feat(byoa): run agents on Qwen Code

docs/BYOA.md

@@ -1,4 +1,4 @@
-# BYOA — Bring Your Own Agent (local Claude Code / Codex / Grok Build / Cursor Agent / OpenCode / pi / Gemini CLI as the engine)
+# BYOA — Bring Your Own Agent (local Claude Code / Codex / Grok Build / Cursor Agent / OpenCode / pi / Gemini CLI / Qwen Code as the engine)
 
 Every Cumora agent has a "brain" and a host. The managed path is
 server-side: `runAgentTurn` in `server/src/agents/turn.ts` runs a
@@ -8,7 +8,7 @@ a per-agent Kubernetes pod (the `agent-computer` image).
 **BYOA** lets a user supply the brain instead: a long-running daemon on
 the user's own machine (laptop **or** VPS) drives a local **Claude Code**,
 **Codex CLI**, **Grok Build** (`grok`), **Cursor Agent** (`cursor-agent`),
-**OpenCode** (`opencode`), **pi** (`pi`), or **Gemini CLI** (`gemini`) as the reasoning engine, on the user's own provider
+**OpenCode** (`opencode`), **pi** (`pi`), **Gemini CLI** (`gemini`), or **Qwen Code** (`qwen`) as the reasoning engine, on the user's own provider
 account — the server never holds the user's provider credentials.
 One daemon hosts **many independent agents** — each with its own dedicated
 home directory, memory, skills, and notes. In Cumora these still appear
@@ -45,7 +45,7 @@ managed cloud agents and local agents into the same picture.
   user to set up; it's always online.
 - **Your computers** — machines you pair (your Mac, a VPS). Each runs the
   `cumora agent computer` daemon with a local engine (Claude Code /
-  Codex / Grok Build / Cursor Agent / OpenCode / pi / Gemini CLI). Agents you place here are BYOA agents.
+  Codex / Grok Build / Cursor Agent / OpenCode / pi / Gemini CLI / Qwen Code). Agents you place here are BYOA agents.
 
 ```
 Computers
@@ -194,13 +194,13 @@ from their own agenda — Kanban cards and due calendar slots — via
 ## Engine integration
 
 `server/src/agents/computer/engine.ts` defines one `EngineAdapter` per
-engine (`claude`, `codex`, `grok`, `cursor`, `opencode`, `gemini`). Persistent per-agent
+engine (`claude`, `codex`, `grok`, `cursor`, `opencode`, `gemini`, `qwen`). Persistent per-agent
 sessions are preferred when the CLI exposes one; Cursor and OpenCode use
 one-shot `run()` for every wake and resume the session id reported by the CLI.
 
