🌀
← Back to Articles & Artefacts
artefactseast

STCMastery-ACP-daee8f3e-e23b-4c9a-947d-2bc92abf3c3f 260207

IAIP Research
stcmastery-acp-daee8f3e-e23b-4c9a-947d-2bc92abf3c3f-260207

STCMastery-ACP-daee8f3e-e23b-4c9a-947d-2bc92abf3c3f 260207#

Agent Client Protocol

'gemini --experimental-acp or copilot --acp' exist and give us some capabilities, our local agent will have to make simple experiment to use this protocol, gather :

  • Academic literature survey of what is implied is such protocol and draft an article that goes thru each sources and their finding (2 pages max that teach me a little what is the state of the academic field in this regards and help understands the reseach questions (inquiries) that are currently in progress.
  • Potential Github or other sources that could help the local agents educate themselves on the topic. Maybe you could draft a claude-code skills that enable agents to fully understand what we are talking about...

I've completed your research on Agent Communication Protocol (ACP) and related agent interoperability standards. Here's what I've prepared:

Summary of Findings#

The Experimental Commands You Mentioned ARE Real:

  • gemini --experimental-acp - Starts Gemini CLI in ACP mode12
  • copilot --acp - Starts GitHub Copilot CLI as an ACP server345

Both are production features (in public preview) enabling local agents to communicate via standardized protocols.

Documents Created#

1. Academic Literature Survey (2 pages)#

A focused academic survey covering:

State of the Field:

  • Four major protocols emerged 2024-2025: MCP, ACP, A2A, ANP
  • Each addresses different interoperability layers (agent↔tools vs agent↔agent)
  • Classification framework: context-oriented vs inter-agent, general vs domain-specific67

Current Research Questions:

  1. Semantic Interoperability - Extending beyond symbolic reasoning to multimodal data8
  2. Communication Efficiency - Goal-oriented messaging for resource-constrained environments9
  3. Security & Trust - OAuth2, mTLS, audit trails across protocol lifecycles710
  4. Scalability - Router/supervisor patterns for collective intelligence1112
  5. Protocol Coexistence - Phased adoption: MCP→ACP→A2A→ANP7
  6. Dynamic Discovery - Runtime capability negotiation vs static registries137
  7. Standardization - W3C standardization expected 2026-202714

Key Insight: MCP and ACP are complementary, not competing. MCP connects agents to tools/data; ACP connects agents to each other.1512

2. Practical Implementation Guide#

A hands-on resource including:

Quick Start Commands:

  • Configuration examples for Gemini CLI and Copilot CLI
  • IDE integration (IntelliJ IDEA)1

Essential GitHub Repositories:

  • Official specs: modelcontextprotocol, agentclientprotocol1617
  • Implementations: i-am-bee/acp, microsoft/mcp1819
  • Example agents: BeeAI Chat, Slack+MCP, RAG with LlamaIndex20

SDK Implementations:

  • Python: acp-sdk-python (enterprise-ready)10
  • TypeScript: @agentclientprotocol/sdk17
  • Rust, Kotlin: Full-featured SDKs available17

Step-by-Step Tutorial: Complete walkthrough from choosing a framework to deploying your first ACP agent with OAuth2 security.

Claude-Code Skill: A structured skill definition enabling agents to fully understand ACP, MCP, and multi-agent architectures.

Key Architectural Patterns for Your Local Agents#

Pattern 1: MCP + ACP Hybrid (Recommended)

``` Your Agent uses MCP to access tools/data uses ACP to coordinate with other agents ```

Pattern 2: Router Agent with Specialists

