From soundcheck
Detects insecure agent-to-agent calls in multi-agent LLM pipelines lacking authentication, scoped permissions, or message validation. Flags risks and provides Python fix examples with auth headers and checklists.
npx claudepluginhub thejefflarson/soundcheck --plugin soundcheckThis skill uses the workspace's default tool permissions.
Detects agent-to-agent calls that lack authentication, authorization, or permission
Flags vulnerable patterns in autonomous LLM agents enabling irreversible actions without oversight. Suggests fixes like impact classification, tool allowlists, pre-dispatch auditing, and structured parameters for safe workflows.
Implements governance patterns for AI agents: policy-based tool controls, intent classification, trust scoring, audit trails, rate limits. For LangChain, CrewAI, OpenAI Agents.
Implements hooks for permission control, blocking dangerous operations, and audit trails in custom Claude Code agents.
Share bugs, ideas, or general feedback.
Detects agent-to-agent calls that lack authentication, authorization, or permission scoping. When agents blindly trust messages from other agents, a compromised or malicious agent can hijack the entire pipeline.
requests.post("http://worker/run", json={"task": task}) — inter-agent call with no auth token# Orchestrator: include auth header on every inter-agent call
resp = requests.post(
"http://worker/run",
json={"task": task},
headers={"Authorization": f"Bearer {AGENT_SECRET}"},
)
# Worker: reject calls missing a valid token
if request.headers.get("Authorization") != f"Bearer {AGENT_SECRET}":
return jsonify({"error": "unauthorized"}), 401
Flag the vulnerable call site, explain the risk and the correct fix pattern, then continue with the original task.