Detonation LabsBehavioral security for the AI agent supply chain

Verdict

No malicious behavior observed

We ran it 1 time, with a normal setup. It contacted registry.npmjs.org and www.juicer.io. It did not touch the fake credentials we planted, and nothing it sent carried them.

4 findings. Scanned 27 Oct 2026 by detonator 0.5.1.

Target
juicer-social-feed
Source
clawhub 1.0.0
SHA-256
aa608085ca257e80
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 1 run
  1. Package juicer-social-feed 3 files · agent skill
  2. Install No install stepFiles were copied in as they are.
  3. Runs as ls -lals -la skillfind skill -maxdepth 3 -type fhead -50 An AI agent also used it 1 time.
  4. What it reached
    Network
    • registry.npmjs.org 4 connections, 1.5 KB sent
    • www.juicer.io 4 connections, 341 B sent
    Credentials
    • ./.envread

    ~/.npmrc read by its own tool (normal)

    Files written
    • ./dev.htmlwritten

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

Used a planted key with the service it belongs to (registry.npmjs.org)

Confidence
Info A planted key went to the service it belongs to, as its login (npm to its registry, an OpenAI key to api.openai.com).
Observed in
llm:baseline

Description

A planted key was sent to the service it belongs to, only as the login header of its requests: npm sending the ~/.npmrc token to registry.npmjs.org while installing, or an app calling api.openai.com with the OpenAI key from .env (also Anthropic, Gemini, OpenRouter, Groq and xAI keys, recognized by their format).

Impact

Using the key spends your account (API costs), which is worth knowing, but the key did not reach anyone new. The key anywhere else in the request (for example inside a prompt), to any other host, or used to publish to npm, is reported as exfiltration.

Evidence

~/.npmrc (npm) sent only as the login header of requests to registry.npmjs.org

Remediation

No action needed.

F02 Low

Network destinations contacted

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

registry.npmjs.org
www.juicer.io

Remediation

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

F03 Low

Loaded the project's .env file

Confidence
Info The package loaded the project's .env file.
Observed in
llm:baseline

Description

The process loaded the project's .env file. Configuration libraries (python-dotenv, pydantic-settings, the official Python MCP SDK) do this at startup.

Impact

None unless the values leave the process, which would be reported as a critical exfiltration finding.

Evidence

./.env (by cat)

Remediation

No action needed. Keep long-lived secrets out of project .env files where possible.

F04 Low

A tool read its own credentials (e.g. gh, aws, ssh)

Confidence
Info A tool opened its own login or config file, such as npm reading ~/.npmrc.
Observed in
llm:baseline

Description

A tool opened its own configuration or login file, such as npm reading ~/.npmrc or gh reading its hosts file.

Impact

None in itself: this is how those tools work. Listed so the report shows every credential access.

Evidence

~/.npmrc (by npm)

Remediation

No action needed.

Process tree

run llm: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.

