Wechat Use

wechat — macOS CLI

Setup failures and user reports (v1.18.7+)

Release status: v1.18.7 is published and marked latest. The capabilities below apply to the released build; users on older versions can upgrade with the README install command.

Do not require users to identify their macOS version or whether this is a fresh install before helping. The setup window collects redacted environment and service facts automatically. Read the WXS-… problem code from an error screenshot; if more detail is needed, the user can click “复制诊断信息” on any setup page, including a waiting page. The window also provides a minimal report when core cannot start. Never request keys.json, activation tokens, raw memory output, or private chat data.

Missing system developer-tool authorization is detected before initialization and uses “授权并继续” with the native macOS administrator confirmation. Preserve existing login and grants. Do not suggest reinstalling or globally resetting permissions for an unexplained initialization failure. A v1.18.6 error window can use its existing “修复安装” button to upgrade once to the current release; then follow the specific diagnosis rather than repeatedly reinstalling.

Unified CLI for WeChat on macOS. Send messages and query the encrypted local databases for sessions, contacts, chat history, group members, Moments, favorites. Conversation selection runs inside the managed clone without activating or raising its windows.

Managed WeChat 4.1.9 (v1.18.0+)

The installer automatically copies a supported local 4.1.9 or downloads a pinned Tencent DMG (no prompt). It installs ~/Applications/WeChat-4.1.9-wechat-use.app with its own bundle ID and sandbox container, disables that clone's updates, and saves ~/.wx-rs/managed-wechat.json.

All CLI, daemon, HTTP bridge, and MCP operations default to that clone. Check its current state first and reuse an existing login. If it is closed, launch it with open -g "$HOME/Applications/WeChat-4.1.9-wechat-use.app". Request first-time login only when it is actually required, then run wechat-use init. Do not substitute open -a WeChat, sign the primary /Applications/WeChat.app, remove the managed selection, or fall back to the primary when the clone is closed. Only 4.1.9 receives ongoing compatibility work. When repairing installation, rerun the installer; do not download the latest WeChat.

Unified setup and recovery (v1.18.6+)

Use wechat-use setup --json to inspect the current setup stage without opening UI. For a normal user, the curl installer and interactive wechat-use init use the same native setup window. It preserves existing activation, login and keys, selects compatible installed developer tools for the tool only, starts the official background service, and verifies actual service ownership and permissions.

If a user action is needed, run wechat-use setup or open ~/Applications/WechatUseSetup.app. Let the window handle activation, login, system authorization, drag-and-drop and rechecking. Do not give users sequences of doctor, init, launchctl, or xcode-select repair commands. Never launch a Terminal-owned daemon as a permission workaround. The service must be owned by the official bridge.

The bridge's /setup/status remains available before keys or Accessibility grants are ready. It is diagnostic only; sending remains gated. Setup control uses local signals to the validated launchd process, not an unauthenticated HTTP write endpoint.

An unconfirmed setup test is persisted and is not automatically replayed after closing or reopening the window. --skip-verify explicitly skips sending; permission readiness alone is not delivery evidence. Human identity confirmation and actual macOS grants remain user-controlled.

User setup guide

Background execution

When System Settings shows the tool enabled but the actual launchd bridge still rejects Accessibility, do not treat the displayed switch as proof. Check the installed path and signature, refresh only that tool's switch, and recheck a freshly loaded official service. If those checks still fail, the specific old permission entry may be stale. An authorized local computer-control agent can remove and re-add only the failing tool entry using the current installed binary. Identify the exact row before each action; do not use blind coordinates or reset other apps' permissions. Preserve the dedicated WeChat process and keys. Afterwards verify HTTP health, the daemon's real Accessibility result, and that its parent is the official bridge. A user identity challenge remains a user action.

Normal operations must preserve the user's foreground window. Do not call open -a, activate/raise/unminimize the clone, switch applications, or use foreground/global hotkeys as routine preparation. Use process-targeted input and background conversation navigation, and verify the exact recipient. This also applies to installation, initialization, permission errors, and service recovery. Return actionable diagnostics without automatically opening Settings, Finder, permission dialogs, or retrying a send. doctor --fix-tcc is an explicit interactive recovery command and opens System Settings. Foreground recovery is a last resort only after available background methods are shown to be unavailable. Never activate and blindly resend an unconfirmed message; inspect delivery first. A blank AX snapshot alone does not justify foreground recovery or restarting WeChat.

