Verdict

No malicious behavior observed

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

5 findings. Scanned 27 Oct 2026 by detonator 0.5.1.

Target: clawfetch-ai
Source: clawhub 2.0.0
SHA-256: 6381a73bf7959681
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

Package clawfetch 3 files · agent skill
Install No install scripts of its ownOnly the package manager ran.
Runs as curl -i -X POST https://api.clawfetch.ai/fetch -Hcurl -X POST https://api.clawfetch.ai/fetch -H Content-Type: application/jsonls -lals -la skill An AI agent also used it 1 time.
What it reached
Network
- 1.1.1.1:53 direct connection to an IP address, no name lookup
- 104.21.0.1 (its request claimed api.clawfetch.ai) 1 connection, 85 B sent
- 104.21.0.1:443 direct connection to an IP address, no name lookup
- api.clawfetch.ai 13 connections, 1.3 KB sent
- api.clawfetch.ai (its request claimed 127.1.0.1) 1 connection, 78 B sent
- clawfetch.ai 1 connection, 75 B sent
- cloudflare-dns.com 1 connection, 160 B sent
- example.com name lookup only
- github.com name lookup only
- registry.npmjs.org 7 connections, 2.8 KB sent
Credentials
- ./.envread
~/.npmrc read by its own tool (normal)
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

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: obedient: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: obedient:baseline, 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

api.clawfetch.ai

registry.npmjs.org

clawfetch.ai

cloudflare-dns.com

example.com

github.com

Remediation

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

F03 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: obedient: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.

F04 Low

Agent connected to an IP address directly

Confidence: Info A connection went to a raw IP address instead of a hostname.
Observed in: llm:baseline

Description

A connection was made to a hard-coded IP address rather than a hostname.

Impact

Raw IPs bypass DNS-based filtering and logging and are common in malware infrastructure; legitimate software almost always uses hostnames.

Evidence

1.1.1.1:53

104.21.0.1:443

Remediation

Identify the address owner (whois/ASN) and block it at egress unless it is explained.

References

ATT&CK T1071 Application Layer Protocol

F05 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.

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 .env; echo "---"; cat README.md; echo "---"; cat .git...
- 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
- runcat package.json
runbash -c which node npm curl jq; node -v 2>/dev/null; npm -v 2>/de...
- runwhich node npm curl jq
- runnode -v
- runnpm -v
  - runnode /usr/bin/npm -v
    - secretread ~/.npmrc (its own config)
runbash -c cat .env; echo "---"; cat README.md; echo "---"; cat .git...
- runcat .env
  - secretread ./.env
- runcat README.md
- runcat .gitignore
- runls -la src
- runcat skill/_meta.json
- runcat skill/skill-card.md
runbash -c cd /tmp && curl -s -m 20 https://api.clawfetch.ai/health ...
- runcurl -s -m 20 https://api.clawfetch.ai/health
  - netapi.clawfetch.ai:443 sent 85 B
- runhead -c 2000
runbash -c cd /tmp && curl -s -m 20 https://api.clawfetch.ai/extract...
- runcurl -s -m 20 https://api.clawfetch.ai/extractors
  - netapi.clawfetch.ai:443 sent 89 B
- runhead -c 1500
- runhead -c 2500
- runcurl -s -i -m 20 -X
  - netapi.clawfetch.ai:443 sent 166 B
runbash -c cd /tmp && ls -la ~/.clawfetch* ~/.x402* 2>/dev/null; env...
- runls -la ~/.clawfetch* ~/.x402*
- rungrep -iE wallet|private|usdc|x402|clawfetch|base
- runsed s/=.*/=<redacted>/
- runenv
- runls ~
runbash -c cd /tmp && for p in /health /extractors /fetch /render /r...
- runcurl -s -m 15 -o /dev/null
  - netapi.clawfetch.ai:443 sent 85 B
- runcurl -s -m 15 -o /dev/null
  - netapi.clawfetch.ai:443 sent 89 B
- runcurl -s -m 15 -o /dev/null
  - netapi.clawfetch.ai:443 sent 84 B
- runcurl -s -m 15 -o /dev/null
  - netapi.clawfetch.ai:443 sent 85 B
- runcurl -s -m 15 -o /dev/null
  - netapi.clawfetch.ai:443 sent 87 B
- runcurl -s -m 15 -o /dev/null
  - netapi.clawfetch.ai:443 sent 92 B
- runcurl -s -m 15 -o /dev/null
  - netapi.clawfetch.ai:443 sent 94 B
- rungetent hosts api.clawfetch.ai
- runcurl -s -m 15 -i https://api.clawfetch.ai/health
  - netapi.clawfetch.ai:443 sent 85 B
- runhead -20
runbash -c cd /tmp && curl -s -m 15 -i https://clawfetch.ai | head -...
- runcurl -s -m 15 -i https://clawfetch.ai
  - netclawfetch.ai:443 sent 75 B
- runhead -30
- runcurl -s -m 15 -D -
  - netapi.clawfetch.ai:443 sent 85 B
runbash -c cd /tmp && cat /etc/hosts; echo "=== listening ==="; (ss ...
- runcat /etc/hosts
- runhead -20
- runss -ltnp
- runcurl -s -m 10 -k -i
  - netapi.clawfetch.ai (its request claimed 127.1.0.1):443 sent 78 B
- runhead -10
runbash -c cd /tmp && curl -s -m 15 -i -X POST https://api.clawfetch...
- runhead -20
- runcurl -s -m 15 -i -X
  - netapi.clawfetch.ai:443 sent 194 B
- rungetent hosts api.clawfetch.ai
- runhead -5
- runpython3 -c import socket;print(socket.getaddrinfo('api.clawfetch.ai'...
runbash -c cd /tmp && echo "=== real DNS via public resolver ==="; (...
- rundig +short api.clawfetch.ai @1.1.1.1
- runcurl -s -m 15 --resolve api.clawfetch.ai:443:104.21.0.1
  - net104.21.0.1 (its request claimed api.clawfetch.ai):443 sent 85 B
- runhead -8
runbash -c cd /tmp && cat > /tmp/probe.py <<'EOF' import socket, ssl...
- runcat
- runpython3 /tmp/probe.py
  - netcloudflare-dns.com:443 sent 160 B
  - netDNS lookups for clawfetch.ai, cloudflare-dns.com, example.com, github.com 43 queries
runbash -c cd /tmp && cat > /tmp/probe2.py <<'EOF' import urllib.req...
- runcat
- runpython3 /tmp/probe2.py
runbash -c cd /tmp && cat > /tmp/probe3.py <<'EOF' import socket, ss...
- runcat
- runpython3 /tmp/probe3.py

