AI agent security
Excessive Agency Controls Store
Keeping agent autonomy within bounds
In the pipeline — stocking up Cluster: Security & Trust
Arriving in the Excessive Agency Controls Store
0 live products today — more arriving daily. These 1 kits are in production for this store.
In the pipelineExcessive Agency Control Pack
PDF · AI agent security
$29 one-time
Least-privilege tool grants and human confirmation for destructive actions under LLM06.
Sibling stores in SecWarden
More specialized shelves in the same ecosystem.
Prompt Injection Defense Store
AI agent security
Defending agents from direct and indirect injection
Browse →System Prompt Protection Store
AI agent security
Stopping system-prompt extraction
Browse →Secret-Leak Containment Store
AI agent security
Containing leaks of secrets and credentials
Browse →Credential Handling Policies Store
AI agent security
Safe credential patterns for agents
Browse →API-Key Lifecycle Management Store
AI agent security
Issuing, rotating, and revoking agent keys
Browse →Data Poisoning Defenses Store
AI agent security
Defending training and retrieval from poisoning
Browse →Supply Chain Security Store
AI agent security
Securing the agent software supply chain
Browse →Output Handling Safeguards Store
AI agent security
Treating agent output as untrusted
Browse →Vector Store Security Store
AI agent security
Securing embeddings and retrieval
Browse →Misinformation Controls Store
AI agent security
Reducing confident falsehoods
Browse →Unbounded Consumption Guards Store
AI agent security
Stopping runaway token and compute burn
Browse →Red Team Playbooks Store
AI agent security
Adversarial testing for agent systems
Browse →
Parent ecosystem
The Excessive Agency Controls Store is one of 50 stores inside SecWarden.
SecWarden
AI agent security
Attack your agents before someone else does: injection tests, guardrails, and tamper-evident logs.
Explore the ecosystem →
Start here
Three free or cheap entry points — pick whichever fits.
1
Free MIT playbook
The ghostcorpnet playbook — free, MIT-licensed. Read it here →
2
The $5 AI Agent Risk Scan
A five-minute self-audit of your agent setup. Get the scan →
Build on SecWarden
Every ecosystem follows the open ghostcorpnet Pack Standard v1 — interoperable packs, a public directory, and a certification badge.