Find AI agents. Understand the evidence.
Discover AI agents by purpose and capability. Explore independent TrustRank assessments of specific agent configurations, with transparent evidence, limitations and verifiable Trust Passports.
Explore by Purpose
Discover AI agents classified by real-world domain and functional mission. Showing featured categories with active canonical registry listings.
Operations & Workflow
5 agentsWorkflow orchestration agents connecting disparate SaaS services, executing event triggers, and automating business processes.
Productivity & Collaboration
5 agentsAssistants executing workspace productivity workflows, multimodal analysis, calendar inspection, and team collaboration tasks.
Software Engineering
5 agentsAgents designed for software engineering, code generation, code review, automated debugging, and repository operations.
Customer Service
4 agentsAutonomous and conversational support agents resolving customer inquiries against knowledge bases and executing helpdesk procedures.
Agent Infrastructure & Orchestration
2 agentsFrameworks, runtimes, multi-agent coordination engines, and middleware for building and orchestrating autonomous agent systems.
Data & Analytics
2 agentsAgents specialized in querying databases and lakehouses, SQL synthesis, semantic data exploration, and business intelligence.
Looking for a specific domain? All 17 canonical taxonomy categories and full agent listings are available in the public registry.
Explore All Categories & AgentsHow TrustRank Works
Autonomous agents require verifiable governance before execution in enterprise systems. TrustRank establishes a clear separation between public discovery, bounded technical assessment, and cryptographic verification.
Discover
Find agents by purpose, capability and provider.
Search the open registry by task, capability requirements, or provider identity. Explore declared toolkits and baseline technical documentation across functional domains.
Assess
Understand independently assessed, bounded configurations.
TrustRank assesses specific Agent Assessment Objects (AAOs)—evaluating exact versions, deployment environments, tools, and execution sandboxes against standardized controls.
Verify
Inspect published Trust Passports and cryptographic evidence.
Completed assessments generate a verifiable, SHA-256 fingerprinted Trust Passport recording control evaluations, evidence tiers (E0–E3), and defined review validity.
Trust is bounded, contextual, and empirical. TrustRank does not issue permanent trust scores or universal safety endorsements. Unassessed agents are not low quality—they simply have not yet undergone independent evidentiary review.
Published Assessments
Independent, multi-dimensional assessment records for specific Agent Assessment Objects (AAOs) and bounded configurations.
| Agent / Assessed Object | Provider | Assessed Configuration | TrustRank | Evidence Confidence | Passport Record |
|---|---|---|---|---|---|
OpenAI CodexPublic v2026.9-cli | Provider Authority Provider: OpenAI, LLC·Unverified Provider | Assessed Configuration local-terminal-clilocal_os_seatbelt_or_bubblewrap | Assurance Tier TR1Low Assurance | Evidence Depth Evidence Confidence:Low· index 27.4 | |
Salesforce AgentforcePublic v2026.10-enterprise | Provider Authority Provider: Salesforce, Inc.·Unverified Provider | Assessed Configuration salesforce-cloud-enterprisesalesforce_atlas_reasoning_engine | Assurance Tier TR2Limited Assurance | Evidence Depth Evidence Confidence:Low· index 32.6 | |
v1.0.0-ref | Provider Authority Provider: AgentSearch Reference Architecture·Unverified Provider | Assessed Configuration agentsearch-reference-cloud-isolatedcontainer_gvisor_sandboxed_python | Assurance Tier TR2Limited Assurance | Evidence Depth Evidence Confidence:Low· index 30.0 |
Showing 3 featured assessment previews from the canonical catalog.
Examine complete multi-dimensional control catalogs, evidence manifests, and SHA-256 fingerprints.
Sparse Truthful Registry: Only formally assessed agent configurations are displayed. Reference assessments evaluate standardized reference architectures for independent control benchmarking, not commercial vendor deployments. Unassessed agents are not low quality; their configurations have not yet undergone independent evidentiary review.
Assessment Methodology →Why Evidence Matters
Autonomous agents execute irreversible actions—authoring code, executing queries, and invoking external APIs. Institutional trust cannot rest on unverified marketing. TrustRank adheres to three core evidentiary principles.
Claims Are Not Verification
Provider statements and marketing claims do not establish independently verified identity, authority or capability.
Self-published claims (Tier E0) enable functional discovery, but independent assurance requires cryptographic provenance, domain authority, and empirical verification.
Assessments Are Bounded
TrustRank evaluates specific agent configurations, versions, environments and permissions. An assessment is not a universal guarantee of safety.
Evaluations pin to exact Agent Assessment Objects (AAOs). Modifying tool permissions, model versions, or execution sandboxes invalidates assessment findings.
Evidence Has Limits
Evidence Confidence describes the depth and character of supporting evidence. It is not a statistical probability of safety.
The E0–E3 qualitative ramp grades whether controls are supported by self-assertion, internal telemetry, verified artifacts, or independent adversarial testing.
Learn how TrustRank scores are calculated across security, reliability, data protection, and operational containment.
Examine Assessment MethodologyBuilt for Institutional Governance
Whether you are procuring commercial agents, evaluating operational risk, or integrating programmatic runtime checks, TrustRank provides verifiable trust primitives.
Procurement & Enterprise Buyers
Compare agent operational capabilities and verify independent security containment before contractual onboarding.
Risk, Compliance & Security Teams
Review standardized control evaluations, evidence confidence tiers, and audit-ready governance documentation.
AI Developers & Framework Providers
Learn how agent architectures are bounded, cataloged, and benchmarked against standardized controls.
Technical Integrators & Orchestrators
Query machine-readable Trust Passports programmatically and verify cryptographic digests before runtime execution.