Core Maintainer Meeting 2025-12-11

core-maintainers/2025-12-11.md

@@ -0,0 +1,43 @@
+# Core Maintainers Meeting - 2025-12-11
+
+## Attendees
+
+## Agenda
+
+- **RFDs**
+  - [Proxy Chains](https://github.com/agentclientprotocol/agent-client-protocol/pull/243) (@nikomatsakis)
+    - Working prototype being refined for proxy middleware functionality
+    - Enables two key features:
+      - Initialize as proxy with message forwarding to client/agent predecessors/successors. Able to compose components together, so you can affect your ACP setup without implementing an entire agent or client
+        - Extending something like "claude skills" across multiple agents
+      - "ACP" MCP transport
+    - Ready for draft merge pending final tweaks, asked for vibe check from the team
+      - no major concerns raised
+    - Demo planned for end of week
+  - [ACP Agent Registry](https://github.com/agentclientprotocol/agent-client-protocol/pull/289) (@ignatov)
+    - Move from links on GitHub to downloading agents from Registry
+    - Zed could move to support downloading agents from the registry
+    - could also tie-in to proxy components, host in additional to agents themselves
+    - Another goal is to allow people to use a custom registry that works with a common spec
+      - basically just a json file with a specific format
+      - would be nice to have this ~Q2 2026
+  - Authentication flow for agents [auth](https://github.com/agentclientprotocol/agent-client-protocol/pull/330/files) (@anna239)
+    - currently, most agents don't supply auth methods
+    - in general, can we provide agents with the ability to present a gui or request data from the user?
+    - for JetBrains it would be nice to have a separate flow for authentication
+      - different UX flow needed
+      - need to be careful to avoid phishing issues, especially once we're downloading agents from the registry
+      - Most agents right now are supporting api keys, allow to reuse key from the keychain
+  - Next Edit Suggestion [NES](https://github.com/agentclientprotocol/agent-client-protocol/pull/325), @ignatov
+    - Current RFD isn't where we'd want it for an API, but it at least provides an anchor for discussion
+    - less of a common workflow across edit prediction providers, especially around how they handle context gathering
+      - some are just a model, you send them a prompt they provide a diff
+      - others have an API with specific fields they want
+      - Agus: tradeoff of value protocol brings vs constraints it imposes
+      - can we get an idea from the providers on what they would expect from an API?
+- Update about the governance, @ignatov, @benbrandt
+
+## Action items
+
+- [ ] [@all] everyone take a look at the auth RFD and give comments
+- [ ] [@ignatov] will gather feedback from providers, @jetbrains + @zed EP teams will give input too