DevAgentRadar → Repo Harness Rule Patch

Goal: update our repository's permanent harness instructions based on this release.

Update the file that tool actually reads. Do not dump everything into one file.

• Claude Code → CLAUDE.md plus .claude/ (skills, hooks, settings, agents, commands). It does not read AGENTS.md natively; start CLAUDE.md with @AGENTS.md.
• Codex, Copilot, Cursor, Factory Droid, Grok Build, Roo Code, Goose, OpenCode, Amp, Zed, Aider → AGENTS.md.
• Gemini CLI / Antigravity → GEMINI.md. AGENTS.md only if context.fileName is set.
• Cursor glob-scoped rules → .cursor/rules/*.mdc (plain .md is ignored), not a second constitution.
• Codex MCP → .codex/config.toml, not .mcp.json.
• Copilot extra instructions → .github/copilot-instructions.md; custom agents → .github/agents/.
• Windsurf → .windsurf/rules (do not assume AGENTS.md).
• Cline → .clinerules.
• Procedures → a skill (SKILL.md). Enforcement the model must not skip → a hook. Isolated roles → subagents.
• Do not fork the same rule into three tool files.

Context

Assistant: Qwen Code Proposal: Persist Qwen daemon tokens with qwen daemon start --save-token Summary: Qwen Code v0.26.1 restores breaking qqbot per-group isolation and adds opt-in owner-only daemon tokens via qwen daemon start --save-token. Use idle-only Managed context compression, update MCP without recreating sessions, and keep default public subagent delegation. Primary source: https://github.com/QwenLM/qwen-code/releases/tag/v0.26.1

Why it matters

v0.26.1 restores per-group qqbot session isolation under thread scope (breaking). Daemon auth is now opt-in persistent: qwen daemon start --save-token writes the bearer token to an owner-only file, qwen daemon token manages it, and daemon stop removes it. Workspace-bound Managed Sessions get idle-only manual context compression via the public API and WebShell context-usage popover while keeping audit history. MCP servers can be changed without recreating sessions; public subagent delegation is on by default; max-sessions persists in settings.json. WebShell exposes model output language in settings, sidebar workspace reordering, and a unified recent-sessions list. OpenTUI input, drafts, and /statusline now match Ink. Adopt these as the local/Managed path so tokens are not left world-readable, qqbot groups do not share thread state, and sessions are not churned to change MCP or compress context.

Suggested actions

1. qwen daemon start --save-token
2. qwen daemon token
3. After this breaking qqbot fix, keep one session per group under thread scope; do not reuse or merge conversation state across groups in the same thread.
4. When a Workspace-bound Managed Session is idle, compress context via the public API or the WebShell context-usage popover so audit history is preserved; do not compress while the session is active.
5. On existing Managed sessions, add or change MCP servers in place (do not recreate the session). Leave public subagent delegation enabled (now default). Persist max-sessions in settings.json and set model output language in the WebShell settings panel.

Config surfaces this release may change

• settings files — check your repo before applying

After you finish

Do not report this as applied to DevAgentRadar. You cannot write the visitor's loop.

Tell the human: open https://devagentradar.com/proposals/qwen-code-v0-26-1-persist-qwen-daemon-tokens-with-qwen-daemon-start-save and mark Applied, Skipped, or Failed. Proposal id: d1ffe96b-38e1-436c-9b97-984e94847ae2

Instructions:

1. Read the existing context files and settings for the tools in this repo.
2. Draft an explicit Git diff. Update ONLY agent harness rules or tool configuration.
3. Call out deprecated flags, obsolete habits, or changed permission boundaries.
4. Do not touch application logic.

Start with the proposed diff and say which layer and which tool's file it belongs in.