feat(contrib/dashboard): DUODUO.md manifest, scripts/, and sample descriptor

[6m1w]

Summary

Follow-up to PR #25, addressing antmanler's review feedback:

• DUODUO.md with full YAML frontmatter manifest (name, type, version, requires, installs, scripts) — ready for future duoduo contrib install tooling
• scripts/ subdirectory — organized helper scripts:
  • sync-channels.sh: cron wrapper for automated channel refresh
  • fill-empty-descriptors.py: populates stub descriptors from template when bot joins new groups
  • refresh-channel-map.py and save-api.py moved into scripts/
• config/sample-descriptor.md — default channel descriptor template with participation rules framework

Design notes

• Uses DUODUO.md (not CLAUDE.md) per antmanler's guidance to avoid prompt injection
• YAML frontmatter is program-parseable for future CLI tooling
• Markdown body is agent-readable for self-service installation
• All scripts are standalone with proper shebangs
• Original flat files (refresh-channel-map.py, save-api.py) can be removed from contrib/dashboard/ root after merge

Test plan

• Verify DUODUO.md YAML frontmatter parses correctly
• Verify scripts/sync-channels.sh runs refresh + fill-empty-descriptors
• Verify config/sample-descriptor.md has correct placeholder format

🤖 Generated with Claude Code

[antmanler]

Thanks for the follow-up! Good instincts on organizing scripts and adding the sample descriptor. A few things need to be addressed before we can merge.

Must fix

1. Hardcoded paths

All scripts hardcode ~/ENG/openduo/openduo. Extensions must discover paths at runtime via the system.runtime.info RPC, which returns runtime_dir and kernel_dir. Example:

import json, urllib.request
def get_runtime_paths(daemon_url="http://localhost:20233"):
    req = urllib.request.Request(f"{daemon_url}/rpc",
        data=json.dumps({"jsonrpc":"2.0","id":1,"method":"system.runtime.info"}).encode(),
        headers={"Content-Type": "application/json"})
    with urllib.request.urlopen(req) as r:
        result = json.loads(r.read())["result"]
    return result["runtime_dir"], result["kernel_dir"]

Same for dashboard.html — loadFile("/home/duoduo/aladuo/config/sample-descriptor.md") must use a path derived from the RPC response, not hardcoded.

See contrib/README.md for the full convention.

2. Rebase on current main

Your branch is based on an older version. The current main already has:

• shortActor(k, evt) with sessionDisplayNames map, display_name support, and subconscious partition name resolution (sub:<name>)
• Signal bar tooltips showing display_name and last_error

Your PR reverts all of these back to the old shortActor(k). Please rebase on current main and preserve these changes.

3. Title "WATU" → "Duoduo"

The dashboard title and brand name must stay "Duoduo", not "WATU".

4. Remove hardcoded entity skip

if (ent.file === "antmanler.md" || ent.file === "feishu-user-primary.md") return;

This is instance-specific. Remove it.

5. No sudo in scripts

fill-empty-descriptors.py uses os.system(f'sudo cp ...') — this is a security concern and won't work in most environments. Use standard file I/O. If there's a permissions issue, document it as a prerequisite rather than working around it with sudo.

Frontmatter

We've finalized the contrib frontmatter convention (see updated contrib/README.md). Please simplify to:

---
name: dashboard-channels-entities
type: dashboard
version: 0.1.0
author: 6m1w
description: Enhanced dashboard with Channels and Entities tabs
requires:
  - python3
  - feishu-credentials
---

The installs and scripts blocks should move into the Markdown body — the agent reads the body for installation steps.

New files

The scripts/ reorganization and config/sample-descriptor.md are good additions. Just fix the path hardcoding and they're ready.

Looking forward to the update!