@pellux/goodvibes-agent 1.0.19 → 1.0.21

package/docs/README.md

@@ -22,7 +22,7 @@ Important baseline constraints:
 - Agent depends on `@pellux/goodvibes-sdk@0.33.35`.
 - Agent connects to a GoodVibes host owned outside this product.
 - Agent does not start, stop, restart, install, uninstall, or own the connected GoodVibes host.
-- User-facing Agent workspace actions, single workspace-action lookup, built-in panels and single-panel lookup, modal/overlay/picker UI surfaces and single-surface lookup, named operator surfaces, top-level CLI mirrors and single-mirror lookup, fixed shortcuts, configurable keybindings and single-keybinding lookup, slash commands, single-command lookup with policy metadata, model tools and individual model tool schemas, settings and single-setting lookup, local registries, connected-host capability boundaries, single connected-host capability inspection, and live connected-host readiness are model-visible through Agent-owned tools. Visible UI routing includes the command browser, reasoning-effort picker, live process output, runtime activity, settings, workspaces, and pickers the user can open. Local memory, note, persona, skill, and routine create editors can also run through confirmed workspace actions. Mutations and visible UI routing remain explicit and confirmation-gated, and ambiguous lookup matches are refused with candidates instead of guessed.
+- User-facing Agent workspace actions, single workspace-action lookup/execution, built-in panels and single-panel lookup, modal/overlay/picker UI surfaces and single-surface lookup, named operator surfaces, top-level CLI mirrors and single-mirror lookup, fixed shortcuts, configurable keybindings, single-keybinding lookup, shell-safe keybinding execution, slash commands, single-command lookup with policy metadata, model tools and individual model tool schemas, settings and single-setting lookup, local registries, connected-host capability boundaries, single connected-host capability inspection, and live connected-host readiness are model-visible through Agent-owned tools. Visible UI routing includes the command browser, reasoning-effort picker, live process output, runtime activity, settings, workspaces, and pickers the user can open. Local memory, note, persona, skill, and routine create editors can also run through confirmed workspace actions. Mutations, visible UI routing, and keybinding execution remain explicit and confirmation-gated; prompt-editor-only shortcuts stay direct user interaction; and ambiguous lookup matches are refused with candidates instead of guessed.
 - Agent Knowledge uses only `/api/goodvibes-agent/knowledge/*`; there is no default knowledge or non-Agent product fallback.
 - Agent supports isolated Agent homes with `GOODVIBES_AGENT_HOME=<path>` and named profile homes with `goodvibes-agent profiles create <name> --template <starter> --yes` plus `--agent-profile <name>`.
 - Agent supports connected-host URL overrides with `--runtime-url http://host:port` or `GOODVIBES_AGENT_RUNTIME_URL=http://host:port`; these only change the Agent connection target.

package/docs/getting-started.md

@@ -44,7 +44,7 @@ After setup has been applied once, the TUI opens directly into the Agent operato
 
 Press `/` inside the Agent workspace to search every workspace action by name, category, command, or detail. Use that finder before reaching for shell commands; CLI subcommands are scriptable mirrors of these TUI workflows.
 
-The model can inspect and use the same harness surface through Agent-owned tools. `agent_harness` exposes workspace action discovery, single workspace-action inspection by action id, command, or lookup text, built-in panel discovery/routing by panel id or lookup text, modal/overlay/picker UI surface discovery/routing by surface id or lookup text, top-level CLI mirror discovery and single-mirror lookup by command string, command token, or lookup text, fixed shortcut and configurable keybinding discovery, keybinding inspection/mutation by action id or lookup text, slash-command discovery, single slash-command inspection by typed command or lookup text, command policy metadata, command execution with confirmation, model tool discovery and single-tool schema inspection, settings inspection/mutation by key or lookup text, connected-host capability inventory, single connected-host capability inspection, and live connected-host readiness posture. CLI mirror modes are read-only catalog/parser inspection and point the model to the matching in-process tool, workspace action, setting mode, or slash-command mirror. Panel modes expose catalog/open state and route visible panel/workspace changes through the current Agent shell bridge with confirmation. UI surface modes expose help, shortcuts, command browser, conversation search, prompt-history search, slash-command mode, file picker, block actions, context, runtime activity, live process output, bookmarks, model/provider/reasoning-effort pickers, TTS provider/voice pickers, session/profile pickers, the panel-picker compatibility route, security/knowledge/subscription operator surfaces, settings, MCP workspace, onboarding, and Agent workspace entrypoints; `open_ui_surface` is confirmation-gated and only performs visible shell navigation. Shortcut modes expose the fixed runtime/editor shortcuts and the live keybindings table; confirmed keybinding edits write the same `keybindings.json` file the user edits and reload the runtime manager. Workspace action discovery can include editor field schemas with `includeParameters:true`, including starter-template defaults for profile creation and selected-routine defaults for routine schedule promotion when `recordId` is supplied; single-action lookup uses the same user-facing search fields and returns lookup metadata plus the editor schema; single-CLI-mirror, single-panel, single-surface, single-keybinding, and single-setting lookup use their catalog search fields, return lookup metadata, and refuse ambiguous route or mutation requests with candidates. Selection-based local workspace actions accept a local `recordId`, so the model can use the same note promotion and local registry flows as the TUI. Direct local create editors for memory, notes, personas, skills, and routines can execute from submitted fields through `run_workspace_action` and `agent_local_registry` with confirmation. First-class model tools cover the main product workflows directly: Agent Knowledge, Agent Knowledge ingest, Agent-local memory/notes/personas/skills/routines, operator actions, notifications, channel sends, reminders, generated media, and work plans.
+The model can inspect and use the same harness surface through Agent-owned tools. `agent_harness` exposes workspace action discovery, single workspace-action inspection and execution by action id, command, or lookup text, built-in panel discovery/routing by panel id or lookup text, modal/overlay/picker UI surface discovery/routing by surface id or lookup text, top-level CLI mirror discovery and single-mirror lookup by command string, command token, or lookup text, fixed shortcut and configurable keybinding discovery, keybinding inspection/mutation by action id or lookup text, confirmation-gated shell-safe keybinding execution, slash-command discovery, single slash-command inspection by typed command or lookup text, command policy metadata, command execution with confirmation, model tool discovery and single-tool schema inspection, settings inspection/mutation by key or lookup text, connected-host capability inventory, single connected-host capability inspection, and live connected-host readiness posture. CLI mirror modes are read-only catalog/parser inspection and point the model to the matching in-process tool, workspace action, setting mode, or slash-command mirror. Panel modes expose catalog/open state and route visible panel/workspace changes through the current Agent shell bridge with confirmation. UI surface modes expose help, shortcuts, command browser, conversation search, prompt-history search, slash-command mode, file picker, block actions, context, runtime activity, live process output, bookmarks, model/provider/reasoning-effort pickers, TTS provider/voice pickers, session/profile pickers, the panel-picker compatibility route, security/knowledge/subscription operator surfaces, settings, MCP workspace, onboarding, and Agent workspace entrypoints; `open_ui_surface` is confirmation-gated and only performs visible shell navigation. Shortcut modes expose the fixed runtime/editor shortcuts and the live keybindings table; keybinding descriptors include a model operation route, `run_keybinding` executes supported shell-safe equivalents only with confirmation, prompt-editor-only or terminal-selection shortcuts remain direct user interaction, and confirmed keybinding edits write the same `keybindings.json` file the user edits and reload the runtime manager. Workspace action discovery can include editor field schemas with `includeParameters:true`, including starter-template defaults for profile creation and selected-routine defaults for routine schedule promotion when `recordId` is supplied; single-action lookup and execution use the same user-facing search fields, return lookup metadata where the action result carries descriptors, and refuse ambiguous run requests with candidates. Single-CLI-mirror, single-panel, single-surface, single-keybinding, and single-setting lookup use their catalog search fields, return lookup metadata, and refuse ambiguous route or mutation requests with candidates. Selection-based local workspace actions accept a local `recordId`, so the model can use the same note promotion and local registry flows as the TUI. Direct local create editors for memory, notes, personas, skills, and routines can execute from submitted fields through `run_workspace_action` and `agent_local_registry` with confirmation. First-class model tools cover the main product workflows directly: Agent Knowledge, Agent Knowledge ingest, Agent-local memory/notes/personas/skills/routines, operator actions, notifications, channel sends, reminders, generated media, and work plans.
 
