Security audit
Security checks across malware telemetry and agentic risk
This skill appears to be a real model-routing tool, but it needs review because it can run provider CLIs with broad local environment access and auto-approved Gemini CLI execution.
Install only if you are comfortable with a router that can pass your task text and derived context to configured model providers, load local OpenClaw environment settings, and run provider CLIs. Review or disable Gemini auto-approval before sensitive use, keep LangSmith prompt/output previews off unless needed, avoid remote Ollama endpoints unless trusted, and treat saved prompts, logs, ledgers, and generated artifacts as sensitive data.
json.dump(build_gemini_temperature_settings(normalized_model, temperature), settings_file)
env[GEMINI_SYSTEM_SETTINGS_ENV_VAR] = settings_path
result = subprocess.run(
command,
capture_output=True,
text=True,)
try:
with urllib.request.urlopen(request, timeout=timeout) as response:
data = json.loads(response.read().decode("utf-8"))
except urllib.error.HTTPError as exc:
body = exc.read().decode("utf-8", errors="replace")json.dump(build_gemini_temperature_settings(normalized_model, temperature), settings_file)
env[GEMINI_SYSTEM_SETTINGS_ENV_VAR] = settings_path
result = subprocess.run(
command,
capture_output=True,
text=True,if client_cls is None:
return None
kwargs: Dict[str, Any] = {}
api_key = os.environ.get("LANGSMITH_API_KEY", "").strip() or os.environ.get("LANGCHAIN_API_KEY", "").strip()
endpoint = os.environ.get("LANGSMITH_ENDPOINT", "").strip() or os.environ.get("LANGCHAIN_ENDPOINT", "").strip()
workspace_id = os.environ.get("LANGSMITH_WORKSPACE_ID", "").strip()
if api_key:if client_cls is None:
return None
kwargs: Dict[str, Any] = {}
api_key = os.environ.get("LANGSMITH_API_KEY", "").strip() or os.environ.get("LANGCHAIN_API_KEY", "").strip()
endpoint = os.environ.get("LANGSMITH_ENDPOINT", "").strip() or os.environ.get("LANGCHAIN_ENDPOINT", "").strip()
workspace_id = os.environ.get("LANGSMITH_WORKSPACE_ID", "").strip()
if api_key:export SOURCE_SRC="$SRC" CONTEXT_OUT="$CTX" ROUTER_PROMPT="$PROMPT" ROUTER_LOG="$LOG" HTML_OUT="$OUT" RUN_DIR
printf '[step] collect context\n' | tee -a "$STATUS"
python3 "$RUN_DIR/collect_context.py" 2>&1 | tee -a "$STATUS"
COLLECT_EXIT=${PIPESTATUS[0]}
if [ "$COLLECT_EXIT" -ne 0 ]; thenjson.dump(build_gemini_temperature_settings(normalized_model, temperature), settings_file)
env[GEMINI_SYSTEM_SETTINGS_ENV_VAR] = settings_path
result = subprocess.run(
command,
capture_output=True,
text=True,def load_openclaw_env_file(env_file: str | None = None) -> None:
"""Load KEY=VALUE pairs from OpenClaw's .env without overriding explicit env."""
configured_path = (env_file or os.environ.get("OPENCLAW_ENV_FILE", "")).strip()
if configured_path:
env_path = os.path.expanduser(configured_path)env_path = os.path.expanduser(configured_path)
else:
openclaw_home = os.path.expanduser(os.environ.get("OPENCLAW_HOME", "~/.openclaw"))
env_path = os.path.join(openclaw_home, ".env")
if not env_path or not os.path.exists(env_path):
return### Post-Completion Artifact Verification For router runs that create or modify artifacts (reports, JSON logs, database ingests, generated files), do not trust the router final stdout alone. Before reporting success to the user: 1. Inspect the declared output artifacts directly (`stat`, read the JSON/Markdown headers, count expected sections/cards when relevant). 2. Compare the router final summary against the saved artifacts and machine-readable logs.
65/65 vendors flagged this skill as clean.
No suspicious patterns detected.