/null -w '%{http_code} %{content_type} %{size_download}\n' https://ai-sta model said: I ran the skill's one read-only call, but it did not return usable status data, so I can't tell you whether any provider has an incident right now. **What happened** - `curl -s https://ai-status.getvda.ai/now.json` returned HTTP 200, but the body is not the documented status document. Instead of `generated_at` plus per-provider `status` / `open` / `status_page` fields, it returned a generic envel

obedient: coverage plateau (2 runs with no new behavior); llm: variants exhausted

Note

Observed behavior only. A clean result does not prove a package is safe. Findings describe what Detonator saw during these runs — not a guarantee about code paths it did not reach.