|
6 | 6 | [](https://github.com/Zijian-Ni/awesome-ai-agents-2026) |
7 | 7 | [](https://opensource.org/licenses/MIT) |
8 | 8 | [](CONTRIBUTING.md) |
9 | | -[](#) |
10 | | -[](#) |
| 9 | +[](#) |
| 10 | +[](#) |
11 | 11 | [](#) |
12 | | -[](#οΈ-status-legend) |
| 12 | +[](#οΈ-status-legend) |
13 | 13 | [](README.zh-CN.md) |
14 | 14 | [](README.ja.md) |
15 | 15 |
|
@@ -449,6 +449,7 @@ Entries may carry one or more status tags so readers can judge maturity at a gla |
449 | 449 |
|
450 | 450 | - [OpenAI Agents SDK](https://github.com/openai/openai-agents-python) - π [Major update April 15, 2026](https://openai.com/index/the-next-evolution-of-the-agents-sdk/): native sandbox execution, first-class MCP integration, sub-agent / handoff patterns, and Codex-style filesystem tools for production-ready multi-agent workflows.  |
451 | 451 | - [Agentic AI Foundation](https://www.linuxfoundation.org/) - π Linux Foundation fund co-founded by Anthropic, Block, and OpenAI to govern open agent standards. |
| 452 | +- [Kuberna Labs](https://github.com/kawacukennedy/kuberna-labs) - β οΈ **Unverified.** Cross-chain intent execution protocol for AI agents. Claims ERC-8004 on-chain identity, zkTLS/TEE attestation, and a typed intent schema enabling agents to autonomously execute transactions across NEAR, Base, and Mantle with verifiable execution proofs. New repo, independent adoption unverified β listed for visibility, evaluate before depending on it. |
452 | 453 |
|
453 | 454 | --- |
454 | 455 |
|
@@ -567,6 +568,7 @@ Entries may carry one or more status tags so readers can judge maturity at a gla |
567 | 568 |
|
568 | 569 | *Tools and frameworks for securing AI agents against prompt injection, data leaks, and misuse.* |
569 | 570 |
|
| 571 | +- [AgentGate](https://github.com/ElamOlame31/agentgate-public) - π Pre-execution authorization PDP for autonomous AI agents. Scores trust across 4 dimensions per request, detects 24h kill-chain patterns (BULK_READ_THEN_EXFIL, SENSITIVITY_RAMP), Merkle-chained audit trail. MIT license, drop-in with LangGraph, LangChain, and AutoGen. [tryagentgate.com](https://www.tryagentgate.com/) |
570 | 572 | - [prompt-firewall](https://github.com/Zijian-Ni/prompt-firewall) - Firewall for LLM prompts β detect and block prompt injection attacks.  |
571 | 573 | - [LLM Guard](https://github.com/protectai/llm-guard) - The Security Toolkit for LLM Interactions β input/output scanners for AI.  |
572 | 574 | - [Rebuff](https://github.com/protectai/rebuff) - π¦ **Archived** (2024-08). Self-hardening prompt injection detector β detect, deflect, and report. Listed for historical reference; no longer maintained.  |
@@ -809,6 +811,7 @@ Entries may carry one or more status tags so readers can judge maturity at a gla |
809 | 811 | - [Gemini Intelligence](https://blog.google/products-and-platforms/products/chrome/bringing-chrome-ai-to-android/) - π **May 12, 2026** (Android Show: I/O Edition). Proactive agentic AI features integrated into Googlebooks laptops, Wear OS, Android Auto, Android XR, and starting on the latest Samsung Galaxy + Pixel devices. Auto-creates shopping carts from grocery lists, books spin classes, filler-word removal via the Rambler speech-to-text. |
810 | 812 | - [Gemini Spark](https://9to5google.com/2026/05/14/gemini-spark-insight/) - π **May 14, 2026** (pre-I/O leak / insight). Upcoming branded agent capability inside the Gemini app for autonomously running multi-step processes; sits above Gemini 3.1 Pro reasoning stack. |
811 | 813 | - [QwenPaw](https://github.com/agentscope-ai/QwenPaw) - π π¨π³ **May 2026 rebrand from CoPaw**. Self-hostable personal assistant in the Qwen / AgentScope family. Local-first memory, hot-loadable skills, multi-agent collaboration, multi-channel (DingTalk / Feishu / WeChat / Discord / Telegram), tool guard + skill scanner. Apache-2.0.  |
| 814 | +- [AI Growth Agents for Marketers](https://github.com/thaolst/ai-growth-agents-for-marketers) - π Growth marketing prompts and Python agents built from real fintech campaigns in Southeast Asia. Covers campaign briefs, MEU planning, and A/B test analysis with multi-agent workflows. Agent Skills format β installable via `npx skills add`. Bilingual VI + EN. MIT. |
812 | 815 |
|
813 | 816 | --- |
814 | 817 |
|
@@ -849,6 +852,7 @@ Entries may carry one or more status tags so readers can judge maturity at a gla |
849 | 852 | - [Bristol Myers Squibb β Claude Enterprise](https://news.bms.com/news/corporate-financial/2026/Bristol-Myers-Squibb-Announces-Strategic-Agreement-with-Anthropic-to-Position-Claude-Enterprise-as-the-Shared-Intelligence-Platform-Across-Its-Global-Operations/default.aspx) - π **May 20, 2026**. BMS standardises on Claude Enterprise as its shared intelligence platform for **30,000+ employees**, embedding agentic Claude into drug-discovery / development / delivery pipelines. First top-5 pharma to make a public, company-wide Claude commitment. |
850 | 853 | - [Kore.ai Artemis Agent Platform](https://venturebeat.com/technology/kore-ai-launches-artemis-ai-agent-platform-expands-challenge-to-microsoft-and-salesforce) - π **May 22, 2026** (launched on Azure). AI-native enterprise agent platform built around the new YAML-style **Agent Blueprint Language (ABL)** for declarative multi-agent workflows. Kore.ai's structural challenge to Copilot Studio and Agentforce. |
851 | 854 | - [FPT Flezi Foundryβ’](https://lasvegassun.com/news/2026/may/22/fpt-launches-flezi-foundry-advancing-ai-augmented-/) - π **May 22, 2026**. AI-augmented delivery platform with two governed Service-as-a-Software modes β **Agentic Development Lifecycle (ADLC)** for full SDLC agent crews and **Agentic Managed Services (AMS)** for incident-resolution agents on top of existing ITOps. |
| 855 | +- [GreenOps Agent](https://github.com/raghu-putta/greenops-agent) - π A 4-agent GCP cost and carbon optimization pipeline built on Google ADK, Gemini Flash, and Cloud Run. Detects idle VMs, unattached disks, and unused reserved IPs to calculate COβ footprint and execute cleanups with human approval. |
852 | 856 |
|
853 | 857 | ## π Agent Evaluation & Observability |
854 | 858 |
|
|
0 commit comments