copilot-relay 0.3.10 → 0.4.1

package/README.md

@@ -13,12 +13,11 @@
 - **WebSearch that fits the conversation** — relay-managed search, followed by an answer that can still use your tools.
 - **Your models, your settings** — configurable GPT/Opus routes and reasoning effort; existing selections survive upgrades.
 - **Use the advertised capacity** — discovered context/output limits, without silently shortening your input.
-- **Know what works** — model discovery and opt-in per-model deep checks with clear results.*
+- **Know what works** — model discovery and opt-in per-model deep checks with clear results.
+- **Choose Claude's upstream protocol** — `claudeUpstreamApi: chat-completions` stays the default; opt into `auto` or `messages` for native Claude transport.
+- **Diagnose offline** — debug mode captures full observed bodies for `copilot-relay replay`; captures contain unredacted prompts and must never be shared wholesale.
 - **Run it your way** — foreground CLI or background service on macOS, Windows, and Linux.
 
-\* Model discovery, deep checks, and the Opus 5.5 fresh-install default require
-**v0.3.10 or later**.
-
 ## Quick start
 
 You need **Node.js 22+**, **Claude Code**, and a GitHub account with Copilot access
@@ -42,7 +41,7 @@ Configuration lives in `~/.copilot-relay/config.yaml`. Model access depends on y
 account and organization policy; if startup rejects a model, choose an available
 one using the [configuration guide](wiki/EN-Configuration.md).
 
-Discover models and optionally test one (v0.3.10+):
+Discover models and optionally test one:
 
 ```sh
 copilot-relay models

package/config.default.yaml

@@ -2,8 +2,8 @@
 #
 # This file is copied to ~/.copilot-relay/config.yaml on first start.
 # Edit the user config file, not this template, for local changes.
-# Most values hot-reload while the service is running. host and port still
-# require restart because the listening socket is already bound.
+# Valid complete edits hot-reload without rewriting this file.
+# host, port, and claudeSetup require restart.
 #
 # Values here apply to a fresh install only. copilot-relay writes the resolved
 # config back to disk, so once a key exists in your config.yaml it is yours:
@@ -26,13 +26,12 @@ claudeSetup: true
 
 # Log verbosity:
 #   error - startup/preflight/request failures only
-#   info  - error logs plus startup/preflight status, model/effort summaries,
-#           and local HTTP status codes
-#   debug - info logs plus detailed Copilot timings and request payloads
+#   info  - errors plus status, model/effort and semantic completion summaries
+#   debug - info plus full private bodies in ~/.copilot-relay/captures; may contain secrets
 # Any other value is invalid and stops startup.
 logLevel: info
 
-# Number of days to keep files in ~/.copilot-relay/logs.
+# Number of days to keep relay logs and settled debug captures.
 # Counted in local calendar days: 3 keeps today plus the previous two days.
 # The active log file is copilot-relay.<local-date>.log and rotates at local
 # midnight, which is what makes this setting take effect at all.
@@ -52,6 +51,11 @@ upstreamTimeoutSeconds: 180
 # Leave empty to use gptModel.
 webSearchBackend:
 
+# Claude upstream protocol: auto selects advertised native Messages support.
+# Keep the proven chat route until native cache non-regression is measured.
+# Explicit alternatives: auto, messages.
+claudeUpstreamApi: chat-completions
+
 # Model routing: requests containing "opus" use opusModel; all others use gptModel.
 #
 # Upstream Copilot model used for non-Opus requests.