Back to Details

sales-char

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Char Platform Help

Char平台帮助

Step 1 — Gather context

步骤1 — 收集上下文

If 
references/learnings.md
exists, read it first for accumulated platform knowledge.
  1. What do you need help with?
    • A) Installing Char or first-time setup on macOS
    • B) Transcription provider selection (local Cactus vs 10 cloud STT providers)
    • C) LLM configuration (OpenRouter, BYOK, Ollama/LM Studio for offline)
    • D) Speaker identification not working in group meetings
    • E) Building a plugin or using automation hooks
    • F) CLI usage (template management, database operations)
    • G) Comparing Char to other note-takers
    • H) Privacy, telemetry, or GPL licensing questions
  2. Which plan are you on?
    • A) Free (on-device transcription, BYOK)
    • B) Lite ($8/mo — cloud STT + LLM, speaker ID, integrations)
    • C) Pro ($25/mo — advanced sharing, team features, selective sync)
    • D) Team (custom)
    • E) Not sure / evaluating
Skip-ahead rule: if the user's prompt already contains enough context, skip to Step 2.
如果
references/learnings.md
存在，请先阅读该文档以获取累积的平台知识。
  1. 你需要哪方面的帮助？
    • A) 在macOS上安装Char或首次设置
    • B) 转录提供商选择（本地Cactus vs 10款云STT提供商）
    • C) LLM配置（OpenRouter、BYOK、Ollama/LM Studio实现离线）
    • D) 群组会议中说话人识别功能失效
    • E) 构建插件或使用自动化钩子
    • F) CLI使用（模板管理、数据库操作）
    • G) 对比Char与其他笔记工具
    • H) 隐私、遥测或GPL许可相关问题
  2. 你使用的是哪个套餐？
    • A) 免费版（设备端转录、BYOK）
    • B) Lite版（每月8美元——云STT+LLM、说话人识别、集成功能）
    • C) Pro版（每月25美元——高级分享、团队功能、选择性同步）
    • D) 团队版（定制）
    • E) 不确定/正在评估
跳步规则：如果用户的提示已包含足够上下文，直接跳至步骤2。

Step 2 — Route or answer directly

步骤2 — 转派或直接解答

Problem domainRoute to
Comparing Char to other note-takers
/sales-note-taker {user's question}
Reviewing a specific call for coaching feedback
/sales-call-review {user's question}
Building a coaching program
/sales-coaching {user's question}
General tool integration patterns (Zapier, webhooks, iPaaS)
/sales-integration {user's question}
When routing, provide the exact command.
问题领域转派至
对比Char与其他笔记工具
/sales-note-taker {用户问题}
针对特定通话获取教练式反馈复盘
/sales-call-review {用户问题}
构建教练项目
/sales-coaching {用户问题}
通用工具集成模式（Zapier、webhooks、iPaaS）
/sales-integration {用户问题}
转派时，请提供准确的命令。

Step 3 — Char platform reference

步骤3 — Char平台参考资料

Read 
references/platform-guide.md
 for the full platform reference — modules, pricing, STT/LLM providers, plugin SDK, hooks, CLI, integrations.
Answer the user's question using only the relevant section. Don't dump the full reference.
**阅读
references/platform-guide.md
**获取完整平台参考——包括模块、定价、STT/LLM提供商、插件SDK、钩子、CLI、集成功能。
仅使用相关章节解答用户问题，不要直接输出完整参考资料。

Step 4 — Actionable guidance

步骤4 — 可执行指导

You no longer need the platform guide — focus on the user's specific situation.
First-time setup priority: Download from char.com → grant macOS audio permissions → select audio devices → test a short recording → choose STT provider (Cactus for local, Deepgram/AssemblyAI for cloud) → configure LLM (Ollama for offline, OpenRouter for cloud).
Fully offline setup: Free plan + Cactus local STT + Ollama local LLM. No cloud calls, no API keys needed.
Speaker ID requires Lite+ — the Free plan does not include speaker identification. If group meeting attribution matters, budget for Lite ($8/mo) minimum.
If you discover a gotcha, workaround, or tip not covered in 
references/learnings.md
, append it there.
你无需再查阅平台指南——专注于用户的具体场景。
首次设置优先级：从char.com下载→授予macOS音频权限→选择音频设备→测试短录音→选择STT提供商（本地使用Cactus，云端使用Deepgram/AssemblyAI）→配置LLM（离线使用Ollama，云端使用OpenRouter）。
完全离线设置：免费版 + Cactus本地STT + Ollama本地LLM。无需调用云端服务，无需API密钥。
说话人识别需要Lite版及以上——免费版不包含说话人识别功能。如果群组会议中的说话人归属很重要，至少需要预算Lite版（每月8美元）。
如果你发现
references/learnings.md
未覆盖的陷阱、临时解决方案或技巧，请补充到该文档中。

