mixdog 1.4.4 → 1.4.5
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/package.json +1 -2
- package/src/cli.mjs +5 -1
- package/src/defaults/shipped-history.json +26 -2
- package/src/defaults/skills/browser-use/SKILL.md +148 -273
- package/src/defaults/skills/browser-use/references/devtools.md +14 -0
- package/src/defaults/skills/browser-use/references/page-flows.md +66 -0
- package/src/defaults/skills/browser-use/references/sign-in.md +70 -0
- package/src/defaults/skills/code-tidy/SKILL.md +60 -84
- package/src/defaults/skills/code-tidy/references/agent-cleanup.md +27 -2
- package/src/defaults/skills/computer-use/SKILL.md +145 -240
- package/src/defaults/skills/computer-use/references/flows.md +35 -0
- package/src/defaults/skills/computer-use/references/operations.md +22 -0
- package/src/defaults/skills/computer-use/references/platforms.md +12 -0
- package/src/defaults/skills/docx/SKILL.md +23 -31
- package/src/defaults/skills/docx/references/native-authoring.md +26 -3
- package/src/defaults/skills/docx/references/redlining.md +15 -0
- package/src/defaults/skills/image/SKILL.md +37 -77
- package/src/defaults/skills/image/references/prompt-forms.md +51 -0
- package/src/defaults/skills/image/references/text.md +6 -0
- package/src/defaults/skills/local-provider/SKILL.md +75 -105
- package/src/defaults/skills/local-provider/references/lifecycle.md +14 -0
- package/src/defaults/skills/local-provider/references/maintenance.md +19 -0
- package/src/defaults/skills/local-provider/references/recovery.md +30 -0
- package/src/defaults/skills/memory-management/SKILL.md +16 -20
- package/src/defaults/skills/pdf/SKILL.md +21 -23
- package/src/defaults/skills/pdf/references/blocks.md +20 -0
- package/src/defaults/skills/pdf/references/forms.md +12 -0
- package/src/defaults/skills/pdf/references/html.md +2 -2
- package/src/defaults/skills/pdf/references/ocr.md +7 -0
- package/src/defaults/skills/pdf/references/security.md +8 -0
- package/src/defaults/skills/pptx/SKILL.md +30 -109
- package/src/defaults/skills/pptx/references/audit.md +45 -0
- package/src/defaults/skills/pptx/references/brief-critique.md +47 -0
- package/src/defaults/skills/pptx/references/editing.md +24 -0
- package/src/defaults/skills/pptx/references/html.md +1 -1
- package/src/defaults/skills/pptx/references/script.md +33 -0
- package/src/defaults/skills/setup/SKILL.md +34 -102
- package/src/defaults/skills/setup/references/actions.md +72 -13
- package/src/defaults/skills/setup/references/surfaces.md +23 -8
- package/src/defaults/skills/setup-guide/SKILL.md +28 -115
- package/src/defaults/skills/setup-guide/references/features.md +36 -0
- package/src/defaults/skills/setup-guide/references/glossary.md +38 -0
- package/src/defaults/skills/setup-guide/references/workflow-models.md +62 -0
- package/src/defaults/skills/test-suite/SKILL.md +15 -127
- package/src/defaults/skills/test-suite/references/case-management.md +17 -9
- package/src/defaults/skills/test-suite/references/diagnose.md +23 -0
- package/src/defaults/skills/test-suite/references/prune.md +27 -0
- package/src/defaults/skills/test-suite/references/write.md +44 -0
- package/src/defaults/skills/video/SKILL.md +28 -42
- package/src/defaults/skills/video/references/prompting.md +31 -0
- package/src/defaults/skills/visualize/SKILL.md +73 -107
- package/src/defaults/skills/visualize/references/craft.md +1 -2
- package/src/defaults/skills/visualize/references/feedback.md +36 -0
- package/src/defaults/skills/visualize/references/showing.md +13 -0
- package/src/defaults/skills/xlsx/SKILL.md +23 -50
- package/src/defaults/skills/xlsx/references/editing.md +15 -0
- package/src/defaults/skills/xlsx/references/model-conventions.md +1 -1
- package/src/defaults/skills/xlsx/references/recalculation.md +37 -0
- package/src/defaults/skills/xlsx/references/report-design.md +18 -2
- package/src/lib/keychain-cjs.cjs +22 -1
- package/src/lib/rules-builder.cjs +1 -1
- package/src/rules/agent/AGENT.md +3 -0
- package/src/rules/lead/LEAD.md +3 -1
