| 1 | --- |
| 2 | name: "client-compatibility" |
| 3 | description: "Platform detection and adaptive spawning for CLI vs VS Code vs other surfaces" |
| 4 | domain: "orchestration" |
| 5 | confidence: "high" |
| 6 | source: "extracted" |
| 7 | --- |
| 8 | |
| 9 | ## Context |
| 10 | |
| 11 | Squad runs on multiple Copilot surfaces (CLI, VS Code, JetBrains, GitHub.com). The coordinator must detect its platform and adapt spawning behavior accordingly. Different tools are available on different platforms, requiring conditional logic for agent spawning, SQL usage, and response timing. |
| 12 | |
| 13 | ## Patterns |
| 14 | |
| 15 | ### Platform Detection |
| 16 | |
| 17 | Before spawning agents, determine the platform by checking available tools: |
| 18 | |
| 19 | 1. **CLI mode** — `task` tool is available → full spawning control. Use `task` with `agent_type`, `mode`, `model`, `description`, `prompt` parameters. Collect results via `read_agent`. |
| 20 | |
| 21 | 2. **VS Code mode** — `runSubagent` or `agent` tool is available → conditional behavior. Use `runSubagent` with the task prompt. Drop `agent_type`, `mode`, and `model` parameters. Multiple subagents in one turn run concurrently (equivalent to background mode). Results return automatically — no `read_agent` needed. |
| 22 | |
| 23 | 3. **Fallback mode** — neither `task` nor `runSubagent`/`agent` available → work inline. Do not apologize or explain the limitation. Execute the task directly. |
| 24 | |
| 25 | If both `task` and `runSubagent` are available, prefer `task` (richer parameter surface). |
| 26 | |
| 27 | ### VS Code Spawn Adaptations |
| 28 | |
| 29 | When in VS Code mode, the coordinator changes behavior in these ways: |
| 30 | |
| 31 | - **Spawning tool:** Use `runSubagent` instead of `task`. The prompt is the only required parameter — pass the full agent prompt (charter, identity, task, hygiene, response order) exactly as you would on CLI. |
| 32 | - **Parallelism:** Spawn ALL concurrent agents in a SINGLE turn. They run in parallel automatically. This replaces `mode: "background"` + `read_agent` polling. |
| 33 | - **Model selection:** Accept the session model. Do NOT attempt per-spawn model selection or fallback chains — they only work on CLI. In Phase 1, all subagents use whatever model the user selected in VS Code's model picker. |
| 34 | - **Scribe:** Cannot fire-and-forget. Batch Scribe as the LAST subagent in any parallel group. Scribe is light work (file ops only), so the blocking is tolerable. |
| 35 | - **Launch table:** Skip it. Results arrive with the response, not separately. By the time the coordinator speaks, the work is already done. |
| 36 | - **`read_agent`:** Skip entirely. Results return automatically when subagents complete. |
| 37 | - **`agent_type`:** Drop it. All VS Code subagents have full tool access by default. Subagents inherit the parent's tools. |
| 38 | - **`description`:** Drop it. The agent name is already in the prompt. |
| 39 | - **Prompt content:** Keep ALL prompt structure — charter, identity, task, hygiene, response order blocks are surface-independent. |
| 40 | |
| 41 | ### Feature Degradation Table |
| 42 | |
| 43 | | Feature | CLI | VS Code | Degradation | |
| 44 | |---------|-----|---------|-------------| |
| 45 | | Parallel fan-out | `mode: "background"` + `read_agent` | Multiple subagents in one turn | None — equivalent concurrency | |
| 46 | | Model selection | Per-spawn `model` param (4-layer hierarchy) | Session model only (Phase 1) | Accept session model, log intent | |
| 47 | | Scribe fire-and-forget | Background, never read | Sync, must wait | Batch with last parallel group | |
| 48 | | Launch table UX | Show table → results later | Skip table → results with response | UX only — results are correct | |
| 49 | | SQL tool | Available | Not available | Avoid SQL in cross-platform code paths | |
| 50 | | Response order bug | Critical workaround | Possibly necessary (unverified) | Keep the block — harmless if unnecessary | |
| 51 | |
| 52 | ### SQL Tool Caveat |
| 53 | |
| 54 | The `sql` tool is **CLI-only**. It does not exist on VS Code, JetBrains, or GitHub.com. Any coordinator logic or agent workflow that depends on SQL (todo tracking, batch processing, session state) will silently fail on non-CLI surfaces. Cross-platform code paths must not depend on SQL. Use filesystem-based state (`.squad/` files) for anything that must work everywhere. |
| 55 | |
| 56 | ## Examples |
| 57 | |
| 58 | **Example 1: CLI parallel spawn** |
| 59 | ```typescript |
| 60 | // Coordinator detects task tool available → CLI mode |
| 61 | task({ agent_type: "general-purpose", mode: "background", model: "claude-sonnet-4.5", ... }) |
| 62 | task({ agent_type: "general-purpose", mode: "background", model: "claude-haiku-4.5", ... }) |
| 63 | // Later: read_agent for both |
| 64 | ``` |
| 65 | |
| 66 | **Example 2: VS Code parallel spawn** |
| 67 | ```typescript |
| 68 | // Coordinator detects runSubagent available → VS Code mode |
| 69 | runSubagent({ prompt: "...Fenster charter + task..." }) |
| 70 | runSubagent({ prompt: "...Hockney charter + task..." }) |
| 71 | runSubagent({ prompt: "...Scribe charter + task..." }) // Last in group |
| 72 | // Results return automatically, no read_agent |
| 73 | ``` |
| 74 | |
| 75 | **Example 3: Fallback mode** |
| 76 | ```typescript |
| 77 | // Neither task nor runSubagent available → work inline |
| 78 | // Coordinator executes the task directly without spawning |
| 79 | ``` |
| 80 | |
| 81 | ## Anti-Patterns |
| 82 | |
| 83 | - ❌ Using SQL tool in cross-platform workflows (breaks on VS Code/JetBrains/GitHub.com) |
| 84 | - ❌ Attempting per-spawn model selection on VS Code (Phase 1 — only session model works) |
| 85 | - ❌ Fire-and-forget Scribe on VS Code (must batch as last subagent) |
| 86 | - ❌ Showing launch table on VS Code (results already inline) |
| 87 | - ❌ Apologizing or explaining platform limitations to the user |
| 88 | - ❌ Using `task` when only `runSubagent` is available |
| 89 | - ❌ Dropping prompt structure (charter/identity/task) on non-CLI platforms |