Fast path (read this first)

Send a WeChat message in one call:

wechat-use send "早上好" Lisa                 # fuzzy name match (remark / nick / alias)
wechat-use send "hi" filehelper              # wxid — zero DB lookup, fastest
wechat-use send "提醒一下" 20590343959@chatroom  # group wxid (ends in @chatroom)

Resolution rules (applied in order):

1. RECIPIENT matches a wxid shape (wxid_…, …@chatroom, gh_…, biz_…, or reserved like filehelper) → skip all DB work and send directly.
2. Otherwise, search the local contact DB (remark / nickname / alias / wxid) with session-recency bias:
   • single match → send
   • multiple matches but only one has recent activity (30d) → send to that one
   • otherwise → exit 2 + JSON {"status":"ambiguous","candidates":[...]}; the agent picks and retries with the explicit wxid

On ambiguous, a sample response:

{
  "status": "ambiguous",
  "hint": "Lisa",
  "candidates": [
    {"wxid": "lishuang683451", "display_name": "lisa", "last_seen": "2026-04-20 05:34:55"},
    {"wxid": "wxid_xxx", "display_name": "Lisa (另一个)", "last_seen": ""}
  ],
  "note": "multiple matches; pass one of the wxids explicitly: wechat-use send <text> <wxid>"
}

Agent should: read candidates[0].wxid, retry wechat-use send "<text>" <wxid>. Don't ask the user unless the top candidate has no recent activity or multiple candidates do.

HTTP Bridge for agent integration (v1.10+)

wechat-bridge is a separate binary that wraps the daemon's RPCs as a stable localhost HTTP surface. Use this when wiring WeChat into agent platforms (Hermes, n8n, Dify, LangChain, custom bots) — HTTP is cheaper to integrate than spawning the CLI per call.

# Start bridge (binds 127.0.0.1:18400 by default)
wechat-bridge &

# Health + send-readiness
curl http://127.0.0.1:18400/health

# Send
curl -X POST http://127.0.0.1:18400/send \
  -H 'Content-Type: application/json' \
  -d '{"wxid":"filehelper","text":"hi"}'

# SSE message stream — ⚠️ ALWAYS pass ?since=<epoch>
# Without ?since, default is 0 = backfills entire local message history
# (1MB+ in seconds for typical accounts). For agent / long-running flows
# always pass a since timestamp; pick "now" for live-only or last-checkpoint.
SINCE=$(date +%s)
curl -N "http://127.0.0.1:18400/messages/stream?since=$SINCE"

Endpoints:

Account readiness and image sends

Image sending and self-message reads resolve the current account in each message shard; File Transfer Assistant warmup samples are no longer required on the managed 4.1.9 path. /health reports wechat.state and wechat.account_state separately from daemon liveness. not_ready means the account is not ready for sending; needs_wechat_login in doctor requires completing WeChat login, not reinitializing or deleting caches. WeChat may still require phone confirmation or a QR login. The tool never bypasses that confirmation.

Ordinary files, videos and cards are not supported by native send yet.

Images over HTTP

The image path must name a file on the bridge host. Native requests accept image_path (or imagePath); Hermes requests use imagePath. Omit the text field for an image-only request. Send text and images separately; combining both in one request is rejected instead of silently dropping content.

curl -X POST http://127.0.0.1:18400/send \
  -H 'Content-Type: application/json' \
  -d '{"wxid":"filehelper","image_path":"/absolute/path/photo.png"}'

SSE payload shape (v1.10.28 — Wechaty-aligned + isMentioned)

/messages/stream emits event: messages carrying a JSON array of:

{
  messageId: string,
  chatId: string,                // wxid (DM) or groupid@chatroom
  senderId: string,              // in group: sender's wxid; in DM: the other party's wxid
  senderName: string,
  chatName: string,
  isGroup: boolean,
  body: string,                  // human-readable text. For URL / quote / mini_program, body is the title — raw XML is NOT exposed here.
  hasMedia: boolean,
  mediaType: "image"|"voice"|"video"|"file"|"",
  mediaUrls: string[],           // first entry is CDN URL when applicable
  mentionedIds: string[],        // v1.10.25+ — authoritative @-mention list resolved by daemon
  isMentioned: boolean,          // v1.10.28+ — bridge-authoritative "this row @-mentions ME". Self-sent rows are always false.
  quotedParticipant: string,     // v1.10.27+ — populated from refer.fromUser on quote replies
  botIds: string[],              // legacy heuristic self-marker; NEW consumers should rely on fromSelf instead
  fromSelf: boolean,             // v1.10.25+ — bridge-authoritative "this row was produced by our own POST /send"; DROP THESE to avoid self-echo loops
  messageKind: "text"|"image"|"audio"|"video"|"contact"|"emoticon"|"location"|
               "url"|"attachment"|"mini_program"|"chat_history"|"transfer"|
               "red_envelope"|"recalled"|"system"|"unknown",  // v1.10.27+, aligned to Wechaty's MessageType enum
  urlLink?:     { title, description, url, thumbUrl },                             // present iff messageKind=url
  miniProgram?: { title, description, appId, username, pagePath, thumbUrl },        // present iff messageKind=mini_program
  refer?:       { svrId, fromUser, chatUser, displayName, content },              // present on quote replies
  recall?:      { replacedMsgId, text },                                            // present iff messageKind=recalled
  media?:       { aesKey, md5, cdnUrl, cdnThumbUrl, length, durationSeconds, localPath },  // structured metadata for image/audio/video/attachment
  timestamp: number,
}

The full JSON Schema is committed at wx/schema/sse-payload-v1.10.28.schema.json and enforced by a contract test in the daemon build.

Consumer checklist:

• Filter self-echo with fromSelf === true. Do NOT use senderId === myWxid — in DM both directions share the same senderId.
• In groups, only respond when isGroup && isMentioned — the daemon already resolves the authoritative mention comparison, so don't reimplement mentionedIds.includes(myWxid) yourself (your wxid may be a remark / lookup that the daemon resolves correctly). The bridge will also drop non-@ group rows automatically when WECHAT_BRIDGE_GROUP_MENTION_ONLY=1.
• Need the URL only? mediaUrls[0]. Need aesKey + md5 to decrypt or verify? media.cdnUrl / media.aesKey / ….
• For messageKind: "image", do not inline base64 image bytes in chat responses. Call wechat-use image get <messageId> --chat <chatId> --json, parse absolutePath, then use the host agent's file/image Read capability on that path. Default --from auto (since v1.13.11) tries the daemon's in-memory lookup first (fast, works when the user has opened the image at least once in WeChat) and falls back to CDN replay only on miss. If the result is image not yet viewed in WeChat (cache empty), and CDN fallback failed, ask the user to open the image once in WeChat and retry. cdn-expired or needs local-decrypt adaptation means neither path can recover this image — surface that to the user instead of guessing.
• Expect body for URL / quote / mini_program to be the human title. If you were previously parsing raw <appmsg> XML from body, migrate to the dedicated urlLink / miniProgram / refer objects.
• Backward compatible: every pre-v1.10.25 field is preserved in name + type. New fields are additive.

Security notes for agents:

• Bridge binds 127.0.0.1 — not exposed to LAN without tunnelling.
• Set WECHAT_BRIDGE_BEARER=<secret> env var to require Authorization: Bearer <secret> on non-/health routes. Use this if tunnelling via Tailscale / SSH.
• Activation gating is enforced inside wechatd, not in the bridge. A missing / expired wechatuse_ token → HTTP 401 / 402 on /send. Bridge cannot bypass activation.

Command groups

All query commands default to YAML output (agent-friendly, low token). Add --json to get JSON.

🛑 Safety rules (CRITICAL — read before calling send)

Every send call must resolve to a known wxid. No silent default to "current chat" — that flag (--current-chat) was removed pre-1.13; if the resolver can't find a recipient, stop and ask the user.

Correct flows for "给 XXX 发 YYY":

1. Just try it: wechat-use send "YYY" XXX. Fast-path resolver (see top of this doc) handles wxid-shaped targets instantly and fuzzy-matches names against the local contact DB with session-recency bias.
2. On exit 2 + status: "ambiguous": if candidates[0] has last_seen within ~30 days and others are stale/empty, the CLI already auto-picked it and returned success. If it truly was ambiguous (multiple candidates with recent activity), pick one yourself by asking the user — don't guess.
3. On no contact matches "XXX": ask the user for the wxid (or have them confirm a candidate from wechat-use contacts --query XXX --brief).

Hard rules (the agent MUST follow):

• DO NOT guess or fabricate a wxid. If resolution fails, escalate to the user.
• DO NOT scan the filesystem / grep logs / use AppleScript to hunt for a wxid. The CLI already searches the local contact DB via the fast path — trust it. If it can't find the recipient, stop and ask the user.
• DO NOT invoke wechat-use contacts followed by wechat-use send as two separate calls unless the first fast-path send already told you it was ambiguous. The one-liner saves ~400ms and one agent round-trip.