- package/src/runtime/agent/orchestrator/agent-runtime/cache-strategy.mjs +8 -30
- package/src/runtime/agent/orchestrator/agent-runtime/title-completion.mjs +5 -0
- package/src/runtime/agent/orchestrator/mcp/oauth-request.mjs +21 -2
- package/src/runtime/agent/orchestrator/providers/account-pool.mjs +3 -0
- package/src/runtime/agent/orchestrator/providers/admission-scheduler.mjs +2 -2
- package/src/runtime/agent/orchestrator/providers/anthropic-api-transport.mjs +2 -0
- package/src/runtime/agent/orchestrator/providers/anthropic-message-threads.mjs +583 -0
- package/src/runtime/agent/orchestrator/providers/anthropic-oauth-client-version.mjs +1 -1
- package/src/runtime/agent/orchestrator/providers/anthropic-oauth.mjs +81 -19
- package/src/runtime/agent/orchestrator/providers/anthropic-sse.mjs +180 -13
- package/src/runtime/agent/orchestrator/providers/anthropic.mjs +56 -6
- package/src/runtime/agent/orchestrator/providers/cursor-wire.mjs +4 -2
- package/src/runtime/agent/orchestrator/providers/gemini-error-contract.mjs +6 -6
- package/src/runtime/agent/orchestrator/providers/gemini-model-redirect.mjs +44 -0
- package/src/runtime/agent/orchestrator/providers/gemini-reasoning-relay.mjs +5 -0
- package/src/runtime/agent/orchestrator/providers/gemini-request-body.mjs +34 -5
- package/src/runtime/agent/orchestrator/providers/gemini-response.mjs +10 -0
- package/src/runtime/agent/orchestrator/providers/gemini-schema.mjs +11 -6
- package/src/runtime/agent/orchestrator/providers/gemini-sdk-request.mjs +4 -100
- package/src/runtime/agent/orchestrator/providers/gemini-stream.mjs +3 -227
- package/src/runtime/agent/orchestrator/providers/gemini.mjs +226 -118
- package/src/runtime/agent/orchestrator/providers/lib/anthropic-request-utils.mjs +11 -1
- package/src/runtime/agent/orchestrator/providers/lib/claude-bridge-schema.mjs +15 -3
- package/src/runtime/agent/orchestrator/providers/lib/gemini-model-catalog.mjs +2 -1
- package/src/runtime/agent/orchestrator/providers/openai-direct-request.mjs +5 -28
- package/src/runtime/agent/orchestrator/providers/openai-ws-pool.mjs +50 -28
- package/src/runtime/agent/orchestrator/providers/openai-ws-stream/socket-handlers.mjs +14 -10
- package/src/runtime/agent/orchestrator/providers/openai-ws.mjs +1 -3
- package/src/runtime/agent/orchestrator/providers/stream-json-pool/worker-slots/slot-spawn.mjs +8 -3
- package/src/runtime/agent/orchestrator/providers/vertex-ai-maas.mjs +2 -2
- package/src/runtime/agent/orchestrator/runtime-core/config-helpers.mjs +4 -5
- package/src/runtime/agent/orchestrator/runtime-core/goal-tool-defs.mjs +2 -2
- package/src/runtime/agent/orchestrator/runtime-core/provider-request-tools.mjs +15 -8
- package/src/runtime/agent/orchestrator/runtime-core/session-title.mjs +1 -0
- package/src/runtime/agent/orchestrator/runtime-core/todo-list.mjs +9 -32
- package/src/runtime/agent/orchestrator/runtime-core/todo-tool-defs.mjs +4 -5
- package/src/runtime/agent/orchestrator/session/compact/summary-route.mjs +1 -3
- package/src/runtime/agent/orchestrator/session/loop/compact-policy.mjs +5 -2
- package/src/runtime/agent/orchestrator/session/loop/no-tool-turn/todo-check.mjs +2 -17
- package/src/runtime/agent/orchestrator/session/loop/usage.mjs +7 -3
- package/src/runtime/agent/orchestrator/session/manager/ask-turn-settle.mjs +9 -2
- package/src/runtime/agent/orchestrator/session/manager/session-close.mjs +26 -13
- package/src/runtime/agent/orchestrator/session/manager/session-lifecycle.mjs +4 -5
- package/src/runtime/agent/orchestrator/session/manager/usage-metrics.mjs +16 -5
- package/src/runtime/agent/orchestrator/session/store/lifecycle-barriers.mjs +22 -1
- package/src/runtime/agent/orchestrator/session/store/live-state.mjs +4 -0
- package/src/runtime/agent/orchestrator/tools/builtin/cache-layers.mjs +11 -9
- package/src/runtime/agent/orchestrator/tools/builtin/grep-formatting.mjs +2 -8
- package/src/runtime/agent/orchestrator/tools/builtin/lib/search-input-helpers.mjs +1 -9
- package/src/runtime/agent/orchestrator/tools/builtin/local-search-telemetry.mjs +2 -2
