Change default serve port from 3100 to 3200

Port 3100 conflicts with Loki (Grafana log aggregation) which uses it
as its default. Updated src, vite proxy, tests, and all documentation.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: Chris Shellenbarger

CHANGELOG.md

@@ -15,7 +15,7 @@ The harness now has a dual-interface architecture: the original terminal TUI for
 - 55 unit tests covering frontmatter parsing, tool filtering, and agent context
 
 **Phase 2: Serve Mode Backend**
-- `--serve` starts a Fastify HTTP/WebSocket server (default port 3100)
+- `--serve` starts a Fastify HTTP/WebSocket server (default port 3200)
 - REST API: agent roster, session CRUD, usage tracking
 - WebSocket: real-time streaming (text tokens, thinking tokens, tool calls, sub-agent progress)
 - Token-based authentication via `~/.mastersof-ai/access.yaml` (SHA-256 hashed tokens)

CLAUDE.md

@@ -31,7 +31,7 @@ Both share agent loading, tools, sub-agents, sessions, and memory.
 ```bash
 npx tsx bin/mastersof-ai.js                         # TUI with default agent
 npx tsx bin/mastersof-ai.js --agent researcher      # TUI with specific agent
-npx tsx bin/mastersof-ai.js --serve                 # Web UI server on port 3100
+npx tsx bin/mastersof-ai.js --serve                 # Web UI server on port 3200
 npx tsx bin/mastersof-ai.js --serve --port 5000     # Web UI on custom port
 npx tsx bin/mastersof-ai.js --agent researcher --sandbox  # Bubblewrap sandbox
 npx tsx bin/mastersof-ai.js --card                  # Output Agent Card JSON

DESIGN.md

@@ -4,7 +4,7 @@ Agent runtime with full system prompt control. Two interfaces, one core.
 
 ```
 Terminal TUI:  mastersof-ai [--agent x]           Single user, local, React/Ink
-Web UI:        mastersof-ai --serve [--port 3100] Multi-user, token auth, Fastify + React SPA
+Web UI:        mastersof-ai --serve [--port 3200] Multi-user, token auth, Fastify + React SPA
 ```
 
 Both share: agent loading, IDENTITY.md parsing, Claude Agent SDK, MCP tools, sub-agents, sessions, memory.
@@ -61,7 +61,7 @@ Detailed documentation for each area:
 ```bash
 npx tsx bin/mastersof-ai.js                         # TUI with default agent
 npx tsx bin/mastersof-ai.js --agent analyst          # TUI with specific agent
-npx tsx bin/mastersof-ai.js --serve                  # Web server on port 3100
+npx tsx bin/mastersof-ai.js --serve                  # Web server on port 3200
 npx tsx bin/mastersof-ai.js --serve --port 5000      # Custom port
 npx tsx bin/mastersof-ai.js --agent analyst --sandbox # Bubblewrap sandbox
 npx tsx bin/mastersof-ai.js --card                   # Output Agent Card JSON

README.md

@@ -123,7 +123,7 @@ mastersof-ai --list-agents            # show available agents
 ### Web UI
 
 ```bash
-mastersof-ai --serve                  # start on port 3100
+mastersof-ai --serve                  # start on port 3200
 mastersof-ai --serve --port 5000      # custom port
 ```
 

docs/configuration.md

@@ -294,7 +294,7 @@ mastersof-ai --resume [name|#N]        Resume a session (TUI)
 mastersof-ai --sandbox                 Run in bubblewrap sandbox
 mastersof-ai --list-agents             Show available agents with metadata
 mastersof-ai --init                    Force first-run setup
-mastersof-ai --serve                   Start web UI server (port 3100)
+mastersof-ai --serve                   Start web UI server (port 3200)
 mastersof-ai --serve --port 5000       Custom port
 mastersof-ai --serve --host 127.0.0.1  Bind to specific host
 mastersof-ai --card                    Output A2A Agent Card JSON and exit

docs/design-decisions.md

@@ -24,7 +24,7 @@
 | Separate frontend SPA | `web/` dir, Vite + React, deploys independently | Frontend on CDN (Cloudflare Pages), backend on VPS — decoupled scaling |
 | Token auth via access.yaml | SHA-256 hashed tokens, constant-time comparison | Simple, no external auth service, hot-reloadable, secure at rest |
 | Per-user session isolation | Sessions stored under `sessions/{user}/` | No cross-user data leakage, supports per-user cost caps |
-| Default port 3100 for serve | Not 3000 (common conflict) | Avoids collision with Vite dev server and other common services |
+| Default port 3200 for serve | Not 3000 (common conflict) | Avoids collision with Vite dev server and other common services |
 
 ## Agent Definition
 

src/index.tsx

@@ -198,7 +198,7 @@ if (getFlag("card")) {
 // --- Flag: --serve (server mode) ---
 
 if (getFlag("serve")) {
-  const port = parseInt(getFlagValue("port") ?? "3100", 10);
+  const port = parseInt(getFlagValue("port") ?? "3200", 10);
   const host = getFlagValue("host") ?? "0.0.0.0";
 
   const { loadAccessConfig } = await import("./access.js");

tests/test-ember-flow.mjs

@@ -1,11 +1,11 @@
 // Ember conversation flow test — tests multi-turn tool use
 // Usage: node web/test-ember-flow.mjs
-// Requires: backend running on :3100
+// Requires: backend running on :3200
 
 import { WebSocket } from "ws";
 
 const TOKEN = "b3f88d0b-e226-486e-b0f6-80a5b98535cc";
-const WS_URL = `ws://localhost:3100/ws?token=${TOKEN}`;
+const WS_URL = `ws://localhost:3200/ws?token=${TOKEN}`;
 const TIMEOUT = 120_000; // 2 minutes — tool use can be slow
 
 function connectAndSubscribe() {

tests/test-ember-tools.mjs

@@ -4,7 +4,7 @@
 import { WebSocket } from "ws";
 
 const TOKEN = "b3f88d0b-e226-486e-b0f6-80a5b98535cc";
-const WS_URL = `ws://localhost:3100/ws?token=${TOKEN}`;
+const WS_URL = `ws://localhost:3200/ws?token=${TOKEN}`;
 
 function connect(sessionId) {
   return new Promise((resolve, reject) => {

tests/test-integration.sh

@@ -1,12 +1,12 @@
 #!/usr/bin/env bash
 # Integration test: validates web frontend API calls against the running backend.
-# Requires: backend running on :3100 (mastersof-ai --serve)
+# Requires: backend running on :3200 (mastersof-ai --serve)
 #
 # Usage: bash web/test-integration.sh
 
 set -uo pipefail
 
-BACKEND="http://localhost:3100"
+BACKEND="http://localhost:3200"
 VALID_TOKEN="b3f88d0b-e226-486e-b0f6-80a5b98535cc"
 INVALID_TOKEN="bad-token-12345"
 PASS=0