Settings

The Settings page uses a left nav + content area layout. After changing options, click Save Settings in the top-right corner to persist changes locally.

Settings Tabs

Tab	Description
General	Theme, log level, body/WebSocket capture limits, workspace paths
Proxy	Listen port, bind address, upstream proxy
Browser	Default browser, custom path, incognito, CDP debug port
Certificate	CA path, export/regenerate, GM SPKI display
About	App version

Where to configure AI

Model, providers, MCP, Engram, and Skills live under AI window → Settings (see AI Features). The main Settings page has no AI tab.

Workspace paths

Under General → Local Workspace, view and open the app data root, config file, database, certificates, reports, and plugin directories.

General

Option	Description
Theme	Dark / Light / System / Custom (presets and color tuning)
Log level	Info or Debug
Max request body capture	MB, default 10
Max WebSocket message capture	MB, default 1

Proxy

Option	Default	Description
Proxy port	8080	Local MITM listen port
Listen address	127.0.0.1	Bind IP
Upstream proxy	empty	Optional, e.g. http://proxy.corp:8080

Domain filtering

Domain exclude rules are configured in the Forwarder filter bar, not in Settings.

Browser

Option	Description
Default browser	Auto-detected Chrome / Firefox / Edge, etc.
Custom browser path	Override executable path
Incognito / private mode	Add privacy launch flags
Auto-clean profile on exit	Remove the current project's browser profile before each launch (browser-profile/{projectId}/); other projects are unaffected
Extra launch args	e.g. --window-size=1280,720
CDP remote debug port	Browser Agent entry for Chromium, often 9222

Launch Browser in the title bar injects proxy and certificate-related flags automatically.

Certificate

Action	Description
View CA path	Generated on first proxy start
Export root CA	Export for OS/browser trust store
Regenerate CA	New key pair; reinstall required
CA key password	Optional, stored locally
GM SPKI fingerprint	SM3 display only; signing algorithm unchanged

Platform install guides are shown on the Certificate tab.

AI configuration

These sections map to AI window → Settings navigation (shared local config with main Settings).

Chat Model

Option	Description
Active Provider	Provider used for chat
Model	Pick from list or type a custom model name
Max Tokens	Max tokens per reply
Temperature	Sampling temperature 0–2
Tool round limit	Max consecutive tool calls per user message
Context window preview	Click Re-resolve for current model context limit

Extensions (modals)

Click a card to open a modal. Changes apply to the current draft immediately; click Save Settings to persist.

Card	Contents
Provider	API key, base URL, model list, default context tokens
Engram memory	Cross-session recall, summaries, enable switch
MCP Servers	Stdio / HTTP-SSE transport, reconnect status
Dynamic Tools	Shell subprocess tools, parameter schema
AI Agent	Personas and system prompts
Skills	SKILL.md packs and tool scope
Advanced	Tool permissions, context budget, Ollama /api/show, Python sandbox

MCP example (Stdio)

In the MCP Servers modal:

Field	Example
Name	filesystem
Transport	Stdio
Command	npx
Args	-y, @modelcontextprotocol/server-filesystem, /path/to/dir

Save, then Reconnect in the modal to verify connected servers and tool counts.

Config file

Advanced options are stored in the app data directory and maintained when you save settings. Manual editing is not recommended. For bulk deployment or enterprise customization, contact support.

Troubleshooting

Settings not applied

1. Confirm Save Settings was clicked
2. Restart and set log level to Debug under General

Certificate / HTTPS capture issues

1. Install CA into the trusted root store
2. Re-export after CA regeneration
3. Firefox uses a separate certificate store

AI connection failures

1. Check API key and base URL under AI window → Settings → Provider
2. Reconnect MCP servers and verify status
3. Confirm you clicked Save Settings