Observatory · Catalog v0.4.0

Agent sources

Research index, not a benchmark ranking. Records distinguish source documentation from independently reproduced measurements. No record on this page is marked independently verified.

← Observatory Models →Protocols →

Search catalogSource status

Showing 5 of 5 records

Governance Commons · Needs review

GC Agent Roster

Curated development-role model roster; not an independently benchmarked set of agent implementations.

Version: Source metadata 1.3.0

Evidence: Documentation only · Independent system evaluation: Not performed

Source review: Pending · Source snapshot: Not captured

Internal source last manually verified 2026-08-07; private roster entries and rankings are not republished.

Primary source ↗

LangChain · Needs review

LangGraph

Agent orchestration framework with graph-based state and workflow control.

Version: Implementation version pending

Evidence: Documentation only · Independent system evaluation: Not performed

Source review: Pending · Source snapshot: Not captured

Framework entry, not a benchmarked agent instance.

Primary source ↗

CrewAI · Needs review

CrewAI

Framework for composing and running role-based AI agent workflows.

Version: Implementation version pending

Evidence: Documentation only · Independent system evaluation: Not performed

Source review: Pending · Source snapshot: Not captured

Specific agent deployments and evaluation results are not yet cataloged.

Primary source ↗

Microsoft Research · Needs review

AutoGen

Framework for developing multi-agent applications.

Version: Implementation version pending

Evidence: Documentation only · Independent system evaluation: Not performed

Source review: Pending · Source snapshot: Not captured

Check lifecycle and version status against upstream before production recommendations.

Primary source ↗

OpenAI · Needs review

OpenAI Agents SDK

Software toolkit for building agent workflows with tools and handoffs.

Version: SDK version pending

Evidence: Documentation only · Independent system evaluation: Not performed

Source review: Pending · Source snapshot: Not captured

Toolkit, not a separately measured autonomous agent.

Primary source ↗

Methodology: A primary-source link is not proof of a current model identifier, interoperability, security, conformance or performance. All entries require editorial verification and dated snapshots before claims or comparative rankings can be published.