Gotchas

注意事项

Best-effort from research — review these, especially items about plan-gated features and integration gotchas that may be outdated.
  • Speaker identification is Lite+ only and still maturing — multiple HN users reported it failing in group meetings. The team acknowledged outdated marketing. Test with your meeting size before committing.
  • "Local-first" includes telemetry by default — Char ships with PostHog analytics and Sentry error tracking. Opt-out exists but isn't surfaced during onboarding. Check Settings → Privacy.
  • GPL-3.0 license has copyleft implications — unlike MIT-licensed Meetily, corporate legal teams may flag GPL. Self-hosted enterprise use is fine; distributing modified binaries triggers copyleft.
  • macOS only until Q2 2026 — Windows and Linux builds are in progress but not publicly available yet. Don't plan a cross-platform deployment today.
  • Public REST API is "coming soon" — no programmatic access to meetings/transcripts beyond the CLI 
    char db
    commands. The hooks system (beforeListeningStarted, afterListeningStopped) is the only automation surface.
  • AudioTap API issues on some Macs — M4 Pro Mini users reported the app bouncing on the dock without opening. No workaround beyond waiting for a fix.
  • CRM integrations are planned, not shipped — HubSpot, Salesforce, Slack, Notion appear on the roadmap but are not available yet.
基于研究的尽力总结——请查看这些内容，尤其是与套餐专属功能和集成陷阱相关的可能过时的条目。
  • 说话人识别仅适用于Lite版及以上，且仍在完善中——多个HN用户反馈该功能在群组会议中失效。团队已承认营销信息过时。在付费前，请针对你的会议规模进行测试。
  • “本地优先”默认包含遥测功能——Char默认搭载PostHog分析和Sentry错误追踪。存在退出选项，但在入门流程中未显示。请前往设置→隐私选项进行操作。
  • GPL-3.0协议具有Copyleft影响——与MIT协议的Meetily不同，企业法务团队可能会标记GPL协议。自托管企业使用没问题，但分发修改后的二进制文件会触发Copyleft条款。
  • 截至2026年第二季度前仅支持macOS——Windows和Linux版本正在开发中，但尚未公开可用。目前不要计划跨平台部署。
  • 公开REST API“即将推出”——除CLI的
    char db
    命令外，暂无程序化访问会议/转录内容的途径。钩子系统（beforeListeningStarted、afterListeningStopped）是唯一的自动化接口。
  • 部分Mac机型存在AudioTap API问题——M4 Pro Mini用户反馈应用在程序坞弹跳但无法打开。除等待修复外暂无临时解决方案。
  • CRM集成功能仍在规划中，尚未发布——HubSpot、Salesforce、Slack、Notion出现在路线图中，但目前不可用。

Related skills

相关技能

  • /sales-note-taker
    — Comparing Char to other note-takers (Fathom, Fireflies, Avoma, Gong, etc.) or wiring transcript APIs into CRM/warehouse
  • /sales-meetily
    — Meetily: MIT-licensed open-source alternative with Whisper/Parakeet transcription
  • /sales-granola
    — Granola: another bot-free AI notepad, closed-source with HubSpot/Attio CRM sync
  • /sales-call-review
    — Review a specific call for coaching feedback and scoring
  • /sales-coaching
    — Build coaching programs that consume call data
  • /sales-integration
    — General CRM/tool integration patterns (Zapier, webhooks, iPaaS)
  • /sales-do
    — Not sure which skill to use? The router matches any sales objective to the right skill. Install: 
    npx skills add sales-skills/sales --skill sales-do
  • /sales-note-taker
    — 对比Char与其他笔记工具（Fathom、Fireflies、Avoma、Gong等），或将转录API接入CRM/数据仓库
  • /sales-meetily
    — Meetily：采用MIT协议的开源替代工具，支持Whisper/Parakeet转录
  • /sales-granola
    — Granola：另一款无机器人AI记事本，闭源，支持HubSpot/Attio CRM同步
  • /sales-call-review
    — 针对特定通话进行教练式反馈和评分复盘
  • /sales-coaching
    — 构建使用通话数据的教练项目
  • /sales-integration
    — 通用CRM/工具集成模式（Zapier、webhooks、iPaaS）
  • /sales-do
    — 不确定使用哪个技能？该路由可将任何销售目标匹配到合适的技能。安装方式：
    npx skills add sales-skills/sales --skill sales-do

Examples

示例

Example 1: Setting up fully offline meeting notes

示例1：设置完全离线会议笔记

User says: "I need meeting transcription that works 100% offline — no data leaving my machine." Skill does:
  1. Confirms Char Free plan supports fully local operation
  2. Guides through Cactus local STT engine setup (no API keys needed)
  3. Configures Ollama as the LLM provider for local summarization
  4. Tests with a short recording to verify end-to-end local processing
  5. Notes that speaker identification requires Lite plan (cloud) Result: Fully air-gapped meeting transcription and summarization.
用户提问：“我需要100%离线工作的会议转录工具——数据不会离开我的设备。” 技能处理：
  1. 确认Char免费版支持完全本地操作
  2. 指导设置Cactus本地STT引擎（无需API密钥）
  3. 配置Ollama作为本地摘要的LLM提供商
  4. 通过短录音测试以验证端到端本地处理
  5. 说明说话人识别需要Lite版（云端功能） 结果：实现完全离线的会议转录和摘要功能。

