feat: / Skill Invocation in Group Channels

.agents/skills/verify-deepseekbot/features/README.md

@@ -70,7 +70,7 @@ One file per user-facing feature. Each file has four sections: `Sub-features`, `
 | Historical Overview: idle PersonaBot (retired by #2026) | no | `scripts/e2e-activity-idle.mjs` | [activity-center-idle.md](activity-center-idle.md) |
 | Temporary Workspace Grant authorization, read and revocation | no | `scripts/e2e-workspace-grant.mjs` | [workspace-grant.md](workspace-grant.md) |
 | Default Bot Browser open, observe, stop and Access refusal | no | `scripts/e2e-browser-lifecycle.mjs` | [browser-lifecycle.md](browser-lifecycle.md) |
-| Skill Invocation from `/` in a Bot DM | yes for the reply | `verify-app` Bot DM and `skills-list` / `send --skill` | [skill-invocation.md](skill-invocation.md) |
+| Skill Invocation from `/` in a Bot DM or Group Channel | yes for the reply | `verify-app` DM/Group; `skills-list`, `send --skill` / `channelSend` | [skill-invocation.md](skill-invocation.md) |
 | Composer footer customization and optional reply speed | yes for a speed sample | none (Web interaction path) | [composer-footer.md](composer-footer.md) |
 | Disposable Memory history, restore and stale-head refusal | no | `scripts/e2e-memory-restore.mjs` | [memory-restore.md](memory-restore.md) |
 | Version-bound Memory images, Markdown previews and CLI export | no | `scripts/e2e-memory-images.mjs`, `scripts/e2e-memory-markdown.mjs`, `scripts/e2e-memory-formats.mjs` | [memory-images.md](memory-images.md) |

.agents/skills/verify-deepseekbot/features/skill-invocation.md

@@ -1,20 +1,25 @@
-# Skill Invocation in a Bot DM
+# Skill Invocation in a Bot DM or Group Channel
 
-A Human types `/` in a PersonaBot DM, picks a Skill, and sends it. The Host stores the Skill's text with the message and the Bot reads it as a `<skill_invocation>` block ([ADR-0346](../../../../docs/adr/0346-skill-invocation-is-snapshotted-text-that-addresses-no-one.md)).
+A Human types `/` in a PersonaBot DM or Group Channel, picks a Skill, and sends it. The Host stores the Skill's text with the message and the Bot reads it as a `<skill_invocation>` block ([ADR-0346](../../../../docs/adr/0346-skill-invocation-is-snapshotted-text-that-addresses-no-one.md)).
 
 ## Sub-features
 
-- `menu` lists composer commands first, then Skills: Host Skills and the Project Skills under the Bot's Memory Repository `.agents/skills` or `.dsh/skills`. Each row shows its owner (Bot avatar or Skill icon), `/name` and a one-line description.
-- `rank` orders Skills by name prefix, then name subsequence, then description text.
+- `menu` lists composer commands first, then Skills: Host Skills and the Project Skills under each conversation member Bot's Memory Repository `.agents/skills` or `.dsh/skills`. Each row shows its owner (Bot avatar or Skill icon), `/name` and a one-line description.
+- `rank` orders Skills by name prefix, then name subsequence, then owning Bot display name or slug substring, then description substring; order within each tier stays stable. Typing `alice` lists all of Alice's Skills.
 - `chip` inserts the picked Skill as one token; Backspace removes it whole; the draft keeps it across reloads.
 - `snapshot` stores `{name, owner, path, content}` with the message; editing or deleting the `SKILL.md` later does not change the sent message.
 - `deliver` gives the Bot the message with `[/name]` and the `<skill_invocation>` block, so DSH does not load the Skill a second time.
-- `cli` `deepseekbot skills-list --bot <slug> [--query]` and `send <slug> --body "/name …" --skill <owner>/<name>`.
+- `group-routing` invokes Alice's Skill while mentioning Bob: only Bob wakes under a mentions-only wake policy, and his Bot Inbox text contains Alice's snapshot, owner `bot:alice` and path. Without a mention, existing per-Bot Group wake policies decide delivery; the Skill owner is never implicitly addressed.
+- `im-mirror` falls back to plain `/name` on Discord/Lark, without `[/name]`, XML or snapshot content.
+- `command-collision` keeps commands first and reserves a bare command name while its same-named Skill row remains selectable. External IM DM command intake recognizes supported bare commands before ordinary intake; Channel/CLI sends require explicit Skill tokens or `--skill`.
+- `cli` `deepseekbot skills-list --bot <slug> [--query]` (or `--channel <id>`) and `send <slug> --body "/name …" --skill <owner>/<name>`.
 
