ClawAudit verdict
ollama-qwen-nothink
qwen-nothink
Creates local Ollama model variants with thinking mode disabled by patching a derived manifest; all operations are local, no network exfiltration, matches stated purpose.
⚠ Flagged for review — coarse, uncorroborated signal, not a confirmed exploit. Review the config yourself before installing.
Automated static analysis — not a human review. ClawAudit flags capabilities, not confirmed intent, and can produce false positives. Disagree with this verdict? Use Dispute below.
Findings (6)
Raw model control tokens — prompt injection via token manipulation
SKILL.md · code · <|im_start|>
HTTP request to bare IP address — common in malicious payloads
scripts/create_qwen_nothink_ollama.py · prose · downgraded · http://127.0.0.1
subprocess execution — runs system commands from Python
scripts/create_qwen_nothink_ollama.py · prose · downgraded · subprocess.run(
Blob URL — may embed executable content
references/manifest-patch.md · prose · downgraded · blob:
Python urllib.request — network access
scripts/create_qwen_nothink_ollama.py · prose · downgraded · urllib.request
Python os.environ.get — reads environment variable
scripts/create_qwen_nothink_ollama.py · prose · downgraded · os.environ.get(
Why the tier is capped
Execution sink present in raw bytes (Hard Floor: class D). Final tier capped at Caution — cannot be lifted by any downgrade, example-payload opt-in, or allowlist.
Permissions & capabilities
No declared permissions — minimal attack surface.
Is this flag fair?
Thanks — recorded.