client-compatibility

Detects Copilot platform and adapts agent spawning for CLI, VS Code, and fallback surfaces.

Updated Jun 2, 2026
One-click install
npx skills add https://github.com/codebytes/btt --skill client-compatibility-codebytes
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: client-compatibility
Source: https://github.com/codebytes/btt/tree/main/.squad/templates/skills/client-compatibility
Command: npx skills add https://github.com/codebytes/btt --skill client-compatibility-codebytes

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Multi-agent orchestration logic breaks when the same coordinator runs on different Copilot surfaces, because spawning tools, model selection, and SQL availability differ between CLI, VS Code, JetBrains, and GitHub.com. ## Core Features & Use Cases - Platform Detection: Identifies the runtime surface by checking for the task tool (CLI), runSubagent/agent (VS Code), or neither (inline fallback). - Adaptive Spawning: Adjusts parameters per platform, dropping agent_type, mode, model, and description on VS Code while batching parallel subagents in a single turn. - Feature Degradation Guidance: Provides a table mapping CLI-only features (SQL tool, per-spawn model selection, fire-and-forget Scribe) to their cross-platform equivalents. - Use Case: A coordinator spawning three parallel agents detects VS Code mode, issues three runSubagent calls in one turn with Scribe last, and skips read_agent polling entirely. ## Quick Start Detect the current Copilot platform and adapt the agent spawning strategy for this session.

Frequently Asked Questions about client-compatibility

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

FAQPage Schema
How do I detect which Copilot platform my agent is running on?

Check which tools are available: the `task` tool indicates CLI mode, `runSubagent` or `agent` indicates VS Code mode, and neither means fallback inline mode. If both `task` and `runSubagent` exist, prefer `task` for its richer parameters.

How do I spawn parallel subagents in VS Code?

Issue multiple `runSubagent` calls in a single turn with the full prompt as the only parameter. They run concurrently automatically, replacing CLI's background mode plus `read_agent` polling, and results return with the response.

Does the SQL tool work on VS Code or GitHub.com?

No, the `sql` tool is CLI-only and silently fails on VS Code, JetBrains, and GitHub.com. Cross-platform workflows must use filesystem-based state in `.squad/` files instead of SQL for todo tracking or session state.

Can I select a specific model per subagent on VS Code?

No, per-spawn model selection only works on CLI. On VS Code, all subagents use the session model chosen in the model picker, so accept it and log the original intent rather than attempting fallback chains.

Why should Scribe be the last subagent in a VS Code parallel group?

VS Code cannot fire-and-forget, so Scribe blocks until completion. Since Scribe only performs light file operations, batching it last in a parallel group keeps the blocking tolerable while preserving result ordering.