 ## How to get to it (user POV)
 
 - Enter Bot mode, open a Bot DM, type `/` and a few letters in the composer, choose a Skill with the arrow keys and Enter (or click it), write the rest of the message, press Enter.
 
+- Open a Group Channel with Alice and Bob, type `/`, inspect both owners and the single Host Skill entry, then pick Alice's Skill and explicitly mention Bob.
+
 ## Driving it with verify-app
 
 Preconditions: `launch` reported a model credential and `doctor` passes.
@@ -25,8 +30,20 @@ Preconditions: `launch` reported a model credential and `doctor` passes.
 4. **Canonical state.** `rpc botharness/channelTimeline '{"channelId":"<dm id>"}'`: the Human entry has `skillInvocations[0]` with `owner {kind: bot}`, the `path` of the seeded file and its body; the Bot entry replies `SKILL_QA_7731`.
 5. **CLI.** `deepseekbot skills-list --bot <slug> --query qa` lists `qa-echo`; `deepseekbot send <slug> --body "/qa-echo" --skill bot:<slug>/qa-echo` returns a committed reply.
 
+### Group Channel path
+
+1. Seed `qa-echo` in Alice's Memory Repository and a distinct Skill in Bob's. Create or open a Group containing Alice and Bob. Set both Group wake policies to mentions-only so an ordinary digest wake cannot obscure the recipient check.
+2. Type `/` in the Group composer: both Project Skills appear with their owning Bot avatars; each Host Skill appears once. Type Alice's display name or slug: all her Skills remain, below name matches and above description-only matches.
+3. Pick Alice's `qa-echo` row, explicitly mention Bob using the mention picker, and send. Bob alone replies `SKILL_QA_7731`. Read `rpc botharness/channelTimeline '{"channelId":"<group id>"}'`: the Human message stores owner `{kind: bot, botSlug: alice}`, path and original content; canonical Bot Inbox/Admission state identifies Bob as the recipient, not Alice.
+4. Repeat without a mention: with both policies mentions-only neither Bot wakes. Change Bob's policy to all messages and repeat: Bob wakes through ordinary routing with Alice's identical snapshot. Editing/deleting the Skill after send leaves the stored snapshot unchanged.
+5. `deepseekbot skills-list --channel <group id> --query alice` lists Alice's Skills by display name and slug. Send with `verify-app rpc botharness/channelSend '{"channelId":"<group id>","body":"/qa-echo","skillInvocations":[{"name":"qa-echo","owner":{"kind":"bot","botSlug":"alice"},"start":0,"end":8}]}'` to exercise the same snapshot and ordinary routing without the UI.
+6. For a same-named command/Skill, inspect commands-first ordering, exercise the bare command, then explicitly select the Skill row and inspect the snapshot.
+7. Through the designated Discord/Lark test destination and the IM browser QA procedure, inspect a mirrored Human message: it shows plain `/qa-echo`, with no brackets, XML or snapshot content. Correlate it with Source Event/Outbox facts. The automated provider-carrier tests alone do not verify browser rendering; do not infer that a Group bridge automatically forwards Human messages.
+
 ## Gotchas
 
-- The `/` menu reads the catalog when the DM opens. A Skill written while the DM is open appears after reopening the DM until S3 (live refresh) lands.
+- The `/` menu reads the catalog when the conversation opens. A Skill written while the conversation is open appears after reopening it until S3 (live refresh) lands.
 - DSH discovers `<memory>/.agents/skills` because the Orchestrator's cwd is the Memory Repository; a Skill placed in `.botharness/` is not found.
 - Recalling an older message with ArrowUp, or restoring a failed send, keeps `/name` as plain text without the chip.
+
+- S2 non-UI automation covers catalog membership, ranking, routing, snapshot delivery and plain-text provider sends. Group menu avatars, command selection and Discord/Lark browser rendering still require the UI/IM journey above.