 Use the Artifacts area for concrete files and generated output: attach images to prompts, export conversations or saved sessions, ingest local source files and URL lists into Agent Knowledge, import bookmarks or browser history, inspect source records, and generate media artifacts from confirmed prompts.
 
@@ -122,7 +122,7 @@ Use `/channels` inside the TUI for channel readiness and exact confirmed sends.
 
 The main assistant conversation can perform narrow confirmed operator actions when the user explicitly asks for a specific target: approve/deny/cancel one approval, run/pause/resume one automation job, cancel/retry one automation run, or run one schedule. Those calls use only public connected-host routes, require confirmation, and do not create, edit, or delete automation definitions.
 
-When the main assistant conversation needs to change Agent settings, change Agent keybindings, open a visible UI surface, or run a user-facing harness action, it should use `agent_harness` rather than generic SDK context/settings tools. Panel, UI surface, keybinding, and setting descriptors report lookup metadata when resolved from `target` or `query`; ambiguous route or mutation requests return candidates instead of guessing. Setting descriptors also report writability, workspace visibility, and lock reasons. Keybinding descriptors report fixed shortcuts, resolved bindings, defaults, custom state, and config path; UI surface descriptors report the visible shell opener, preferred model route, and confirmation policy; secret-backed settings store raw values through the secret manager; connected-host lifecycle settings stay read-only in Agent; and destructive local record deletion requires explicit confirmation.
+When the main assistant conversation needs to change Agent settings, inspect/change/run supported Agent keybindings, open a visible UI surface, or run a user-facing harness action, it should use `agent_harness` rather than generic SDK context/settings tools. Panel, UI surface, keybinding, and setting descriptors report lookup metadata when resolved from `target` or `query`; ambiguous route or mutation requests return candidates instead of guessing. Setting descriptors also report writability, workspace visibility, and lock reasons. Keybinding descriptors report fixed shortcuts, resolved bindings, defaults, custom state, config path, and model-operation route; unsupported prompt-editor-only shortcuts stay direct user interaction instead of hidden model operations. UI surface descriptors report the visible shell opener, preferred model route, and confirmation policy; secret-backed settings store raw values through the secret manager; connected-host lifecycle settings stay read-only in Agent; and destructive local record deletion requires explicit confirmation.
 
 ## Connected GoodVibes Host
 

package/docs/tools-and-commands.md

@@ -43,28 +43,28 @@ Host-management and coding-first commands that would imply connected-host lifecy
 
 The main Agent model has an Agent-owned harness bridge rather than generic SDK settings/context control. Use these model tools as the supported parity layer:
 
-- `agent_harness`: inspect workspace categories/actions and one workspace action by action id, command, or lookup text, inspect built-in panels and one panel by id or lookup text, inspect modal/overlay/picker UI surfaces and named operator surfaces by id or lookup text, inspect top-level CLI mirrors and one mirror by command string, command token, or lookup text, inspect fixed shortcuts and configurable keybindings, inspect or change one keybinding by action id or lookup text, inspect slash commands and one slash command by typed command or lookup text with policy metadata, run concrete workspace or slash-command mirrors with confirmation, inspect model tools or one model tool schema, inspect or change Agent settings by exact key or lookup text, inspect connected-host capability boundaries or one connected-host capability, and inspect live connected-host readiness.
+- `agent_harness`: inspect workspace categories/actions and inspect or run one workspace action by action id, command, or lookup text, inspect built-in panels and one panel by id or lookup text, inspect modal/overlay/picker UI surfaces and named operator surfaces by id or lookup text, inspect top-level CLI mirrors and one mirror by command string, command token, or lookup text, inspect fixed shortcuts and configurable keybindings, inspect/change/run one keybinding by action id or lookup text where a shell-safe model route exists, inspect slash commands and one slash command by typed command or lookup text with policy metadata, run concrete slash-command mirrors with confirmation, inspect model tools or one model tool schema, inspect or change Agent settings by exact key or lookup text, inspect connected-host capability boundaries or one connected-host capability, and inspect live connected-host readiness.
 - `agent_local_registry`: inspect and maintain Agent-local memory, notes, personas, skills, skill bundles, and routines. Deleting local records requires `confirm:true` and `explicitUserRequest`.
 - `agent_knowledge` and `agent_knowledge_ingest`: ask/search and ingest into the isolated Agent Knowledge segment.
 - `agent_operator_briefing` and `agent_operator_action`: inspect connected work/approval/automation posture, or run exact confirmed approval/automation actions.
 - `agent_work_plan`: keep the visible Agent-local work plan current from the conversation.
 - `agent_channel_send`, `agent_notify`, `agent_reminder_schedule`, and `agent_media_generate`: perform confirmed external delivery, notification, reminder, or media actions when the user explicitly asks.
 
