TheBotique

WorkBuddy 自定义模型能力位排查(读图/工具/推理)

排查并修复 WorkBuddy 里自定义模型「不支持读图 / 不调用工具 / 不能推理」的能力位标错问题,以及新增自定义模型(OpenCode Zen、自建或其他厂商的 OpenAI 兼容端点)时一次性探测能力位并生成配置。当用户说「为什么说不支持读图」「模型读不了图片」「附件发不进去」「custom-local 模型能力不对」,或问「加新模型要测什么」「怎么知道这个模型支持读图/思考强度」时使用。含通用探测器 probe_model.py(一条命令测出读图/工具调用/effort 档位)、能力位定位法、直连端点自证法、以及「改完必须重启」这一必踩的坑。

as observed 2026-11-19T03:34:19.311Z
Identifier
workbuddy-model-capability-flags
Source
ClawHub
Version observed
1.0.1
Source repository
not published
Repository observation
No source repository listed
First observed here
2026-10-05T10:33:23.015Z
Observations recorded
2
Installs (reported upstream)
1
Weekly downloads (upstream)
365
Declared license
MIT-0

Observation history

2026-11-19T03:34:19.311Z

Fields that differed: changelog latestVersion topics

FieldBeforeAfter
changelog "首次发布;API key 改读环境变量,探针图缓存移到系统临时目录" "v1.0.1:按简介重制专属图标(纯符号)"
latestVersion "1.0.0" "1.0.1"
topics ["model-capabilities","tool-calling","vision"] ["capability","flags","llm","model","workbuddy"]

Correction

If you maintain this extension and believe anything above is inaccurate, request a correction. Corrections are published, and disputed entries are marked as disputed while under review.