skills.rest
yyy0107
yyy0107COMMUNITY
Explore all skills by this author →
GitHub

workbench-settings

Inspect and update Pi Workbench preferences and Pi agent settings through the workbench_settings tool.

2|Updated Aug 17, 2026
Search Tags:#system prompt#configuration#settings#preferences#compaction#appearance#pi workbench
One-click install
npx skills add https://github.com/yyy0107/pi-workbench --skill workbench-settings-yyy0107
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: workbench-settings
Source: https://github.com/yyy0107/pi-workbench/tree/main/packages/agent-runtime/runtimes/pi/server/src/internal-skills/workbench-settings
Command: npx skills add https://github.com/yyy0107/pi-workbench --skill workbench-settings-yyy0107

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Changing Pi Workbench preferences or Pi agent settings normally requires hunting through UI menus or editing configuration files by hand. This Skill lets the agent read and modify appearance, language, tool switches, system prompts, and compaction settings directly through the workbench_settings tool, with correct scoping, merging, and revision handling. ## Core Features & Use Cases - Workbench preference management: Read and update UI preferences such as locale, theme, fonts, tool switches, sidebar behavior, and model-picker memory via the workbench domain. - Pi agent settings: Manage user- or project-scoped system prompts, additional instructions, and automatic compaction with revision-checked writes via the pi domain. - Safe update semantics: Merge object preferences instead of overwriting, require expectedRevision for Pi writes, and handle truncated outputs and conflict retries correctly. - Use Case: A user asks to switch the interface to dark mode and Chinese. The Skill reads the current appearance object, merges the colorMode change, and updates locale in one confirmed operation. ## Quick Start Ask the agent to switch the Workbench interface to dark mode and set the language to Simplified Chinese using the workbench settings tool.

Frequently Asked Questions about workbench-settings

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I change Pi Workbench appearance settings like theme or font size?▼

Use the workbench_settings tool with the workbench domain. Read the current appearance object with a describe action, merge your change such as colorMode or uiFontSize into it, and send the merged object in an update patch.

How do I set a custom system prompt for the Pi agent?▼

Use workbench_settings with domain pi and patch systemPrompt or appendSystemPrompt, up to 500,000 characters. Every Pi update requires the expectedRevision from a prior describe of the same scope, and changes apply in a new session or after reload.

What is the difference between user scope and project scope for Pi settings?▼

User scope targets the Runtime's Pi user directory and is the default. Project scope targets the registered workspace matching the current conversation and writes that project's .pi directory; a missing workspace is an error, never a fallback to user scope.

Can this tool change providers, credentials, or model selection?▼

No. Providers, credentials, model selection, skills, packages, and trust decisions are outside this tool's scope. Those tasks belong to pi-docs and the corresponding host services; the modelSelector preference only changes the UI's remembered picker choice.

Why did my Pi settings update fail with a stale revision error?▼

Pi writes use optimistic concurrency: the expectedRevision must match the latest read of the same scope. If it is stale, read again, reconcile the requested change with the new content, and retry with the fresh revision rather than dropping it.

What happens when a settings describe result is too large?▼

Results exceeding 50 KiB or 2,000 lines return truncated true with a fullOutputPath to a temporary file containing the complete JSON. Read relevant sections with offset and limit instead of dumping the whole file into context.

Related Skills

Explore More Software Engineering Skills→
ivanrvpereiraivanrvpereira
17

pi-model-config

Configure Pi model defaults, session overrides, and cycling lists.

Community
Intermediate
bugabingabugabinga

pi

Provide grounded Pi reference guidance for coding agents editing extensions and settings.

Community
Intermediate
shiwenwenshiwenwen
1.4k

ha-settings

Read and update Hope Agent settings through conversational workflows.

Community
Advanced
Cogni-AI-OUCogni-AI-OU
3

datadog-agent

Configure Databog Agent deployments and custom checks on Linux via Ansible and Python.

Official
Advanced