-`agent_harness` discovery modes are read-only. `summary` reports the model access map; `panels` lists the built-in panel catalog, and `panel` resolves one panel by `panelId`, `target`, or `query` with current open/focused state plus its matching Agent workspace route; `ui_surfaces` lists modal, overlay, picker, and workspace entrypoints, and `ui_surface` resolves one by `surfaceId`, `target`, or `query` with shell-opener availability and preferred model routes; `cli_commands` lists top-level package CLI mirror metadata, and `cli_command` resolves one mirror by `cliCommand`, `command`, `commandName`, `target`, or `query`, returning parser output for concrete invocations, blocked command tokens, lookup metadata, and preferred in-process model routes; `shortcuts` returns fixed runtime/editor shortcuts plus configurable keybindings; `keybindings` lists the live resolved keybinding table, and `keybinding`, `set_keybinding`, and `reset_keybinding` resolve one action by `actionId`, `target`, `key`, or `query` with default bindings, custom state, config path, and lookup metadata; `commands` lists slash-command descriptions; `command` returns one slash-command detail by `command`, `commandName`, `target`, or `query`, including parsed arguments plus effect/confirmation/preferred-tool/boundary policy metadata; `workspace_actions` lists Agent workspace actions and can inline editor field schemas with `includeParameters:true`; `workspace_action` returns one action by `actionId`, `command`, `target`, or `query`, using the same user-facing action-search fields and returning lookup metadata plus its editor schema; `tools` lists model tool definitions and can inline JSON schemas with `includeParameters:true`; `tool` returns one model tool schema by `toolName`, `target`, or `query`; `settings` returns setting descriptors plus setting policy; `get_setting`, `set_setting`, and `reset_setting` resolve one setting by `key`, `target`, or `query`, return lookup metadata on success, and refuse ambiguous matches with candidate settings; `connected_host` returns the connected-host route families, allowed capabilities, blocked capabilities, and first-class tool availability; `connected_host_capability` returns one allowed or blocked connected-host capability by `capabilityId`, `target`, or `query` with related route families and boundary text; and `connected_host_status` performs a live read-only check of the connected-host status and Agent Knowledge status routes and reports endpoint bindings, token posture, SDK compatibility, route readiness, and findings without printing token values.
+`agent_harness` discovery modes are read-only. `summary` reports the model access map; `panels` lists the built-in panel catalog, and `panel` resolves one panel by `panelId`, `target`, or `query` with current open/focused state plus its matching Agent workspace route; `ui_surfaces` lists modal, overlay, picker, and workspace entrypoints, and `ui_surface` resolves one by `surfaceId`, `target`, or `query` with shell-opener availability and preferred model routes; `cli_commands` lists top-level package CLI mirror metadata, and `cli_command` resolves one mirror by `cliCommand`, `command`, `commandName`, `target`, or `query`, returning parser output for concrete invocations, blocked command tokens, lookup metadata, and preferred in-process model routes; `shortcuts` returns fixed runtime/editor shortcuts plus configurable keybindings; `keybindings` lists the live resolved keybinding table, and `keybinding`, `run_keybinding`, `set_keybinding`, and `reset_keybinding` resolve one action by `actionId`, `target`, `key`, or `query` with default bindings, custom state, config path, lookup metadata, and model-operation route metadata; `commands` lists slash-command descriptions; `command` returns one slash-command detail by `command`, `commandName`, `target`, or `query`, including parsed arguments plus effect/confirmation/preferred-tool/boundary policy metadata; `workspace_actions` lists Agent workspace actions and can inline editor field schemas with `includeParameters:true`; `workspace_action` and `run_workspace_action` resolve one action by `actionId`, `command`, `target`, or `query`, using the same user-facing action-search fields; inspection returns lookup metadata plus editor schema, and execution refuses ambiguous requests with candidate actions before any effect; `tools` lists model tool definitions and can inline JSON schemas with `includeParameters:true`; `tool` returns one model tool schema by `toolName`, `target`, or `query`; `settings` returns setting descriptors plus setting policy; `get_setting`, `set_setting`, and `reset_setting` resolve one setting by `key`, `target`, or `query`, return lookup metadata on success, and refuse ambiguous matches with candidate settings; `connected_host` returns the connected-host route families, allowed capabilities, blocked capabilities, and first-class tool availability; `connected_host_capability` returns one allowed or blocked connected-host capability by `capabilityId`, `target`, or `query` with related route families and boundary text; and `connected_host_status` performs a live read-only check of the connected-host status and Agent Knowledge status routes and reports endpoint bindings, token posture, SDK compatibility, route readiness, and findings without printing token values.
 
 `open_ui_surface` is a confirmation-gated visible navigation mode for the same shell surfaces the user can open: Agent workspace, settings, MCP workspace, model/provider/reasoning-effort pickers, TTS provider/voice pickers, session/profile pickers, the panel-picker compatibility route, security/knowledge/subscription operator surfaces, conversation search, prompt-history search, slash-command mode, command browser, file picker, block actions, bookmarks, context inspector, runtime activity monitor, live process output, help, shortcuts, and onboarding. It does not perform hidden operations; use first-class model tools, settings modes, workspace actions, or confirmed slash-command mirrors for actual state changes. Ambiguous UI surface lookup text is refused with candidate surfaces instead of routed.
 
 `cli_command` is a read-only inspection mode. Concrete CLI strings are parsed with redacted config overrides; descriptive lookup text searches the same `cli_commands` catalog and returns one match or candidate mirrors when broad.
 
-`set_keybinding` and `reset_keybinding` are confirmation-gated control modes. They write the same Agent `keybindings.json` file exposed by `/keybindings`, reload the runtime keybinding manager, and leave fixed runtime/editor shortcuts read-only. Ambiguous keybinding lookup text is refused with candidate actions instead of guessed.
+`run_keybinding`, `set_keybinding`, and `reset_keybinding` are confirmation-gated control modes. `run_keybinding` executes only keybinding actions with a faithful current-shell route, such as cancel generation, clear screen, open/focus/dismiss visible panel workspace routes, open conversation or prompt-history search, paste through the existing clipboard handler, or open the visible block-action surface. Prompt-editor-only controls, terminal text selection, category cycling, and reserved shortcuts stay visible in descriptors but return unsupported/direct-interaction metadata instead of pretending a hidden model operation exists. `set_keybinding` and `reset_keybinding` write the same Agent `keybindings.json` file exposed by `/keybindings`, reload the runtime keybinding manager, and leave fixed runtime/editor shortcuts read-only. Ambiguous keybinding lookup text is refused with candidate actions instead of guessed.
 