Example 2: Speaker identification not working

示例2：说话人识别失效

User says: "I have a 4-person team call and Char can't tell who said what." Skill does:
  1. Confirms the user is on Lite+ plan (speaker ID is plan-gated)
  2. Explains speaker ID is still maturing — known issues with group meetings
  3. Suggests workarounds: separate audio devices per speaker, post-meeting manual attribution
  4. Notes this is a known pain point the Char team is actively improving Result: Realistic expectations set, workarounds provided.
用户提问：“我有一个4人团队通话，Char无法区分是谁在说话。” 技能处理：
  1. 确认用户使用的是Lite版及以上套餐（说话人识别是套餐专属功能）
  2. 说明说话人识别仍在完善中——群组会议中存在已知问题
  3. 建议临时解决方案：每位使用单独的音频设备，会后手动归属说话人
  4. 说明这是Char团队正在积极改进的已知痛点 结果：设定合理预期，提供临时解决方案。

Example 3: Building a custom plugin

示例3：构建自定义插件

User says: "I want to build a Char plugin that sends meeting summaries to Slack." Skill does:
  1. Reads the plugin SDK reference — plugins register as tabs via 
    window.__char_plugins.register()
  2. Notes that plugins are tab-based UI extensions, not background automations
  3. For Slack integration, recommends the hooks system instead: 
    afterListeningStopped
    hook triggers a shell script that reads the transcript and posts to Slack via webhook
  4. Provides the hooks.json configuration path and format Result: Correct architecture choice (hooks, not plugins) for the integration.
用户提问：“我想构建一个Char插件，将会议摘要发送到Slack。” 技能处理：
  1. 查阅插件SDK参考——插件通过
    window.__char_plugins.register()
    注册为标签页
  2. 说明插件是基于标签页的UI扩展，而非后台自动化工具
  3. 对于Slack集成，建议使用钩子系统替代：
    afterListeningStopped
    钩子触发shell脚本，读取转录内容并通过webhook发送到Slack
  4. 提供hooks.json配置路径和格式 结果：为集成选择正确的架构（钩子而非插件）。

Troubleshooting

故障排除

App bounces on dock but won't open

应用在程序坞弹跳但无法打开

Symptom: Clicking Char shows a dock bounce animation but the window never appears, no error messages. Cause: AudioTap API incompatibility on certain Mac models (particularly M4 Pro Mini reported). Solution: Check GitHub Issues for your Mac model. Try running from Terminal (
/Applications/Char.app/Contents/MacOS/Char
) to see error output. Update to the latest version — AudioTap fixes ship frequently.
症状：点击Char后，程序坞显示弹跳动画，但窗口从未出现，无错误提示。 原因：部分Mac机型（尤其是M4 Pro Mini反馈的）存在AudioTap API兼容性问题。 解决方案：查看GitHub Issues中对应Mac机型的情况。尝试从终端运行（
/Applications/Char.app/Contents/MacOS/Char
）查看错误输出。更新到最新版本——AudioTap相关修复会频繁发布。

Speaker identification fails in group meetings

群组会议中说话人识别失效

Symptom: All speech attributed to one speaker or speakers randomly swapped. Cause: Speaker diarization is still maturing in Char. System audio capture (vs per-participant streams) makes diarization harder. Solution: Ensure you're on Lite+ plan (Free has no speaker ID). For critical meetings, consider a bot-based tool (Fathom, Fireflies) that captures per-participant audio streams for better diarization.
症状：所有语音被归为同一说话人，或说话人被随机调换。 原因：Char中的说话人分离功能仍在完善中。系统音频捕获（而非每位参与者的单独流）增加了分离难度。 解决方案：确保你使用的是Lite版及以上套餐（免费版无说话人识别功能）。对于重要会议，考虑使用基于机器人的工具（Fathom、Fireflies），这些工具可捕获每位参与者的音频流以实现更好的分离效果。

Transcription quality poor or missing words

转录质量差或缺失内容

Symptom: Transcripts have gaps, wrong words, or low accuracy. Cause: Local Cactus engine trades accuracy for privacy. Cloud providers vary in quality. Solution: Switch from Cactus to a cloud STT provider (Deepgram or AssemblyAI recommended for accuracy). If staying local, ensure good microphone input and minimal background noise. Check that the correct audio device is selected in Settings.
症状：转录内容存在空白、错误词汇或低准确率。 原因：本地Cactus引擎为了隐私牺牲了准确率。不同云提供商的质量存在差异。 解决方案：从Cactus切换到云STT提供商（推荐Deepgram或AssemblyAI以获得高准确率）。如果坚持使用本地引擎，请确保麦克风输入质量良好，背景噪音最小。检查设置中是否选择了正确的音频设备。