changes/4233-group-skill-ranking.md

@@ -0,0 +1,11 @@
+---
+section: Added
+---
+
+## en
+
+- Added owning Bot display name and slug matching to Skill search in Group Channels and `skills-list`, after Skill name matches and before description matches; invoking another Bot's Skill preserves normal routing and the shared text snapshot ([#4233](https://github.com/BotHarness/BotHarness/issues/4233)).
+
+## zh
+
+- 群聊与 `skills-list` 的 Skill 搜索新增所属 Bot 显示名和 slug 匹配，排在 Skill 名称匹配之后、描述匹配之前；调用其他 Bot 的 Skill 仍遵循原有路由，并保留共享文本快照（[#4233](https://github.com/BotHarness/BotHarness/issues/4233)）。

docs/bot-cli.md

@@ -238,7 +238,7 @@ deepseekbot folder-browse [<folder-path>] [--offset <n>]
 deepseekbot grant-create <bot-id> --workspace <workspace-id>
 ```
 
-`skills-list` returns the Skills the `/` menu offers in that conversation, ranked like the menu by `--query` (name prefix, then name subsequence, then description): Host Skills (`owner.kind` `host`) and the Project Skills under each PersonaBot's Memory Repository `.agents/skills` or `.dsh/skills` (`owner.kind` `bot`), as that Bot's Orchestrator sees them. Pass `--channel <channel-id>` instead of `--bot` for any Channel. `send --skill <owner>/<name>` (owner `host` or `bot:<bot-id>`, repeatable) turns the first unused `/name` token in the body into a Skill Invocation for each flag, in flag order (the token must start and end at whitespace or a body boundary; the send is refused if no unused occurrence remains): the Host stores the Skill's text as it is now with the message and the Bot reads it as a `<skill_invocation>` block. The Skill does not choose who replies; a Skill that is gone refuses with `skill-unavailable`, and Skill text longer than 64,000 characters refuses with `skill-too-large` instead of being truncated. See [ADR-0346](adr/0346-skill-invocation-is-snapshotted-text-that-addresses-no-one.md).
+`skills-list` returns the Skills the `/` menu offers in that conversation, ranked like the menu by `--query` (name prefix, then name subsequence, then owning Bot display name or slug substring, then description substring): Host Skills (`owner.kind` `host`) and the Project Skills under each PersonaBot's Memory Repository `.agents/skills` or `.dsh/skills` (`owner.kind` `bot`), as that Bot's Orchestrator sees them. Pass `--channel <channel-id>` instead of `--bot` for any Channel. `send --skill <owner>/<name>` (owner `host` or `bot:<bot-id>`, repeatable) turns the first unused `/name` token in the body into a Skill Invocation for each flag, in flag order (the token must start and end at whitespace or a body boundary; the send is refused if no unused occurrence remains): the Host stores the Skill's text as it is now with the message and the Bot reads it as a `<skill_invocation>` block. The Skill does not choose who replies; a Skill that is gone refuses with `skill-unavailable`, and Skill text longer than 64,000 characters refuses with `skill-too-large` instead of being truncated. See [ADR-0346](adr/0346-skill-invocation-is-snapshotted-text-that-addresses-no-one.md).
 
 `tool-approval-status` returns `status` and, while the card is still `pending`, its `expiresAt`, the time the Web card counts down to. `--outcome allowed-always-kind` is the `secret_cli` card's "Always allow this kind": it allows this call and saves a rule for the same program, account and environment variable names, which `tool-approval-rules` lists with kind `secret-cli`; any other card refuses it.
 
@@ -479,22 +479,14 @@ printf '{"token":"…"}' | deepseekbot im-app-add discord --credentials-stdin
 deepseekbot im-app-status discord
 deepseekbot im-bind <bot-id> --app <app-id>
 deepseekbot im-status <bot-id>
-deepseekbot im-status <bot-id> --read-mode saved
-deepseekbot messaging-grants <bot-id> --read-mode status
-deepseekbot channel-bridges <channel-id> --read-mode saved
 deepseekbot channel-ingests <channel>
 deepseekbot channel-connect <channel> --bot <bot-id> --conversation <external-id>
-deepseekbot im-reconnect <bot-id> --app <app-id>
 deepseekbot im-unbind <bot-id> --app <app-id>
 deepseekbot im-app-remove discord --app <app-id>
 ```
 
 `im-app-add` covers token-based platforms (Discord, Slack, QQ, DingTalk, WeCom, Matrix); Feishu and WeChat keep `im-authorize`. Credentials arrive only on stdin and are never printed; output lists each app's id, name, connection state and error code. `im-bind` binds a connected app (the `id` from `im-app-status`, also the `ref` in `im-status`) to a Bot, and `im-status` shows each binding's `reception`. `channel-connect` attaches an external group conversation to a Channel as context through that Bot's binding; the conversation must already be known to the binding, so mention the app there once and pick the id from the `candidates` in `channel-ingests`. Otherwise the Host refuses with `conversation-unavailable`.
 