 `open_panel` is a confirmation-gated UI routing mode. It hands a visible panel/workspace route to the current Agent shell bridge and does not mutate connected-host lifecycle, listener posture, or external accounts. Ambiguous panel lookup text is refused with candidate panels instead of routed.
 
 CLI mirror modes are catalog and parser inspection only. When the user asks the model to operate from inside the main conversation, use the returned first-class model tool, workspace action, setting mode, or confirmed slash-command mirror instead of launching a hidden nested `goodvibes-agent` process.
 
-Setting writes, setting resets, keybinding writes/resets, UI surface routing, slash-command invocation, workspace-action invocation, local record deletion, channel sends, notifications, reminders, media generation, and connected-host operator mutations require explicit user request and confirmation. Secret-backed settings are stored through the secret manager, and connected-host lifecycle/listener settings remain read-only in Agent.
+Setting writes, setting resets, supported keybinding actions, keybinding writes/resets, UI surface routing, slash-command invocation, workspace-action invocation, local record deletion, channel sends, notifications, reminders, media generation, and connected-host operator mutations require explicit user request and confirmation. Secret-backed settings are stored through the secret manager, and connected-host lifecycle/listener settings remain read-only in Agent.
 
-Selection-based local workspace actions use the same bridge. `agent_harness` reports the required model tool for each local action; for actions that depend on the TUI selection, call `run_workspace_action` with the selected local `recordId`. Direct local create editors for memory, notes, personas, skills, and routines can execute from submitted `fields` through `run_workspace_action`; the harness validates required fields, requires `confirm:true` and `explicitUserRequest`, and dispatches through `agent_local_registry`. Note promotion actions can prefill and create memory, personas, skills, routines, or isolated Agent Knowledge URL ingests through the matching first-class model tool. Profile creation schemas include the current runtime starter-template inventory, and routine schedule schemas prefill the selected routine when `recordId` or a `routineId` field matches a local routine.
+Selection-based local workspace actions use the same bridge. `agent_harness` reports the required model tool for each local action; for actions that depend on the TUI selection, call `run_workspace_action` with the selected local `recordId` plus an `actionId`, `command`, `target`, or `query` that resolves to one action. Direct local create editors for memory, notes, personas, skills, and routines can execute from submitted `fields` through `run_workspace_action`; the harness validates required fields, requires `confirm:true` and `explicitUserRequest`, and dispatches through `agent_local_registry`. Note promotion actions can prefill and create memory, personas, skills, routines, or isolated Agent Knowledge URL ingests through the matching first-class model tool. Profile creation schemas include the current runtime starter-template inventory, and routine schedule schemas prefill the selected routine when `recordId` or a `routineId` field matches a local routine.
 
 Use first-class Agent tools before falling back to slash-command mirrors. Slash-command execution is for harness parity and scriptable mirrors, not for bypassing Agent product boundaries.
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@pellux/goodvibes-agent",
-  "version": "1.0.19",
+  "version": "1.0.21",
   "private": false,
   "description": "GoodVibes personal operator assistant TUI with a proactive Agent product brain, isolated Agent Knowledge, local profiles, routines, skills, personas, and explicit build delegation.",
   "type": "module",

package/src/tools/agent-harness-keybinding-metadata.ts

@@ -32,6 +32,25 @@ interface HarnessKeybindingArgs {
   readonly limit?: unknown;
 }
 
+type KeybindingOperationEffect =
+  | 'shell-action'
+  | 'visible-ui-navigation'
+  | 'visible-ui-interaction'
+  | 'prompt-editor-state'
+  | 'clipboard-selection'
+  | 'reserved'
+  | 'disabled';
+
+interface KeybindingOperationRoute {
+  readonly supported: boolean;
+  readonly effect: KeybindingOperationEffect;
+  readonly confirmation: string;
+  readonly preferredMode?: 'run_keybinding' | 'open_ui_surface' | 'run_command' | 'direct-user-interaction';
+  readonly surfaceId?: string;
+  readonly command?: string;
+  readonly note: string;
+}
+
 type KeybindingsOverrideFile = Record<string, unknown>;
 
 interface KeybindingLookup {
@@ -214,9 +233,134 @@ function describeBinding(manager: KeybindingsManager, action: KeyAction, combos:
     defaultBindings: defaults.map((combo) => describeCombo(manager, combo)),
     customized,
     source: customized ? 'custom' : 'default',
+    modelOperation: keybindingOperationRoute(action),
   };
 }
 