Capability matrix

Agent: first-use setup

Step 1 — Check wechat is on PATH:

command -v wechat

If missing:

curl -fsSL https://raw.githubusercontent.com/leeguooooo/wechat-use/main/install.sh | bash
# Ensure ~/.local/bin is on PATH
case "${SHELL##*/}" in
  fish) fish_add_path "$HOME/.local/bin" ;;
  zsh)  grep -q '.local/bin' ~/.zshrc  2>/dev/null || echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.zshrc ;;
  bash) grep -q '.local/bin' ~/.bashrc 2>/dev/null || echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc ;;
esac
export PATH="$HOME/.local/bin:$PATH"

Step 2 — Initialize the already-logged-in managed WeChat:

wechat-use init

On the managed WeChat 4.1.9 this scans the existing process for per-database keys; it does not need to quit and relaunch WeChat. Keep the existing account, container and login state. Reuse cached keys while queries work, including after a daemon restart; only rescan when keys are missing or fail to decrypt.

If WeChat itself is on a login screen, the user must complete that genuine login challenge. Do not manufacture a new challenge by restarting a logged-in clone, replacing its bundle identity, or deleting its local account data.

Step 3 — Check service permissions:

wechat-use doctor --json

Check the running daemon's daemon_accessibility result and the bridge's health. Accessibility belongs to the installed helper identities; a signed upgrade should retain those grants. If missing, run wechat-use init --fix-tcc: it opens the correct pane, selects the binaries in Finder for drag-and-drop, then automatically rechecks the running daemon. Do not ask the user to find hidden paths or type recovery commands themselves.

Step 4 — Verify a send without manual chat selection:

wechat-use send "安装验证" filehelper --json

The daemon reads the selected chat identity and uses background conversation navigation to select the exact requested recipient before typing. It checks the identity again before Enter. Missing build calibration, ambiguous context, a changed recipient, or an unavailable target stops the send before dispatch. Do not ask the user to click filehelper or send a warmup message as a normal setup step. The current local candidate requires a matching recipient-context calibration; distribution of this calibration is still pending and must be tested before release.

Only a verified record in the requested chat counts as delivery. A delivery_verify_timeout can be a delayed delivery: inspect history before retrying. If automatic preparation fails, keep the diagnostic output and run wechat-use doctor --json; do not restart the logged-in WeChat or reactivate its account as a workaround.

An empty AX tree or blank automation screenshot is not proof that WeChat is blank, logged out, or frozen. Use process/CLI/database evidence and the user's visible state before diagnosing it. Never quit/relaunch a logged-in clone just to recover an automation snapshot.

Send failure modes (agent self-recovery table)

When wechat-use send returns {ok: false, …} or stderr matches the reason: slot in the support payload, use the playbook below before asking the user to debug. Diagnose in the background and request human action only for a confirmed step that cannot be completed there. Never infer permission to take focus from a failed send.

Key principle: every reason: slot is stable enum; the support payload includes ax_trusted, input_monitoring, binary_fingerprint, cli_version, daemon_version — match on those exact field names, not on the human-readable Chinese error message (it may evolve across releases).

LaunchAgent / TCC handbook for agents: if you ever need to manually reset responsibility chains (e.g. user's machine has a customized LaunchAgent layout), use launchctl bootout + launchctl bootstrap — never launchctl kickstart -k. kickstart doesn't re-read plist EnvironmentVariables and doesn't reset launchd's responsibility-chain cache. The install.sh upgrade flow is the reference implementation.

Usage — send

# Recipient resolves wxid / 群名 / 昵称 / 备注 (fuzzy match against local contact DB)
wechat-use send "你好 🎉" filehelper
wechat-use send "会议 5 分钟后开始" lishuang683451
wechat-use send "早上好" Lisa                # 找不到 → friendly error + 候选

# Group send (resolver also handles group display names)
wechat-use send "今天 19:00 团建" "AI 星球"

# JSON output for agents that parse responses
wechat-use send "ok" filehelper --json

# Dry-run: resolve recipient + validate args, do NOT actually send. Useful when
# the agent wants to verify a fuzzy name → expected wxid before committing.
wechat-use send "draft" "李工" --dry-run --json

send arguments

send --json 三态契约 (v1.13.20+)

所有 --json 输出都带顶层 ok: bool,agent 直接 if (r.ok) {...} else {...} 不需要解析三套 schema:

# 推荐:agent 用 jq 分支
wechat-use send "hi" filehelper --dry-run --json | jq -e '.ok' && echo "✓ resolved" || echo "✗ failed"

stderr 仍然有 human-readable 错误描述(给终端用户看);agent 只需 parse stdout JSON。

Usage — query

# Sessions (recent conversations)
wechat-use sessions -n 20                           # full yaml
wechat-use sessions --brief -n 20                   # 单行/会话, 带未读数
wechat-use sessions --filter group --json -n 20     # 只看群聊 (chat_type: group / private / official_account / folded / other)
# JSON 字段命名:未读数是 `unread_count`(下划线全名), 不是裸 `unread`。brief 视图渲染成 [N unread] 仅是显示, 实际字段是 unread_count。

# Contacts
wechat-use contacts --query 李                      # fuzzy match nickname/remark/wxid
wechat-use contacts --brief -n 50                   # 单行/联系人 (姓名 + wxid)

# Unread
wechat-use unread -n 5

# History (chat positional or --chat flag, both accepted)
wechat-use history "张三" -n 2000
wechat-use history --chat 21263894984@chatroom -n 200
wechat-use history "AI 星球" --since "2026-04-01" --until "2026-04-15" -n 200   # ISO date OK
wechat-use history "AI 星球" --since 1719793200 --until 1720484400 -n 200       # epoch OK too

# Search (FTS5 trigram, v1.16.21+)
wechat-use search "会议纪要"                              # 全局,毫秒级
wechat-use search "报销" --in "财务群"                    # 群限定
wechat-use search "claude" --since "3 days ago"          # 加时间窗
wechat-use search "report" --since "2026-05-01" --until "2026-05-10"

# Search agent UX (v1.16.22+):
wechat-use search "desktop" --in "Helm" --context 5m     # 每条命中带前后 5 分钟同群上下文,文本模式 `>` 标记命中
wechat-use search "report" --timeout-ms 5000             # 硬上限 5s,超时 exit 124 + `search_timeout`
# `--json` meta 多带: elapsed_ms / row_count / chat_resolved / chat_hint / context_secs

# HTML 证据页 (v1.16.23+, v1.16.24 重做 WeChat 主题 UI + 默认 inline 图片):
wechat-use search "desktop" --in "Helm" --report                    # 单文件 HTML,默认 --context 5m,图片 base64 inline
wechat-use search "房价" --in "立水桥" --report --no-media           # 跳过图片(escape hatch,渲染成占位卡片)
wechat-use search "通知" --in "群" --report --out report.html       # 自定义路径,无 sidecar 目录,转发邮件/微信直接带走
# stdout = 写入的文件路径(text 模式)或 {ok, report_path, hit_count, ...}(--json)
# 输出文件 self-contained:inline CSS,无 JS,无 CDN,可直接邮件/微信/Notion 转发

# Search caveats (FTS5 trigram limits):
# - 查询 < 3 字符: 全局会报 `query_too_short`,加 `--in <chat>` 则
#   fallback 到 LIKE 单 shard。例: `wechat-use search 卡 --in <chat>`
# - 首次升级后 daemon 在后台 backfill (266k 消息 ~86s 本机);期间
#   查询会报 `search_index_unavailable`,等一两分钟后再查
# - mirror DB at `~/.wx-rs/search-index.db` (SQLCipher 加密,key 从
#   现有 WeChat key HKDF 派生),mirror 跟 WeChat 数据库一起在 home 目录
# - `--context` 需要配合 `--in <chat>` 才生效(无 chat 时被 daemon 跳过)

# Group members
wechat-use members "AI 星球"

# Stats
wechat-use stats "AI 星球"

history --json payload shape (stable contract for agents)

顶层包装(history / sessions / unread / search / digest 全一致):

{
  "meta": {
    "chat_latest_timestamp": 1778981425,
    "shards_scanned": 2,
    "shards_hit": 2,
    "status": "ok",
    "order": "desc",
    "now_unix": 1779178200,
    "since_resolved": 1778573400,
    "until_resolved": 1779178200
  },
  "rows": [ {...row...}, {...row...} ]
}