 ```ts
 interface EngineAdapter {
-  id: 'claude' | 'codex' | 'grok' | 'cursor' | 'opencode' | 'gemini'
+  id: 'claude' | 'codex' | 'grok' | 'cursor' | 'opencode' | 'gemini' | 'qwen'
   seedHome(home, persona)          // lay out CLAUDE.md/AGENTS.md, skills, dirs
   startSession?(args): EngineSession | null   // persistent session (primary)
   run(args): Promise<…>            // one-shot fallback
@@ -215,16 +215,16 @@ interface EngineSession {
 }
 ```
 
-| Concern | Claude Code | Codex CLI | Grok Build | Cursor Agent | OpenCode | pi | Gemini CLI |
-| --- | --- | --- | --- | --- | --- | --- | --- |
-| Secure default | Claude Code ≥ 2.1.248 with `--restricted`; no Bash/PowerShell/web tools; host-home reads, command network, unsandboxed retry, and subprocess credentials are denied | Codex ≥ 0.138.0 with a custom permission profile: only minimal runtime reads plus the agent home; command network disabled; tool env allowlisted; user/project config, rules, hooks, apps, remote plugins, and multi-agent tools ignored or disabled | disabled | disabled | disabled | disabled | disabled |
-| Platform | macOS, Linux, WSL2; native Windows disabled because Claude's sandbox is unsupported there | macOS, Linux, WSL2, native Windows | compatibility opt-in only | compatibility opt-in only | compatibility opt-in only | compatibility opt-in only | compatibility opt-in only |
-| Persistent session | secure default; `claude -p --input-format stream-json --output-format stream-json --verbose` | compatibility opt-in only; secure default uses one-shot `exec` | compatibility opt-in ACP | none | none | compatibility opt-in RPC | none |
-| Standing prompt | `--append-system-prompt-file <home>/.cumora-standing-prompt.md` | inlined into each secure one-shot wake | compatibility ACP `_meta.rules` | inlined | inlined | compatibility `--append-system-prompt` | inlined |
-| One-shot | sandboxed `claude -p … --output-format stream-json` | sandboxed `codex exec --ignore-user-config --ignore-rules …` | compatibility `grok -p … --always-approve` | compatibility `cursor-agent … --force --trust` | compatibility `opencode run … --auto` | compatibility `pi … -p` | compatibility `gemini … --yolo` |
-| Custom argv | ignored securely; requires the compatibility opt-in | ignored securely; requires the compatibility opt-in | compatibility opt-in required | compatibility opt-in required | compatibility opt-in required | compatibility opt-in required | compatibility opt-in required |
-| Memory / persona file | `CLAUDE.md` | `AGENTS.md` | `AGENTS.md` | `AGENTS.md` | `AGENTS.md` plus `.opencode/skills/` | `AGENTS.md` plus `.pi/skills/` (loaded via `--skill`) | `GEMINI.md` plus `.gemini/skills/` |
-| Triage (small brain) | restricted and tool-free | read-only custom profile and tool env | compatibility only | compatibility only | compatibility only | compatibility only | compatibility only |
+| Concern | Claude Code | Codex CLI | Grok Build | Cursor Agent | OpenCode | pi | Gemini CLI | Qwen Code |
+| --- | --- | --- | --- | --- | --- | --- | --- | --- |
+| Secure default | Claude Code ≥ 2.1.248 with `--restricted`; no Bash/PowerShell/web tools; host-home reads, command network, unsandboxed retry, and subprocess credentials are denied | Codex ≥ 0.138.0 with a custom permission profile: only minimal runtime reads plus the agent home; command network disabled; tool env allowlisted; user/project config, rules, hooks, apps, remote plugins, and multi-agent tools ignored or disabled | disabled | disabled | disabled | disabled | disabled | disabled |
+| Platform | macOS, Linux, WSL2; native Windows disabled because Claude's sandbox is unsupported there | macOS, Linux, WSL2, native Windows | compatibility opt-in only | compatibility opt-in only | compatibility opt-in only | compatibility opt-in only | compatibility opt-in only | compatibility opt-in only |
+| Persistent session | secure default; `claude -p --input-format stream-json --output-format stream-json --verbose` | compatibility opt-in only; secure default uses one-shot `exec` | compatibility opt-in ACP | none | none | compatibility opt-in RPC | none | none — 0.22.3 has no stream-json INPUT mode |
+| Standing prompt | `--append-system-prompt-file <home>/.cumora-standing-prompt.md` | inlined into each secure one-shot wake | compatibility ACP `_meta.rules` | inlined | inlined | compatibility `--append-system-prompt` | inlined | inlined |
+| One-shot | sandboxed `claude -p … --output-format stream-json` | sandboxed `codex exec --ignore-user-config --ignore-rules …` | compatibility `grok -p … --always-approve` | compatibility `cursor-agent … --force --trust` | compatibility `opencode run … --auto` | compatibility `pi … -p` | compatibility `gemini … --yolo` | compatibility `qwen --output-format stream-json --yolo` |
+| Custom argv | ignored securely; requires the compatibility opt-in | ignored securely; requires the compatibility opt-in | compatibility opt-in required | compatibility opt-in required | compatibility opt-in required | compatibility opt-in required | compatibility opt-in required | compatibility opt-in required |
+| Memory / persona file | `CLAUDE.md` | `AGENTS.md` | `AGENTS.md` | `AGENTS.md` | `AGENTS.md` plus `.opencode/skills/` | `AGENTS.md` plus `.pi/skills/` (loaded via `--skill`) | `GEMINI.md` plus `.gemini/skills/` | `QWEN.md` plus `.qwen/skills/` |
+| Triage (small brain) | restricted and tool-free | read-only custom profile and tool env | compatibility only | compatibility only | compatibility only | compatibility only | compatibility only | compatibility only |
 
 Sessions carry a resume id (`~/.cumora/sessions/<agentId>.session`); a
 failed resume falls back to a fresh thread instead of wedging the agent.
@@ -332,6 +332,7 @@ CUMORA_ENGINE_MODEL=local CUMORA_TRIAGE_MODEL=local-small cumora agent computer
     .cursor/skills/                 ← Cursor-native skill directory
     .opencode/skills/               ← OpenCode-native skill directory
     .gemini/skills/                 ← Gemini-native skill directory
+    .qwen/skills/                   ← Qwen-native skill directory
     .pi/skills/                     ← pi-native skill directory (via --skill)
     bin/cumora                     ← compatibility mode only
     memory/MEMORY.md               ← the agent's durable memory index
@@ -384,7 +385,7 @@ CREATE TABLE computers (
   owner_user_id     TEXT,            -- null for the managed Cumora Cloud row
   name              TEXT NOT NULL,   -- "Cumora Cloud", "MacBook Pro", …
   kind              TEXT NOT NULL,   -- 'cloud' | 'local' | 'vps'
-  available_engines JSONB,           -- ['claude','codex','grok','cursor','opencode','pi','gemini'] (daemon-detected)
+  available_engines JSONB,           -- ['claude','codex','grok','cursor','opencode','pi','gemini','qwen'] (daemon-detected)
   status            TEXT NOT NULL,   -- 'online' | 'offline' | 'busy'
   last_seen_at      TIMESTAMP,
   credential_hash   TEXT,            -- SHA256 of the device token

scripts/guard-big-brain.mjs

@@ -80,7 +80,7 @@ export function lineViolations(rel, raw) {
     out.push('big-brain engine spawn (adapter.run) outside the gated daemon path')
   }
   // R4 — a supported engine binary spawned directly, outside the adapter.
-  if (/(?:spawn|execFile\w*|exec)\s*\(\s*['"`](claude|codex|grok|cursor-agent|opencode|pi|gemini)['"`]/.test(code) && !ALLOW.engineSpawn.includes(rel)) {
+  if (/(?:spawn|execFile\w*|exec)\s*\(\s*['"`](claude|codex|grok|cursor-agent|opencode|pi|gemini|qwen)['"`]/.test(code) && !ALLOW.engineSpawn.includes(rel)) {
     out.push('engine binary spawned outside the engine adapter (bypasses the classify=small / run=big split)')
   }
   return out