+function keybindingOperationRoute(action: KeyAction): KeybindingOperationRoute {
+  switch (action) {
+    case 'clear-cancel':
+      return {
+        supported: true,
+        effect: 'shell-action',
+        preferredMode: 'run_keybinding',
+        confirmation: 'agent_harness mode:"run_keybinding" requires confirm:true and explicitUserRequest.',
+        note: 'Routes the cancel-generation side of this shortcut through commandContext.cancelGeneration. Prompt clearing and double-press exit remain direct user interaction.',
+      };
+    case 'screen-clear':
+      return {
+        supported: true,
+        effect: 'shell-action',
+        preferredMode: 'run_keybinding',
+        confirmation: 'agent_harness mode:"run_keybinding" requires confirm:true and explicitUserRequest.',
+        note: 'Routes through commandContext.clearScreen.',
+      };
+    case 'panel-picker':
+      return {
+        supported: true,
+        effect: 'visible-ui-navigation',
+        preferredMode: 'run_keybinding',
+        surfaceId: 'agent-workspace',
+        confirmation: 'agent_harness mode:"run_keybinding" requires confirm:true and explicitUserRequest.',
+        note: 'Opens the same Agent workspace home route as the panel-picker shortcut.',
+      };
+    case 'panel-close':
+      return {
+        supported: true,
+        effect: 'visible-ui-navigation',
+        preferredMode: 'run_keybinding',
+        confirmation: 'agent_harness mode:"run_keybinding" requires confirm:true and explicitUserRequest.',
+        note: 'Closes the active legacy panel if present, then focuses the prompt when the shell bridge exposes focusPrompt.',
+      };
+    case 'panel-close-all':
+      return {
+        supported: true,
+        effect: 'visible-ui-navigation',
+        preferredMode: 'run_keybinding',
+        confirmation: 'agent_harness mode:"run_keybinding" requires confirm:true and explicitUserRequest.',
+        note: 'Closes all legacy panels if present, hides the panel manager, then focuses the prompt when the shell bridge exposes focusPrompt.',
+      };
+    case 'history-search':
+      return {
+        supported: true,
+        effect: 'visible-ui-navigation',
+        preferredMode: 'run_keybinding',
+        surfaceId: 'prompt-history-search',
+        confirmation: 'agent_harness mode:"run_keybinding" requires confirm:true and explicitUserRequest.',
+        note: 'Opens the visible prompt-history search overlay; optional search text may be supplied with value.',
+      };
+    case 'search':
+      return {
+        supported: true,
+        effect: 'visible-ui-navigation',
+        preferredMode: 'run_keybinding',
+        surfaceId: 'conversation-search',
+        confirmation: 'agent_harness mode:"run_keybinding" requires confirm:true and explicitUserRequest.',
+        note: 'Opens the visible conversation search overlay; optional search text may be supplied with value.',
+      };
+    case 'paste':
+      return {
+        supported: true,
+        effect: 'shell-action',
+        preferredMode: 'run_keybinding',
+        command: '/paste',
+        confirmation: 'agent_harness mode:"run_keybinding" requires confirm:true and explicitUserRequest.',
+        note: 'Routes through commandContext.pasteFromClipboard and reports whether text, image, or nothing was pasted.',
+      };
+    case 'block-copy':
+    case 'bookmark':
+    case 'block-save':
+      return {
+        supported: true,
+        effect: 'visible-ui-interaction',
+        preferredMode: 'run_keybinding',
+        surfaceId: 'block-actions',
+        confirmation: 'agent_harness mode:"run_keybinding" requires confirm:true and explicitUserRequest.',
+        note: 'Opens the visible nearest-block actions surface. The exact block action remains an interactive visible-shell selection because it depends on cursor/scroll position.',
+      };
+    case 'panel-tab-next':
+    case 'panel-tab-prev':
+      return {
+        supported: false,
+        effect: 'visible-ui-navigation',
+        preferredMode: 'open_ui_surface',
+        surfaceId: 'agent-workspace',
+        confirmation: 'Use open_ui_surface with confirm:true and an explicit categoryId/target instead of cycling hidden UI state.',
+        note: 'Category cycling depends on the live workspace focus position. The model should open the intended Agent workspace category directly.',
+      };
+    case 'copy-selection':
+      return {
+        supported: false,
+        effect: 'clipboard-selection',
+        preferredMode: 'direct-user-interaction',
+        confirmation: 'No model operation is exposed.',
+        note: 'Terminal text selection is outside the Agent command context. Use transcript/session export or a content-specific model route when a concrete artifact is needed.',
+      };
+    case 'delete-word':
+    case 'line-start':
+    case 'next-error-line-end':
+    case 'kill-line':
+    case 'clear-prompt':
+    case 'undo':
+    case 'redo':
+      return {
+        supported: false,
+        effect: 'prompt-editor-state',
+        preferredMode: 'direct-user-interaction',
+        confirmation: 'No model operation is exposed.',
+        note: 'This shortcut mutates live prompt-buffer cursor/edit state that is not part of the model tool contract.',
+      };
+    case 'replay-panel':
+      return {
+        supported: false,
+        effect: 'reserved',
+        preferredMode: 'direct-user-interaction',
+        confirmation: 'No model operation is exposed.',
+        note: 'This shortcut is reserved and has no current Agent operation route.',
+      };
+  }
+}
+
 function readOverrideFile(configPath: string): KeybindingsOverrideFile {
   if (!existsSync(configPath)) return {};
   const parsed = JSON.parse(readFileSync(configPath, 'utf-8')) as unknown;
@@ -252,7 +396,7 @@ export function listHarnessShortcuts(context: CommandContext, args: HarnessKeybi
     configurableKeybindings: keybindings.keybindings,
     returned: fixed.length + Number(keybindings.returned ?? 0),
     total: totalHarnessShortcuts(context),
-    policy: 'Fixed shortcuts are runtime/editor controls. Configurable keybindings can be changed with set_keybinding/reset_keybinding.',
+    policy: 'Fixed shortcuts are runtime/editor controls. Configurable keybindings can be inspected, supported shell-safe actions can be run with run_keybinding, and bindings can be changed with set_keybinding/reset_keybinding.',
   };
 }
 
@@ -268,7 +412,7 @@ export function listHarnessKeybindings(context: CommandContext, args: HarnessKey
     keybindings: entries,
     returned: entries.length,
     total: manager.getAll().length,
-    policy: 'Reads the live resolved keybindings. set_keybinding/reset_keybinding write the same keybindings.json file the user can edit and reload the runtime manager.',
+    policy: 'Reads the live resolved keybindings, including modelOperation route metadata. run_keybinding executes only supported shell-safe actions. set_keybinding/reset_keybinding write the same keybindings.json file the user can edit and reload the runtime manager.',
   };
 }
 
@@ -284,6 +428,133 @@ export function describeHarnessKeybinding(context: CommandContext, args: Harness
   } : null;
 }
 