不是裸数组。jq 要 .rows[] 不是 .[]。meta.order 是 v1.16.12+ 加的字段,告诉 你 rows 是 "desc"(新→老,history 默认)还是 "asc"(老→新,digest 默认 / wechat history --order asc)。别拿 rows[0] 当"最新"也别当"最老",先看 meta.order。

meta.status 取值:ok / windowed(传了 --since/--until)/ possibly_stale (SessionTable 比 history 领先 > 24h,大概率分片漂)/ possibly_stale_unknown_shards (磁盘有新分片 daemon 不认,要重跑 wechat-use init)。

v1.16.19+ 时间窗回灌(history): meta.now_unix 是查询时的本机 epoch; meta.since_resolved / meta.until_resolved 是 --since / --until 自然语言解析后的 实际 epoch(没传不出现)。自然语言 --since "2 days ago" 解析完后,agent 不用重算时区, 直接拿这三个值跟用户复述"实际查的是 X 到 Y, 当下 Z"。

每条 message row 字段(snake_case):

字段稳定性:增加 = 默认 null / 缺省;不会重命名 / 改类型(契约由 v1.10.27 起的 SSE schema 单测守)。

想跨 chat 拉"我说过什么": 用 wechat-use sent --since "7 days ago" --json(v1.16.12+), 比手动遍历 sessions + filter 干净得多。

群名歧义: wechat-use history "AI 星球" 可能匹配多个(同名群 + 同名联系人)。 撞歧义时 CLI 会列出候选 + 报错,改用 --chat <wxid> 或 --chat <chatroom_id@chatroom> 明确指定。

v1.16.19+ user alias: 长群名 / emoji 前缀群名 fuzzy 经常 miss(詹密群 / 季景铭郡业主三群),用 alias map 一劳永逸:

wechat-use alias add 詹密群 17765974862@chatroom
wechat-use alias add AI星球 20590343959@chatroom
wechat-use alias list                # 看现有
wechat-use alias rm 詹密群           # 删

Resolver 优先级:wxid-shape > alias > contact.db fuzzy。wechat-use send / history / digest / search / recalled / sent / listen --wxid 全部 共享同一查找(写到 ~/.wx-rs/aliases.json)。

# Export
wechat-use export "张三" --format markdown -o zhang.md
wechat-use export "AI 星球" --format json -o ai.json -n 5000

# Incremental (since last checkpoint saved in ~/.wx-rs/cursor.json)
wechat-use new-messages -n 50      # advances checkpoint
wechat-use new-messages --reset    # rewind checkpoint to "now" so next call starts fresh

# Favorites
wechat-use favorites                          # all locally-cached items

# Image media (local in-memory lookup first, CDN fallback)
wechat-use image get <local_id> --chat <chat_id>            # decrypts + writes to ~/.wechat/media-cache/<md5>.jpg
wechat-use image inspect <local_id> --chat <chat_id>        # dump CDN metadata (no key/url leak)

# Voice media — history auto-transcribes by default (v1.13.25+)
wechat-use audio setup [--model small|medium|large]         # one-time: install deps + download model
wechat-use audio transcribe <svr_id> [--language zh]        # single-file pipeline
wechat-use audio get <svr_id>                               # raw SILK_V3 bytes, no decode

Voice — agents reading group history just work (v1.13.25+)

After running wechat-use audio setup once (~2-3 minutes downloads ~1.5GB medium model + builds silk-decoder), wechat-use history automatically transcribes voice messages so the agent sees the spoken content in display_text (and structured in media.transcript) — no more [语音消息] placeholders breaking conversation context. Transcripts are cached by SHA-256 of the audio blob, so re-reading the same chat is near-instant.

# One-time setup
wechat-use audio setup

# Read a chat — voice messages already transcribed inline
# (Output is {meta, rows} not a bare array — use `.rows[]` not `.[]`.)
wechat-use history <chat> --json | jq '.rows[] | {kind: .message_kind, text: .display_text}'

Opt-out (skip transcribe to keep history fast / private):

wechat-use history <chat> --no-transcribe                   # skip transcribe entirely
wechat-use history <chat> --transcribe-model small          # smaller / faster model
wechat-use history <chat> --quiet                           # silence stderr progress lines
                                                        # (auto-on in --json mode)

The media.transcript_status field on each audio row tells the agent where the text came from: cached / transcribed / no_deps (run wechat-use audio setup) / failed / skipped_svr_id_zero / invalid_input.

wechat-use doctor reports audio readiness in two rows so machine consumers can check default-model status without parsing strings:

• audio_transcribe_setup — overall ffmpeg / whisper-cli / silk-decoder presence. Always ok: true (audio is optional; missing tools must not flip overall doctor status to needs_init).
• audio_transcribe_default_model — ggml-medium.bin exists. ok reflects reality, but excluded from the overall status calculation so a user without me

…(truncated)