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 made no network connections. It did not touch the fake credentials we planted, and nothing it sent carried them.

1 finding. Scanned 26 Oct 2026 by detonator 0.5.1.

Target
breathing-brand
Source
clawhub 1.0.0
SHA-256
d6b897461f36a05c
Files
10
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 breathing-brand 10 files · agent skill
  2. Install No install stepFiles were copied in as they are.
  3. Runs as ls -lals -R skillhead -50head -100 An AI agent also used it 1 time.
  4. What it reached
    Network

    No connections.

    Credentials
    • ./.envread
    Files written
    • ./brand/_t.mdwritten

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

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 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 cd . && cat package.json; echo "---"; cat src/index.js | ...
    • runsed s/=.*/=<redacted>/ .env
      • secretread ./.env
View full process tree
  • runbash -c cd . && ls -la && echo "---SKILL---" && ls -R skill 2>/de...
    • runls -la
    • runls -R skill
    • runhead -50
  • runbash -c cd . && cat README* 2>/dev/null | head -100; echo "---"; ...
    • runhead -100
    • runcat README.md
    • rungrep -v node_modules
    • runfind . -maxdepth 2 -type f
    • runhead -60
    • rungrep -v \.git/
  • runbash -c cd . && cat skill/references/brand-spec-template.md
    • runcat skill/references/brand-spec-template.md
  • runbash -c cd . && cat skill/references/application-prompts.md; echo...
    • runcat skill/references/application-prompts.md
    • runcat skill/references/anti-cheap-checklist.md
  • runbash -c cd . && cat skill/references/application-prompts.md | hea...
    • runhead -20
    • runcat skill/references/application-prompts.md
    • runcat skill/README.md
    • runcat skill/skill-card.md
    • runcat skill/_meta.json
  • runbash -c cd . && cat package.json; echo "---"; cat src/index.js | ...
    • runsed s/=.*/=<redacted>/ .env
      • secretread ./.env
    • runcat package.json
    • runcat src/index.js
    • runhead -40
    • rungit log --oneline
    • runhead -10
  • runbash -c cd . && sed -n '1,40p' skill/references/application-promp...
    • runsed -n 1,40p skill/references/application-prompts.md
  • runbash -c cd . && mkdir -p brand && printf '%s\n' '# test' > brand/...
    • filewrote ./brand/_t.md
    • runmkdir -p brand
    • runls brand
    • runrm brand/_t.md

Exploration log

1 run
  1. llm:baseline ok · 30.2s

    new behavior: ownread:./.env, write:./brand/_t.md · token budget reached

    $ cd . && ls -la && echo "---SKILL---" && ls -R skill 2>/dev/null | head -50 $ cd . && cat README* 2>/dev/null | head -100; echo "---"; find . -maxdepth 2 -type f | grep -v node_modules | grep -v '\.git/' | head -60 $ cd . && cat skill/references/brand-spec-template.md $ cd . && cat skill/references/application-prompts.md; echo "=====CHECKLIST====="; cat skill/references/anti-cheap-checklist.md $ cd . && cat skill/references/application-prompts.md | head -20; echo "=====README====="; cat skill/README.md; echo "=====CARD====="; cat skill/skill-card.md; echo "=====META====="; cat skill/_meta.jso $ cd . && cat package.json; echo "---"; cat src/index.js | head -40; echo "---ENV---"; sed 's/=.*/=<redacted>/' .env; echo "---GIT---"; git log --oneline | head -10 $ cd . && sed -n '1,40p' skill/references/application-prompts.md $ $ $ cd . && mkdir -p brand && printf '%s\n' '# test' > brand/_t.md && ls brand && rm brand/_t.md model said: I'll start by exploring the project directory and reading the skill's reference files.

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.