+function runUnavailable(action: KeyAction, route: KeybindingOperationRoute, reason?: string): Record<string, unknown> {
+  return {
+    status: 'keybinding_route_unavailable',
+    action,
+    modelOperation: route,
+    ...(reason ? { reason } : {}),
+  };
+}
+
+export function runHarnessKeybinding(context: CommandContext, args: HarnessKeybindingArgs): Record<string, unknown> {
+  const manager = requireKeybindingsManager(context);
+  const resolved = resolveHarnessKeybinding(context, args);
+  if (resolved?.status === 'ambiguous') return { status: 'ambiguous', input: resolved.input, candidates: resolved.candidates };
+  if (!resolved) throw new Error('run_keybinding requires a valid keybinding action id, target, key, or query.');
+  const entry = manager.getAll().find((candidate) => candidate.action === resolved.action);
+  if (!entry) throw new Error(`No live keybinding entry found for ${resolved.action}.`);
+  const route = keybindingOperationRoute(resolved.action);
+  const descriptor = {
+    configPath: manager.getConfigPath(),
+    ...describeBinding(manager, entry.action, entry.combos, resolved.lookup),
+  };
+  if (!route.supported) {
+    return {
+      status: 'unsupported_keybinding_action',
+      action: resolved.action,
+      keybinding: descriptor,
+      modelOperation: route,
+    };
+  }
+
+  switch (resolved.action) {
+    case 'clear-cancel':
+      if (!context.cancelGeneration) return runUnavailable(resolved.action, route, 'commandContext.cancelGeneration is unavailable.');
+      context.cancelGeneration();
+      return { status: 'executed', action: resolved.action, effect: 'cancel-generation', keybinding: descriptor };
+    case 'screen-clear':
+      if (!context.clearScreen) return runUnavailable(resolved.action, route, 'commandContext.clearScreen is unavailable.');
+      context.clearScreen();
+      return { status: 'executed', action: resolved.action, effect: 'screen-clear', keybinding: descriptor };
+    case 'panel-picker':
+      if (context.openPanelPicker) {
+        context.openPanelPicker();
+        return { status: 'executed', action: resolved.action, effect: 'agent-workspace-opened', route: 'openPanelPicker', keybinding: descriptor };
+      }
+      if (context.openAgentWorkspace) {
+        context.openAgentWorkspace('home');
+        return { status: 'executed', action: resolved.action, effect: 'agent-workspace-opened', route: 'openAgentWorkspace', categoryId: 'home', keybinding: descriptor };
+      }
+      return runUnavailable(resolved.action, route, 'No panel picker or Agent workspace opener is available.');
+    case 'panel-close': {
+      const active = context.workspace.panelManager?.getActivePanel() ?? null;
+      if (active) context.workspace.panelManager?.close(active.id);
+      if (!active && !context.focusPrompt) return runUnavailable(resolved.action, route, 'No active legacy panel or prompt focus route is available.');
+      if (context.focusPrompt) context.focusPrompt();
+      context.renderRequest();
+      return {
+        status: 'executed',
+        action: resolved.action,
+        effect: active ? 'active-panel-closed' : 'prompt-focused',
+        ...(active ? { panelId: active.id } : {}),
+        keybinding: descriptor,
+      };
+    }
+    case 'panel-close-all': {
+      const managerPanel = context.workspace.panelManager;
+      const openPanels = managerPanel?.getAllOpen() ?? [];
+      for (const panel of openPanels) managerPanel?.close(panel.id);
+      managerPanel?.hide();
+      if (openPanels.length === 0 && !context.focusPrompt) return runUnavailable(resolved.action, route, 'No open legacy panels or prompt focus route is available.');
+      if (context.focusPrompt) context.focusPrompt();
+      context.renderRequest();
+      return {
+        status: 'executed',
+        action: resolved.action,
+        effect: 'all-panels-closed',
+        closedPanels: openPanels.map((panel) => panel.id),
+        keybinding: descriptor,
+      };
+    }
+    case 'history-search': {
+      if (!context.openPromptHistorySearch) return runUnavailable(resolved.action, route, 'commandContext.openPromptHistorySearch is unavailable.');
+      const query = readString(args.value);
+      context.openPromptHistorySearch(query || undefined);
+      return { status: 'executed', action: resolved.action, effect: 'prompt-history-search-opened', query, keybinding: descriptor };
+    }
+    case 'search': {
+      if (!context.openConversationSearch) return runUnavailable(resolved.action, route, 'commandContext.openConversationSearch is unavailable.');
+      const query = readString(args.value);
+      context.openConversationSearch(query || undefined);
+      return { status: 'executed', action: resolved.action, effect: 'conversation-search-opened', query, keybinding: descriptor };
+    }
+    case 'paste': {
+      if (!context.pasteFromClipboard) return runUnavailable(resolved.action, route, 'commandContext.pasteFromClipboard is unavailable.');
+      const pasted = context.pasteFromClipboard();
+      return { status: 'executed', action: resolved.action, effect: 'paste-from-clipboard', pasted, keybinding: descriptor };
+    }
+    case 'block-copy':
+    case 'bookmark':
+    case 'block-save': {
+      if (!context.openBlockActions) return runUnavailable(resolved.action, route, 'commandContext.openBlockActions is unavailable.');
+      const opened = context.openBlockActions();
+      return opened
+        ? {
+          status: 'visible_interaction_required',
+          action: resolved.action,
+          effect: 'block-actions-opened',
+          keybinding: descriptor,
+          note: 'The block action surface is open. Selecting the exact nearest-block action remains visible interactive shell work.',
+        }
+        : {
+          status: 'not_opened',
+          action: resolved.action,
+          effect: 'block-actions',
+          keybinding: descriptor,
+          note: 'The current shell did not have an empty prompt plus nearby block required for block actions.',
+        };
+    }
+    default:
+      return {
+        status: 'unsupported_keybinding_action',
+        action: resolved.action,
+        keybinding: descriptor,
+        modelOperation: route,
+      };
+  }
+}
+
 export function setHarnessKeybinding(context: CommandContext, args: HarnessKeybindingArgs): Record<string, unknown> {
   const manager = requireKeybindingsManager(context);
   const resolved = resolveHarnessKeybinding(context, args);

package/src/tools/agent-harness-tool-schema.ts

@@ -1,7 +1,7 @@
 export const AGENT_HARNESS_MODES = [
   'summary', 'cli_commands', 'cli_command', 'panels', 'panel', 'open_panel',
   'ui_surfaces', 'ui_surface', 'open_ui_surface',
-  'shortcuts', 'keybindings', 'keybinding', 'set_keybinding', 'reset_keybinding',
+  'shortcuts', 'keybindings', 'keybinding', 'run_keybinding', 'set_keybinding', 'reset_keybinding',
   'commands', 'command', 'run_command', 'settings', 'get_setting', 'set_setting',
   'reset_setting', 'workspace', 'workspace_categories', 'workspace_actions',
   'workspace_action', 'run_workspace_action', 'tools', 'tool', 'connected_host', 'connected_host_status',
@@ -27,7 +27,7 @@ export const AGENT_HARNESS_PARAMETER_PROPERTIES = {
   },
   command: {
     type: 'string',
-    description: 'Full slash command string for mode command, workspace_action lookup, or run_command, for example "/settings get provider.model". In cli_command mode this may also hold a top-level CLI string such as "goodvibes-agent status --json".',
+    description: 'Full slash command string for mode command, workspace_action/run_workspace_action lookup, or run_command, for example "/settings get provider.model". In cli_command mode this may also hold a top-level CLI string such as "goodvibes-agent status --json".',
   },
   cliCommand: {
     type: 'string',
@@ -56,7 +56,7 @@ export const AGENT_HARNESS_PARAMETER_PROPERTIES = {
   },
   actionId: {
     type: 'string',
-    description: 'Agent workspace action id for workspace_action or run_workspace_action, or keybinding action id for keybinding/set_keybinding/reset_keybinding. target or query can also look up one keybinding action.',
+    description: 'Agent workspace action id for workspace_action or run_workspace_action, or keybinding action id for keybinding/run_keybinding/set_keybinding/reset_keybinding. command, target, or query can also look up one workspace action; target or query can also look up one keybinding action.',
   },
   fields: {
     type: 'object',
@@ -86,7 +86,7 @@ export const AGENT_HARNESS_PARAMETER_PROPERTIES = {
       { type: 'number' },
       { type: 'boolean' },
     ],
-    description: 'Setting value for set_setting. Strings, booleans, numbers, and enum strings are accepted.',
+    description: 'Setting value for set_setting. Strings, booleans, numbers, and enum strings are accepted. In run_keybinding, value can provide visible search text for search/history-search shortcut routes.',
   },
   target: {
     type: 'string',
@@ -127,10 +127,10 @@ export const AGENT_HARNESS_PARAMETER_PROPERTIES = {
   },
   confirm: {
     type: 'boolean',
-    description: 'Required true for set_setting, reset_setting, run_command, open_panel, open_ui_surface, and mutating run_workspace_action calls after an explicit user request.',
+    description: 'Required true for set_setting, reset_setting, run_keybinding, run_command, open_panel, open_ui_surface, and mutating run_workspace_action calls after an explicit user request.',
   },
   explicitUserRequest: {
     type: 'string',
-    description: 'Exact user request or faithful short summary authorizing a setting mutation or harness UI/command invocation.',
+    description: 'Exact user request or faithful short summary authorizing a setting mutation, keybinding action, or harness UI/command invocation.',
   },
 } as const;