-`im-status` and `messaging-grants` include process-local `recovery` metadata: `retrying` with `reason`, `attempt`, `nextRetryAt`, or `action-required` with `reason`. Temporary provider readiness failures retry exponentially from 250 ms to a 30 s cap until recovery. Host restart reconstructs receivers from enabled bindings and grants. `im-reconnect` uses the existing Owner and current revision without changing conversation access. Failed mutations are reported, never automatically replayed.
-
-`im-status`, `messaging-grants` and `channel-bridges` support `--read-mode saved|status`. `saved` returns local configuration promptly with `checks: pending`; capability/availability fields are not verified permissions. `status` checks only bound accounts and grants with `checks: complete`, sharing account inspections only within the current request. Omitting the flag retains full app/setup discovery. Mutations reject this flag and keep their fresh Owner checks. See [ADR-0343](adr/0343-im-settings-project-saved-configuration-before-live-checks.md).
-
 ### IM messaging, approvals and failure alerts from the CLI
 
 ```bash
@@ -2114,5 +2106,3 @@ deepseekbot roles update <roleId> --models all [--as <username>]
 deepseekbot roles update <roleId> --models provider/model,provider/model [--as <username>]
 deepseekbot models allowed [--as <username>]
 ```
-
-Tool effect declarations (E1) preserve the existing tool commands and approval outcomes. Full Access, saved rules, Secrets, device targets and Workspace Grants retain their existing checks; no new route or CLI authority is introduced.

docs/bot-cli.zh.md

@@ -233,7 +233,7 @@ deepseekbot folder-browse [<文件夹路径>] [--offset <n>]
 deepseekbot grant-create <bot-id> --workspace <workspace-id>
 ```
 
-`skills-list` 返回该对话 `/` 菜单提供的 Skill，并按 `--query` 以与菜单相同的规则排序（名称前缀、名称子序列、描述）：包括 Host Skill（`owner.kind` 为 `host`），以及各 PersonaBot Memory Repository 下 `.agents/skills` 或 `.dsh/skills` 中的 Project Skill（`owner.kind` 为 `bot`）。任意 Channel 可用 `--channel <channel-id>` 代替 `--bot`。`send --skill <owner>/<name>`（owner 为 `host` 或 `bot:<bot-id>`，可重复）按参数顺序，为每个参数把正文中第一个尚未选中的 `/name` token 变成 Skill Invocation（token 两端必须为空白或正文边界；没有未用匹配时拒绝发送）：Host 把该 Skill 当前的文本随消息保存，Bot 以 `<skill_invocation>` 块读到它。Skill 不决定由谁回复；已不存在的 Skill 会以 `skill-unavailable` 拒绝；Skill 文本超过 64,000 个字符时会以 `skill-too-large` 拒绝，不会截断。见 [ADR-0346](adr/0346-skill-invocation-is-snapshotted-text-that-addresses-no-one.md)。
+`skills-list` 返回该对话 `/` 菜单提供的 Skill，并按 `--query` 以与菜单相同的规则排序（名称前缀、名称子序列、所属 Bot 显示名或 slug 子串、描述子串）：包括 Host Skill（`owner.kind` 为 `host`），以及各 PersonaBot Memory Repository 下 `.agents/skills` 或 `.dsh/skills` 中的 Project Skill（`owner.kind` 为 `bot`）。任意 Channel 可用 `--channel <channel-id>` 代替 `--bot`。`send --skill <owner>/<name>`（owner 为 `host` 或 `bot:<bot-id>`，可重复）按参数顺序，为每个参数把正文中第一个尚未选中的 `/name` token 变成 Skill Invocation（token 两端必须为空白或正文边界；没有未用匹配时拒绝发送）：Host 把该 Skill 当前的文本随消息保存，Bot 以 `<skill_invocation>` 块读到它。Skill 不决定由谁回复；已不存在的 Skill 会以 `skill-unavailable` 拒绝；Skill 文本超过 64,000 个字符时会以 `skill-too-large` 拒绝，不会截断。见 [ADR-0346](adr/0346-skill-invocation-is-snapshotted-text-that-addresses-no-one.md)。
 
 `tool-approval-status` 返回 `status`；卡片仍为 `pending` 时还返回 `expiresAt`，即网页卡片倒计时的截止时间。`--outcome allowed-always-kind` 对应 `secret_cli` 卡片上的“总是允许此类”：批准这一次，并为同一程序、账号和环境变量名保存一条规则；`tool-approval-rules` 会以 `secret-cli` 类型列出它。其他卡片会拒绝这个选项。
 
@@ -463,22 +463,14 @@ printf '{"token":"…"}' | deepseekbot im-app-add discord --credentials-stdin
 deepseekbot im-app-status discord
 deepseekbot im-bind <bot-id> --app <app-id>
 deepseekbot im-status <bot-id>
-deepseekbot im-status <bot-id> --read-mode saved
-deepseekbot messaging-grants <bot-id> --read-mode status
-deepseekbot channel-bridges <channel-id> --read-mode saved
 deepseekbot channel-ingests <channel>
 deepseekbot channel-connect <channel> --bot <bot-id> --conversation <external-id>
-deepseekbot im-reconnect <bot-id> --app <app-id>
 deepseekbot im-unbind <bot-id> --app <app-id>
 deepseekbot im-app-remove discord --app <app-id>
 ```
 