``` Router Agent decomposes tasks → Data Agent → Code Agent
→ Research Agent ```

Pattern 3: Discovery via Agent Cards

``` HTTP GET /.well-known/agent.json Returns: capabilities, authentication, endpoints ```

Academic Foundation#

Must-Read Papers:

  1. Yang et al. (2025) - "A Survey of AI Agent Protocols" (arXiv:2504.16736)6
  2. Ehtesham et al. (2025) - "Survey of Agent Interoperability Protocols" (arXiv:2505.02279)7
  3. Yan et al. (2025) - "Beyond Self-Talk: Communication-Centric Survey" (arXiv:2502.14321)21

Historical Context:

  • FIPA-ACL (2000): IEEE standard based on speech act theory2223
  • Modern protocols build on these foundations but optimize for LLM agents

Timeline and Predictions#

2025-2026: Protocol convergence, W3C standardization begins 2026-2027: Domain-specific variants (healthcare, finance), edge/IoT protocols 2027-2030: Universal standard emerges ("HTTP for agents"), agent marketplaces14

Next Steps for Your Local Agents#

  1. Experiment with gemini --experimental-acp or copilot --acp
  2. Install Python SDK: pip install acp-sdk-python
  3. Study example implementations at agentcommunicationprotocol.dev24
  4. Build your first agent following the step-by-step guide
  5. Integrate MCP servers for tool access2518

The documents provide everything needed for local agents to self-educate on agent communication protocols, from theoretical foundations to production deployment patterns. 26272829303132333435363738394041424344454647484950515253545556575859

⁂

Footnotes#

  1. https://glaforge.dev/posts/2026/02/01/how-to-integrate-gemini-cli-with-intellij-idea-using-acp/ ↩ ↩2

  2. https://geminicli.com/docs/get-started/configuration/ ↩

  3. https://docs.github.com/fr/enterprise-cloud@latest/copilot/reference/acp-server ↩

  4. https://github.blog/changelog/2026-01-28-acp-support-in-copilot-cli-is-now-in-public-preview/ ↩

  5. https://docs.github.com/en/enterprise-cloud@latest/copilot/reference/acp-server ↩

  6. https://arxiv.org/abs/2504.16736 ↩ ↩2

  7. https://arxiv.org/html/2505.02279v1 ↩ ↩2 ↩3 ↩4 ↩5

  8. https://etasr.com/index.php/ETASR/article/view/2841 ↩

  9. https://arxiv.org/abs/2508.07720 ↩

  10. https://pypi.org/project/acp-sdk-python/ ↩ ↩2

  11. https://agentcommunicationprotocol.dev/core-concepts/architecture ↩

  12. https://devblogs.microsoft.com/semantic-kernel/guest-blog-building-multi-agent-solutions-with-semantic-kernel-and-a2a-protocol/ ↩ ↩2

  13. https://www.gocodeo.com/post/acp-the-protocol-standard-for-ai-agent-interoperability ↩

  14. https://www.ruh.ai/blogs/ai-agent-protocols-2026-complete-guide ↩ ↩2

  15. https://research.ibm.com/blog/agent-communication-protocol-ai ↩

  16. https://github.com/modelcontextprotocol ↩

  17. https://github.com/agentclientprotocol/agent-client-protocol ↩ ↩2 ↩3

  18. https://github.com/microsoft/mcp ↩ ↩2

  19. https://github.com/i-am-bee/acp ↩

  20. https://agentcommunicationprotocol.dev/introduction/example-agents ↩

  21. https://arxiv.org/abs/2502.14321 ↩

  22. https://martinpilat.com/en/multiagent-systems/communication-protocols ↩

  23. https://smythos.com/developers/agent-development/fipa-agent-communication-language/ ↩

  24. https://agentcommunicationprotocol.dev/introduction/quickstart ↩

  25. https://github.com/modelcontextprotocol/servers ↩

  26. https://onereach.ai/blog/power-of-multi-agent-ai-open-protocols/ ↩

  27. https://learnprompting.org/blog/agntcy-open-standard-for-agent-interoperability ↩

  28. https://modelcontextprotocol.io/docs/learn/client-concepts ↩

  29. https://www.ciodive.com/news/-Microsoft-AI-agent-standard-Google-a2a-interoperability/747593/ ↩

  30. https://www.ssonetwork.com/intelligent-automation/columns/ai-agent-protocols-10-modern-standards-shaping-the-agentic-era ↩

  31. https://www.ibm.com/think/topics/agent-communication-protocol ↩

  32. https://fabrix.ai/blog/some-of-the-open-source-standards-used-with-ai-agents-or-agentic-frameworks/ ↩

  33. https://www.reddit.com/r/AgentsOfAI/comments/1mfchfo/10_modern_ai_agent_protocols_standardizing_ai/ ↩

  34. https://www.reddit.com/r/AI_Agents/comments/1k7fdhe/we_tried_building_actual_agenttoagent_protocols/ ↩

  35. https://www.forrester.com/blogs/interoperability-is-key-to-unlocking-agentic-ais-future/ ↩

  36. https://github.com/google-gemini/gemini-cli/issues/7458 ↩

  37. https://docs.github.com/ko/copilot/reference/acp-server ↩

  38. https://modelcontextprotocol.io/examples ↩

  39. https://github.com/google-gemini/gemini-cli/issues/10855 ↩

  40. https://github.com/google-gemini/gemini-cli/issues/8672 ↩

  41. https://docs.github.com/copilot/reference/acp-server ↩

  42. https://modelcontextprotocol.io ↩

  43. https://github.com/AI-Engineer-Foundation/agent-protocol ↩

  44. https://github.com/agi-inc/agent-protocol ↩

  45. https://www.ibm.com/think/tutorials/acp-ai-agent-interoperability-building-multi-agent-workflows ↩

  46. http://www.diva-portal.org/smash/get/diva2:1970755/FULLTEXT01.pdf ↩

  47. https://www.openagentschool.org/concepts/acp ↩

  48. https://smythos.com/developers/agent-development/agent-communication-in-multi-agent-systems/ ↩

  49. https://agentclientprotocol.github.io/typescript-sdk/ ↩

  50. https://www.cs.cmu.edu/~softagents/papers/paper778.pdf ↩

  51. https://onereach.ai/blog/what-is-a2a-agent-to-agent-protocol/ ↩

  52. https://smythos.com/developers/agent-development/types-of-agent-communication-languages/ ↩

  53. https://ieeexplore.ieee.org/document/10983821/ ↩

  54. https://iabtechlab.com/resolve-to-update-your-tech-lab-standards-in-2026/ ↩

  55. https://www.iiia.csic.es/~puyol/SEIAD2001/publicacions/ACL-FIPA.doc.pdf ↩

  56. https://nrc-publications.canada.ca/eng/view/accepted/?id=0c21987d-d963-444d-893d-5fb93a7e8b7b ↩

  57. https://theconversation.com/ai-agents-arrived-in-2025-heres-what-happened-and-the-challenges-ahead-in-2026-272325 ↩

  58. http://www.fipa.org/specs/fipa00018/OC00018.pdf ↩

  59. https://dl.acm.org/doi/10.1145/3670105.3670127 ↩