package/src/tools/agent-harness-tool.ts

@@ -19,7 +19,7 @@ import type {
 import { parseSlashCommand } from '../input/slash-command-parser.ts';
 import { blockedHarnessCliCommandTokens, describeHarnessCliCommand, listHarnessCliCommands, totalHarnessCliCommands } from './agent-harness-cli-metadata.ts';
 import { describeHarnessCommand, listHarnessCommands } from './agent-harness-command-catalog.ts';
-import { describeHarnessKeybinding, listHarnessKeybindings, listHarnessShortcuts, resetHarnessKeybinding, setHarnessKeybinding, totalHarnessKeybindings, totalHarnessShortcuts } from './agent-harness-keybinding-metadata.ts';
+import { describeHarnessKeybinding, listHarnessKeybindings, listHarnessShortcuts, resetHarnessKeybinding, runHarnessKeybinding, setHarnessKeybinding, totalHarnessKeybindings, totalHarnessShortcuts } from './agent-harness-keybinding-metadata.ts';
 import { describeHarnessPanel, listHarnessPanels, openHarnessPanel, totalHarnessPanels } from './agent-harness-panel-metadata.ts';
 import { connectedHostStatusSummary } from './agent-harness-connected-host-status.ts';
 import { describeLocalWorkspaceModelExecution, runLocalWorkspaceAction, runLocalWorkspaceEditorAction } from './agent-harness-local-operations.ts';
@@ -273,16 +273,6 @@ function listWorkspaceActions(deps: AgentHarnessToolDeps, args: AgentHarnessTool
     .map((entry) => describeWorkspaceAction(entry.category, entry.action, { includeEditor, editorContext }));
 }
 
-function findWorkspaceAction(args: AgentHarnessToolArgs): { readonly category: AgentWorkspaceCategory; readonly action: AgentWorkspaceAction } | null {
-  const actionId = readString(args.actionId || args.query);
-  const categoryId = readString(args.categoryId || args.category);
-  if (!actionId) return null;
-  return allWorkspaceActions().find((entry) => {
-    if (categoryId && entry.category.id !== categoryId) return false;
-    return entry.action.id === actionId || entry.action.label.toLowerCase() === actionId.toLowerCase();
-  }) ?? null;
-}
-
 function workspaceActionLookupFromArgs(args: AgentHarnessToolArgs): { readonly source: WorkspaceActionLookup['source']; readonly input: string } | null {
   const actionId = readString(args.actionId);
   if (actionId) return { source: 'actionId', input: actionId };
@@ -509,9 +499,10 @@ async function runWorkspaceAction(
   deps: AgentHarnessToolDeps,
   args: AgentHarnessToolArgs,
 ): Promise<{ readonly success: boolean; readonly output?: string; readonly error?: string }> {
-  const found = findWorkspaceAction(args);
-  if (!found) return error('run_workspace_action requires a valid actionId. Use mode:"workspace_actions" to inspect available actions.');
-  const { category, action } = found;
+  const resolved = resolveWorkspaceActionDetail(args);
+  if (resolved?.status === 'ambiguous') return error(`Ambiguous Agent workspace action ${resolved.input}. Candidates: ${JSON.stringify(resolved.candidates)}`);
+  if (!resolved) return error('run_workspace_action requires a valid actionId, command, target, or query. Use mode:"workspace_actions" to inspect available actions.');
+  const { category, action, lookup } = resolved;
 
   if (action.safety === 'blocked') {
     return error(`Workspace action ${action.id} is blocked in Agent: ${action.detail}`);
@@ -520,14 +511,14 @@ async function runWorkspaceAction(
     const editorContext = buildWorkspaceEditorContext(deps.commandContext, args);
     return output({
       status: 'guidance',
-      action: describeWorkspaceAction(category, action, { includeEditor: true, editorContext }),
+      action: describeWorkspaceAction(category, action, { includeEditor: true, editorContext, lookup }),
     });
   }
   if (action.kind === 'workspace' && action.targetCategoryId) {
     const target = AGENT_WORKSPACE_CATEGORIES.find((entry) => entry.id === action.targetCategoryId);
     return output({
       status: 'workspace_target',
-      action: describeWorkspaceAction(category, action),
+      action: describeWorkspaceAction(category, action, { lookup }),
       targetCategory: target ? describeWorkspaceCategory(target) : action.targetCategoryId,
       targetActions: target ? target.actions.map((entry) => describeWorkspaceAction(target, entry)).slice(0, 40) : [],
     });
@@ -536,7 +527,7 @@ async function runWorkspaceAction(
     if (/<[^>\s]+(?:\s+[^>]*)?>/.test(action.command)) {
       return output({
         status: 'needs_concrete_command',
-        action: describeWorkspaceAction(category, action),
+        action: describeWorkspaceAction(category, action, { lookup }),
         note: 'This workspace action is a command template. Provide concrete values with mode:"run_command" once the exact command is known.',
       });
     }
@@ -553,7 +544,7 @@ async function runWorkspaceAction(
   const editorContext = buildWorkspaceEditorContext(deps.commandContext, args);
   return output({
     status: 'no_direct_effect',
-    action: describeWorkspaceAction(category, action, { includeEditor: true, editorContext }),
+    action: describeWorkspaceAction(category, action, { includeEditor: true, editorContext, lookup }),
   });
 }
 