 `im-app-add` 适用于基于 token 的平台（Discord、Slack、QQ、钉钉、企业微信、Matrix）；飞书和微信仍用 `im-authorize`。凭证只从标准输入读取，绝不打印；输出列出每个应用的 id、名称、连接状态和错误码。`im-bind` 把已连接的应用（`im-app-status` 里的 `id`，即 `im-status` 里的 `ref`）绑定到 Bot，`im-status` 显示每个绑定的 `reception`。`channel-connect` 通过该 Bot 的绑定，把外部群聊作为上下文接入 Channel；这个会话必须已被该绑定知晓，所以先在那里 @ 一次应用，再从 `channel-ingests` 的 `candidates` 里选 id。否则 Host 会以 `conversation-unavailable` 拒绝。
 
-`im-status` 和 `messaging-grants` 返回进程内的 `recovery` 状态：`retrying` 包含 `reason`、`attempt`、`nextRetryAt`；`action-required` 包含 `reason`。Provider 暂未就绪时从 250 毫秒开始指数退避，上限 30 秒，持续重试直至恢复。Host 重启后从启用的绑定和授权重建接收器。`im-reconnect` 通过现有 Owner 和当前修订号重连，不改变会话权限。失败的修改会返回错误，不会自动重放。
-
-`im-status`、`messaging-grants` 和 `channel-bridges` 支持 `--read-mode saved|status`。`saved` 快速返回本地配置，标记 `checks: pending`；其中能力和可用性字段不代表已验证的权限。`status` 只检查绑定的账号和会话，返回 `checks: complete`，账号检查仅在本次请求内共享。不传此参数时仍进行完整应用和接入方式发现。修改操作拒绝此参数，并保留实时 Owner 检查。见 [ADR-0343](adr/0343-im-settings-project-saved-configuration-before-live-checks.md)。
-
 ### 用命令行管理 IM 消息、审批和失败提醒
 
 ```bash
@@ -2055,5 +2047,3 @@ token 或原始响应。
 Web 手动检查请求最多等待 25 秒，关闭自检卡片会取消请求。公网 WebSocket 在验证 Host 身份前将握手消息限制为 1 KiB，并禁用压缩。
 
 公网阶段仅对 DNS、网络、超时及 HTTP 429/5xx 短暂失败重试：最多四次，间隔 1/3/6 秒，仍受同一个 20 秒总时限约束。证书或身份失败不重试。每次自检完成后，界面刷新一次账号电脑列表。
-
-工具效果声明（E1）保持既有工具命令与审批结果。Full Access、保存规则、Secrets、设备目标和 Workspace Grant 仍使用原检查；不新增 route 或 CLI 权限。