agentor

Name: agentor
Rating: 84 (1 reviews)
Author: CelestoAI

MCP Server

CelestoAI/agentor

Agentor is an open-source Python framework designed for building long-running AI agents equipped with durability, observability, and robust security integrations. It supports both the Model Context Protocol (MCP) and Agent-to-Agent (A2A) communication standards.

192 38Apache-2.0Updated 2026-07-30

Great

View on GitHub Open source file

Overview

Agentor provides an end-to-end proprietary agent loop that emits a typed event stream, allowing runs to be persisted and resumed after process crashes. Featuring LiteMCP for FastAPI-compatible server decorators, a tool registry, and support for any OpenAI-compatible provider, it streamlines the creation of robust, interoperable multi-agent systems.

Capabilities

▸Durable agent runs with event persistence and state recovery
▸Model Context Protocol (MCP) server and client implementation
▸Agent-to-Agent (A2A) protocol support with automated agent card generation
▸Extensible function tool registry and decorator system
▸Typed event stream emission for advanced tracing and observability
▸Compatibility with any OpenAI-compatible LLM endpoint or LiteLLM wrapper

Best for

Building long-running AI agents that need to survive service restarts or crashes via durable event logging., Exposing custom tools and resources securely through a FastAPI-compatible Model Context Protocol (MCP) server., Enabling multi-agent communication and orchestration using the Agent-to-Agent (A2A) protocol., Integrating LLMs with external services like Google Workspace (Gmail, Calendar) using lightweight modular packages.

Works with

GitHub CopilotCursorCodex

Evoa Score breakdown= Σ (score × weight)

Task usefulness20%100 → +20.0

実タスクにどれだけ役立つか（機能の豊富さ・用途の明確さ）。 — AIによるcapabilities/use-cases解析

Code quality15%90 → +13.5

実装・指示の品質。 — AIによるSKILL.md/README解析

Maintenance15%100 → +15.0

リポジトリがどれだけ活発に保守されているか。 — GitHub 最終push日時の新しさ

Documentation12%55 → +6.6

ドキュメントの充実度・分かりやすさ。 — README/独自要約の情報量

Security15%80 → +12.0

危険・不審な挙動が無いか。 — AIによるセキュリティレビュー

Originality10%85 → +8.5

ありふれたラッパーではない独自性。 — AIによる独自性判定

Popularity8%61 → +4.9

コミュニティの採用度。 — GitHub Stars/Forks（対数スケール）

Compatibility5%75 → +3.8

対応AIエージェントの広さ。 — AIによる対応エージェント判定

Weighted total84.2 / 100

ライセンス不明/制限あり（Red）のSkillは総合スコアに0.85倍の補正を適用します。ランキングはこのScoreのみで決まり、広告で変わりません。算出方法の詳細 →

Security considerations

Tracing is opt-in and telemetry data should be handled carefully to avoid leaking sensitive prompts or tool outputs to remote endpoints. Ensure secure handling of API keys and local file stores.