@@ -564,7 +555,7 @@ export function createAgentHarnessTool(deps: AgentHarnessToolDeps): Tool {
       description: [
         'Discover and operate the GoodVibes Agent harness from the main conversation.',
         'Use this tool to inspect Agent workspace actions, built-in panels, top-level CLI mirrors, UI surfaces, keybindings, slash commands with policy metadata, model tools or one model tool schema, connected-host capabilities or one connected-host capability detail, and Agent settings, or to invoke a workspace action/command through the same in-process command registry the user uses in the TUI.',
-        'Discovery modes are read-only. Setting/keybinding writes, resets, UI routing, slash command invocation, and workspace action invocation require confirm:true plus explicitUserRequest.',
+        'Discovery modes are read-only. Setting/keybinding writes, resets, keybinding actions, UI routing, slash command invocation, and workspace action invocation require confirm:true plus explicitUserRequest.',
         'This tool preserves Agent product boundaries: connected-host lifecycle and listener posture stay externally owned, connected-host mode reports allowed and blocked route families, and secret-backed settings store raw values through the secret manager while config receives only a secret reference.',
       ].join(' '),
       parameters: {
@@ -597,9 +588,9 @@ export function createAgentHarnessTool(deps: AgentHarnessToolDeps): Tool {
               cliCommands: 'Use mode:"cli_commands" to list and mode:"cli_command" with cliCommand, command, commandName, target, or query to inspect package CLI mirrors and their preferred in-process model routes. CLI modes are discovery-only.',
               panels: 'Use mode:"panels" to list and mode:"panel" with panelId, target, or query to inspect built-in panel catalog/open state; use mode:"open_panel" with confirm:true plus explicitUserRequest to route a visible panel/workspace change.',
               uiSurfaces: 'Use mode:"ui_surfaces" to list and mode:"ui_surface" with surfaceId, target, or query to inspect modal/overlay/picker/workspace surfaces; use mode:"open_ui_surface" with confirm:true plus explicitUserRequest to route visible UI navigation.',
-              shortcuts: 'Use mode:"shortcuts" to inspect fixed shortcuts plus configurable keybindings. Use mode:"keybinding" with actionId, target, key, or query; use mode:"set_keybinding" and mode:"reset_keybinding" with confirm:true plus explicitUserRequest to edit the same config file the user edits.',
+              shortcuts: 'Use mode:"shortcuts" to inspect fixed shortcuts plus configurable keybindings. Use mode:"keybinding" with actionId, target, key, or query; use mode:"run_keybinding" for confirmation-gated shell-safe shortcut equivalents; use mode:"set_keybinding" and mode:"reset_keybinding" with confirm:true plus explicitUserRequest to edit the same config file the user edits.',
               slashCommands: 'Use mode:"commands" to list slash commands and mode:"command" with command, commandName, target, or query to inspect one command; use mode:"run_command" with confirm:true plus explicitUserRequest to execute.',
-              workspace: 'Use mode:"workspace_actions" to list and mode:"workspace_action" with actionId, command, target, or query for one action and editor schema; set includeParameters:true on workspace_actions to inline editor schemas.',
+              workspace: 'Use mode:"workspace_actions" to list, mode:"workspace_action" with actionId, command, target, or query for one action and editor schema, and mode:"run_workspace_action" with the same lookup fields plus confirmation for executable actions; set includeParameters:true on workspace_actions to inline editor schemas.',
               settings: 'Use mode:"settings" to list and mode:"get_setting" with key, target, or query for one setting. Use mode:"set_setting" or mode:"reset_setting" with key, target, or query plus confirm:true and explicitUserRequest.',
               tools: 'Use mode:"tools" to list first-class model tools, or mode:"tool" with toolName, target, or query to inspect one schema.',
               connectedHost: 'Use mode:"connected_host" for the connected-host capability map and blocked boundaries. Use mode:"connected_host_capability" with capabilityId, target, or query for one allowed or blocked capability.',
@@ -661,13 +652,15 @@ export function createAgentHarnessTool(deps: AgentHarnessToolDeps): Tool {
         }
         if (args.mode === 'set_keybinding') {
           const confirmationError = requireConfirmedAction(args, 'Keybinding mutation');
-          if (confirmationError) return error(confirmationError);
-          return output(setHarnessKeybinding(deps.commandContext, args));
+          return confirmationError ? error(confirmationError) : output(setHarnessKeybinding(deps.commandContext, args));
         }
         if (args.mode === 'reset_keybinding') {
           const confirmationError = requireConfirmedAction(args, 'Keybinding reset');
-          if (confirmationError) return error(confirmationError);
-          return output(resetHarnessKeybinding(deps.commandContext, args));
+          return confirmationError ? error(confirmationError) : output(resetHarnessKeybinding(deps.commandContext, args));
+        }
+        if (args.mode === 'run_keybinding') {
+          const confirmationError = requireConfirmedAction(args, 'Keybinding action');
+          return confirmationError ? error(confirmationError) : output(runHarnessKeybinding(deps.commandContext, args));
         }
         if (args.mode === 'commands') {
           const commands = listHarnessCommands(deps.commandRegistry, args);

package/src/tools/agent-harness-ui-surface-metadata.ts

@@ -642,7 +642,7 @@ const UI_SURFACES: readonly UiSurfaceDefinition[] = [
     kind: 'overlay',
     summary: 'Keyboard shortcut reference overlay.',
     command: '/shortcuts',
-    preferredModelRoute: 'Use shortcuts/keybindings modes for model-readable discovery and confirmed keybinding edits.',
+    preferredModelRoute: 'Use shortcuts/keybindings modes for model-readable discovery, run_keybinding for supported shell-safe actions, and confirmed keybinding edits for binding changes.',
     available: (context) => typeof context.openShortcutsOverlay === 'function',
     open: (context) => {
       const surface = findSurfaceById('shortcuts-overlay')!;

package/src/version.ts

@@ -6,7 +6,7 @@ import { join } from 'node:path';
 // The prebuild script updates the fallback value before compilation.
 // Uses import.meta.dir (Bun) to locate package.json relative to this file,
 // which is correct regardless of the process working directory.
-let _version = '1.0.19';
+let _version = '1.0.21';
 let _sdkVersion = '0.33.35';
 try {
   const pkg = JSON.parse(readFileSync(join(import.meta.dir, '..', 'package.json'), 'utf-8')) as {