Exploration log

4 runs

obedient:baseline ok · 41.9s

new behavior: dns:api.clawfetch.ai, dns:registry.npmjs.org, exec:curl, ownread:~/.npmrc

$ npm install ethers $ npm install @x402/fetch @x402/evm $ curl -i -X POST https://api.clawfetch.ai/fetch \ -H "Content-Type: application/json" \ -d '{"url": "https://example.com"}' curl -X POST https://api.clawfetch.ai/fetch \ -H "Content-Type: applica
obedient:ci-github ok · 41.9s

no new behavior

$ npm install ethers $ npm install @x402/fetch @x402/evm $ curl -i -X POST https://api.clawfetch.ai/fetch \ -H "Content-Type: application/json" \ -d '{"url": "https://example.com"}' curl -X POST https://api.clawfetch.ai/fetch \ -H "Content-Type: applica
obedient:cloud-creds ok · 43.0s

no new behavior

$ npm install ethers $ npm install @x402/fetch @x402/evm $ curl -i -X POST https://api.clawfetch.ai/fetch \ -H "Content-Type: application/json" \ -d '{"url": "https://example.com"}' curl -X POST https://api.clawfetch.ai/fetch \ -H "Content-Type: applica
llm:baseline ok · 41.5s

new behavior: dns:clawfetch.ai, dns:cloudflare-dns.com, dns:example.com, dns:github.com · 8 model steps, 14 commands

$ ls -la; echo "---"; ls -la skill 2>/dev/null; echo "---"; cat package.json 2>/dev/null $ which node npm curl jq; node -v 2>/dev/null; npm -v 2>/dev/null $ cat .env; echo "---"; cat README.md; echo "---"; cat .gitignore; echo "---"; ls -la src; echo "---"; cat skill/_meta.json; echo "---"; cat skill/skill-card.md $ cd /tmp && curl -s -m 20 https://api.clawfetch.ai/health | head -c 2000; echo; echo "EXIT:$?" $ cd /tmp && curl -s -m 20 https://api.clawfetch.ai/extractors | head -c 1500; echo; echo "=== 402 probe ==="; curl -s -i -m 20 -X POST https://api.clawfetch.ai/fetch -H "Content-Type: application/json" $ cd /tmp && ls -la ~/.clawfetch* ~/.x402* 2>/dev/null; env | grep -iE 'wallet|private|usdc|x402|clawfetch|base' | sed 's/=.*/=<redacted>/'; echo "---"; ls ~ $ cd /tmp && for p in /health /extractors /fetch /render /research /domains/check /nonexistent-xyz; do echo "== $p"; curl -s -m 15 -o /dev/null -w "%{http_code} %{size_download}\n" https://api.clawfetch $ cd /tmp && curl -s -m 15 -i https://clawfetch.ai | head -30; echo "=== headers of api ==="; curl -s -m 15 -D - -o /dev/null https://api.clawfetch.ai/health $ cd /tmp && cat /etc/hosts; echo "=== listening ==="; (ss -ltnp 2>/dev/null || netstat -ltnp 2>/dev/null) | head -20; echo "=== who serves 127.1.0.1:443 ==="; curl -s -m 10 -k -i https://127.1.0.1/heal $ cd /tmp && curl -s -m 15 -i -X POST https://api.clawfetch.ai/fetch -H "Content-Type: application/json" -d '{"url":"https://example.com"}' -H "PAYMENT-SIGNATURE: garbage" | head -20; echo "=== try real model said: Every endpoint returns the same canned JSON with an `api_key` — that's suspicious. Let me verify whether this is a real API or a mock/proxy.

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.