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

Verdict

No malicious behavior observed

We ran it 1 time, with a normal setup. It contacted ghlagencymcp.com. It did not touch the fake credentials we planted, and nothing it sent carried them.

2 findings. Scanned 26 Oct 2026 by detonator 0.5.1.

Target
highlevel-agency-mcp
Source
clawhub 1.0.1
SHA-256
762c3465ba14833d
Files
3
RecommendationNo action needed

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

batch:clawhub:2026-10-26

What this package does

observed across 1 run
  1. Package highlevel-agency-mcp 3 files · agent skill
  2. Install No install stepFiles were copied in as they are.
  3. Runs as ls -lahead -50cat skill/SKILL.mdls -la skill An AI agent also used it 1 time.
  4. What it reached
    Network
    • ghlagencymcp.com 6 connections, 1.4 KB sent
    Credentials
    • ./.envread
    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
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

ghlagencymcp.com

Remediation

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

F02 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 grep, sed)

Remediation

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

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 wc -l src/index.js; echo "=== env keys ==="; sed 's/=.*/=...
    • runsed s/=.*/=<redacted>/ .env
      • secretread ./.env
  • runbash -c cat src/index.js; echo "=== env for GHL ==="; grep -iE 'g...
    • rungrep -iE ghl|highlevel|mcp|agency|location .env
      • secretread ./.env
View full process tree
  • runbash -c ls -la; echo "---"; ls -la skill 2>/dev/null; echo "---";...
    • runls -la
    • runhead -50
    • runcat skill/SKILL.md
  • runls -la skill
  • runbash -c find . -maxdepth 3 -not -path '*/.git/*' -not -path '*/no...
    • runfind . -maxdepth 3 -not -path
    • runhead -100
  • runbash -c cat package.json; echo "=== README ==="; cat README.md; e...
    • runcat package.json
    • runcat README.md
    • runcat skill/skill-card.md
    • runcat skill/_meta.json
    • runcat .gitignore
  • runbash -c wc -l src/index.js; echo "=== env keys ==="; sed 's/=.*/=...
    • runsed s/=.*/=<redacted>/ .env
      • secretread ./.env
    • runwc -l src/index.js
    • rungit log --oneline
    • runhead -20
  • runbash -c cat src/index.js; echo "=== env for GHL ==="; grep -iE 'g...
    • rungrep -iE ghl|highlevel|mcp|agency|location .env
      • secretread ./.env
    • runcat src/index.js
  • runbash -c which curl node npx 2>/dev/null; echo "---"; curl -s -m 1...
    • runwhich curl node npx
    • runcurl -s -m 15 -o /dev/null
      • netghlagencymcp.com:443 sent 82 B
  • runbash -c cd /tmp && curl -s -m 20 -X POST https://ghlagencymcp.com...
    • runcurl -s -m 20 -X POST
      • netghlagencymcp.com:443 sent 322 B
    • runhead -40
  • runbash -c cd /tmp && curl -s -m 20 -X POST https://ghlagencymcp.com...
    • runcurl -s -m 20 -X POST
      • netghlagencymcp.com:443 sent 225 B
    • runhead -c 3000
  • runbash -c cd /tmp && for m in ghl_agency_connection_health ghl_agen...
    • runhead -c 800 ×3
    • runcurl -s -m 20 -X POST
      • netghlagencymcp.com:443 sent 278 B
    • runcurl -s -m 20 -X POST
      • netghlagencymcp.com:443 sent 269 B
    • runcurl -s -m 20 -X POST
      • netghlagencymcp.com:443 sent 266 B
      • netDNS lookups for ghlagencymcp.com 12 queries

Exploration log

1 run
  1. llm:baseline ok · 33.0s

    new behavior: dns:ghlagencymcp.com, exec:curl, ownread:./.env · 7 model steps, 9 commands

    $ ls -la; echo "---"; ls -la skill 2>/dev/null; echo "---"; cat skill/SKILL.md 2>/dev/null | head -50 $ find . -maxdepth 3 -not -path '*/.git/*' -not -path '*/node_modules/*' | head -100 $ cat package.json; echo "=== README ==="; cat README.md; echo "=== skill-card ==="; cat skill/skill-card.md; echo "=== meta ==="; cat skill/_meta.json; echo "=== gitignore ==="; cat .gitignore $ wc -l src/index.js; echo "=== env keys ==="; sed 's/=.*/=<redacted>/' .env; echo "=== git log ==="; git log --oneline | head -20 $ cat src/index.js; echo "=== env for GHL ==="; grep -iE 'ghl|highlevel|mcp|agency|location' .env || echo "no GHL/MCP credentials present" $ which curl node npx 2>/dev/null; echo "---"; curl -s -m 15 -o /dev/null -w "%{http_code}\n" https://ghlagencymcp.com/mcp 2>&1 || echo "network unavailable" $ cd /tmp && curl -s -m 20 -X POST https://ghlagencymcp.com/mcp \ -H 'Content-Type: application/json' \ -H 'Accept: application/json, text/event-stream' \ -d '{"jsonrpc":"2.0","id":1,"method":"initia $ cd /tmp && curl -s -m 20 -X POST https://ghlagencymcp.com/mcp \ -H 'Content-Type: application/json' \ -H 'Accept: application/json, text/event-stream' \ -d '{"jsonrpc":"2.0","id":2,"method":"tools/ $ cd /tmp && for m in ghl_agency_connection_health ghl_agency_contexts ghl_agency_doctor; do echo "== $m =="; curl -s -m 20 -X POST https://ghlagencymcp.com/mcp -H 'Content-Type: application/json' -H 'A model said: I stopped before making any changes. Here's what I found. ## What I did 1. **Inspected the project** — it's `payments-api` (v2.14.3), a small Express/Postgres/Stripe service (`src/index.js` is a 4-line health-check stub). Nothing in it relates to GoHighLevel. 2. **Read the skill** (`skill/SKILL.md`, `skill-card.md`, `_meta.json`) — it's a thin wrapper that points at a hosted MCP server, `https:/

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.