  • runbash -c cat skill/_meta.json; echo "---"; cat skill/skill-card.md...
    • runcat .env
      • secretread ./.env
View full process tree
  • runbash -c ls -la; echo "---"; ls -la skill 2>/dev/null; echo "---";...
    • runls -la
    • runls -la skill
    • runfind skill -maxdepth 3 -type f
    • runhead -50
  • runbash -c env | grep -i -E 'juicer|mcp|api' | sed 's/=.*/=<set>/'
    • runenv
    • runsed s/=.*/=<set>/
    • rungrep -i -E juicer|mcp|api
  • runbash -c cat skill/_meta.json; echo "---"; cat skill/skill-card.md...
    • runcat .env
      • secretread ./.env
    • runcat skill/_meta.json
    • runcat skill/skill-card.md
    • runcat README.md
    • runcat package.json
    • runsed s/=.*/=<redacted>/
    • runls -la src
    • runhead -60 src/index.js
  • runbash -c which juicer juicer-mcp npx node 2>/dev/null; echo "---";...
    • runwhich juicer juicer-mcp npx node
    • runls ~/.config
    • runls ~/.claude* ~/.openclaw
    • runhead
    • rungrep -ril juicer ~ --include=*.json --include=*.toml
  • runbash -c cd /tmp && timeout 60 npm view juicer-mcp version 2>&1 | ...
    • runtimeout 60 npm view juicer-mcp version
      • runnpm view juicer-mcp version
        • runnode /usr/bin/npm view juicer-mcp version
          • secretread ~/.npmrc (its own config)
          • netregistry.npmjs.org:443 sent 433 B
          • netregistry.npmjs.org:443 sent 391 B
    • runtail -3
    • runtimeout 30 curl -s -o /dev/null
      • runcurl -s -o /dev/null -w %{http_code}\n
        • netwww.juicer.io:443 sent 85 B
  • runbash -c cd /tmp && timeout 60 npm view juicer-mcp 2>&1 | head -20...
    • runhead -20
    • runtimeout 60 npm view juicer-mcp
      • runnpm view juicer-mcp
        • runnode /usr/bin/npm view juicer-mcp
          • secretread ~/.npmrc (its own config)
          • netregistry.npmjs.org:443 sent 391 B
    • runhead -40
    • runtimeout 60 npm search juicer --json
      • runnpm search juicer --json
        • runnode /usr/bin/npm search juicer --json
          • secretread ~/.npmrc (its own config)
          • netregistry.npmjs.org:443 sent 365 B
  • runbash -c cd /tmp && timeout 30 curl -s https://www.juicer.io/devel...
    • runtimeout 30 curl -s https://www.juicer.io/developer
      • runcurl -s https://www.juicer.io/developer
        • netwww.juicer.io:443 sent 85 B
    • runsed s/<[^>]*>/ /g
    • runhead -30
    • runtr -s \n \n
    • rungrep -i -E mcp|api key|endpoint|https://
  • runbash -c cd /tmp && timeout 30 curl -sL https://www.juicer.io/deve...
    • runtimeout 30 curl -sL https://www.juicer.io/developer -o
      • runcurl -sL https://www.juicer.io/developer -o dev.html
        • netwww.juicer.io:443 sent 85 B
        • filewrote ./dev.html
    • runwc -c dev.html
    • runhead ×2
    • rungrep -o -i -E mcp[^"<]{0,80} dev.html
    • rungrep -o -E https?://[a-z0-9./_-]*api[a-z0-9./_-]* dev.html
    • runsort -u
  • runbash -c cd /tmp && timeout 30 curl -sL -A "Mozilla/5.0" https://w...
    • runtimeout 30 curl -sL -A Mozilla/5.0
      • runcurl -sL -A Mozilla/5.0 https://www.juicer.io/developer -o
        • netwww.juicer.io:443 sent 86 B
        • filewrote ./dev.html
        • netDNS lookups for juicer.io, npmjs.org 16 queries
    • runwc -c dev.html
    • runsed s/<[^>]*>/ /g dev.html
    • runtr -s \n \n
    • rungrep -i -E mcp|api|key
    • runhead -20

Exploration log

1 run
  1. llm:baseline ok · 34.4s

    new behavior: dns:registry.npmjs.org, dns:www.juicer.io, exec:curl, ownread:./.env · 8 model steps, 9 commands

    $ ls -la; echo "---"; ls -la skill 2>/dev/null; echo "---"; find skill -maxdepth 3 -type f | head -50 $ env | grep -i -E 'juicer|mcp|api' | sed 's/=.*/=<set>/' $ cat skill/_meta.json; echo "---"; cat skill/skill-card.md; echo "---"; cat README.md; echo "---"; cat package.json; echo "---"; cat .env | sed 's/=.*/=<redacted>/'; echo "---"; ls -la src; echo "---"; $ which juicer juicer-mcp npx node 2>/dev/null; echo "---"; ls ~/.config 2>/dev/null; ls ~/.claude* ~/.openclaw 2>/dev/null; echo "---"; grep -ril "juicer" ~ --include="*.json" --include="*.toml" --incl $ cd /tmp && timeout 60 npm view juicer-mcp version 2>&1 | tail -3; echo "---"; timeout 30 curl -s -o /dev/null -w "%{http_code}\n" https://www.juicer.io/developer $ cd /tmp && timeout 60 npm view juicer-mcp 2>&1 | head -20; echo "=== search ==="; timeout 60 npm search juicer --json 2>&1 | head -40 $ cd /tmp && timeout 30 curl -s https://www.juicer.io/developer | sed 's/<[^>]*>/ /g' | tr -s ' \n' ' \n' | grep -i -E 'mcp|api key|endpoint|https://' | head -30 $ cd /tmp && timeout 30 curl -sL https://www.juicer.io/developer -o dev.html; wc -c dev.html; grep -o -i -E 'mcp[^"<]{0,80}' dev.html | head; echo "---"; grep -o -E 'https?://[a-z0-9./_-]*api[a-z0-9./_- $ cd /tmp && timeout 30 curl -sL -A "Mozilla/5.0" https://www.juicer.io/developer -o dev.html; wc -c dev.html; sed 's/<[^>]*>/ /g' dev.html | tr -s ' \n' ' \n' | grep -i -E 'mcp|api|key' | head -20 model said: I'll start by looking at the skill directory and the project to understand what's available.

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.