- package/src/runtime/agent/orchestrator/tools/builtin/native-search-client.mjs +1 -1
- package/src/runtime/agent/orchestrator/tools/builtin/runtime-capabilities.mjs +12 -3
- package/src/runtime/agent/orchestrator/tools/code-graph/ast-calls.mjs +9 -2
- package/src/runtime/agent/orchestrator/tools/code-graph/build.mjs +137 -7
- package/src/runtime/agent/orchestrator/tools/code-graph/calls-cache.mjs +43 -14
- package/src/runtime/agent/orchestrator/tools/code-graph/calls-sidecar-worker.mjs +33 -0
- package/src/runtime/agent/orchestrator/tools/code-graph/disk-cache.mjs +146 -11
- package/src/runtime/agent/orchestrator/tools/code-graph/dispatch.mjs +31 -24
- package/src/runtime/agent/orchestrator/tools/code-graph/graph-binary/spawn-jsonl.mjs +172 -25
- package/src/runtime/agent/orchestrator/tools/code-graph/graph-model.mjs +100 -2
- package/src/runtime/agent/orchestrator/tools/code-graph/main-thread-slice.mjs +18 -0
- package/src/runtime/agent/orchestrator/tools/code-graph/mask-pool.mjs +219 -0
- package/src/runtime/agent/orchestrator/tools/code-graph/mask-worker.mjs +12 -0
- package/src/runtime/agent/orchestrator/tools/code-graph/memory-cache.mjs +6 -1
- package/src/runtime/agent/orchestrator/tools/code-graph/modes/calls.mjs +9 -2
- package/src/runtime/agent/orchestrator/tools/code-graph/modes/dependents.mjs +5 -1
- package/src/runtime/agent/orchestrator/tools/code-graph/modes/structure.mjs +8 -2
- package/src/runtime/agent/orchestrator/tools/code-graph/modes/symbols.mjs +14 -11
- package/src/runtime/agent/orchestrator/tools/code-graph/search-references.mjs +41 -18
- package/src/runtime/agent/orchestrator/tools/code-graph/search.mjs +391 -83
- package/src/runtime/agent/orchestrator/tools/code-graph/symbol-index.mjs +109 -9
- package/src/runtime/agent/orchestrator/tools/code-graph/text-mask.mjs +73 -29
- package/src/runtime/agent/orchestrator/tools/code-graph-prewarm-worker.mjs +6 -5
- package/src/runtime/agent/orchestrator/tools/lib/native-spawn-client.mjs +7 -3
- package/src/runtime/channels/lib/crash-log.mjs +21 -1
- package/src/runtime/channels/lib/worker-main.mjs +8 -25
- package/src/runtime/media/jobs.mjs +5 -0
- package/src/runtime/media/store.mjs +146 -3
- package/src/runtime/office/core/office-write-ledger.mjs +2 -2
- package/src/runtime/shared/agent-reap-state.mjs +19 -0
- package/src/runtime/shared/desktop-notice-host.d.mts +13 -0
- package/src/runtime/shared/desktop-notice-host.mjs +87 -0
- package/src/runtime/shared/file-lock.mjs +2 -2
- package/src/runtime/shared/heap-cap.mjs +9 -11
- package/src/runtime/shared/llm/http-agent.mjs +11 -1
- package/src/runtime/shared/llm/system-proxy.mjs +100 -25
- package/src/runtime/shared/llm/usage-accounting.mjs +18 -7
- package/src/runtime/shared/llm/usage-context.mjs +2 -2
- package/src/runtime/shared/llm/usage-ledger-requests.mjs +10 -3
- package/src/runtime/shared/llm/usage-ledger.mjs +13 -1
- package/src/runtime/shared/provider-accounts.mjs +22 -0
- package/src/runtime/shared/provider-auth-binding.mjs +2 -2
- package/src/runtime/shared/resource-admission.mjs +2 -2
- package/src/runtime/shared/root-async-context.mjs +51 -0
- package/src/runtime/shared/run-child.mjs +18 -1
- package/src/runtime/shared/slow-span.mjs +2 -2
- package/src/runtime/shared/tool-execution-owner.mjs +2 -2
- package/src/runtime/shared/tool-work-units.mjs +2 -0
- package/src/runtime/shared/turn-snapshot-store.mjs +75 -18
- package/src/runtime/shared/turn-worktree-snapshot.mjs +38 -8
- package/src/runtime/shared/user-cwd.mjs +2 -2
- package/src/runtime/shared/worker-requests.mjs +3 -1
- package/src/runtime/shared/worktrees/lock.mjs +2 -2
- package/src/runtime/visualize-bridge/client.mjs +26 -4
- package/src/runtime/visualize-bridge/tool-defs.mjs +46 -5
- package/src/session-runtime/boot/apis.mjs +1 -0
- package/src/session-runtime/boot/begin.mjs +2 -0
- package/src/session-runtime/media-api.mjs +15 -0
- package/src/session-runtime/provider-readiness.mjs +10 -2
- package/src/session-runtime/runtime-core.mjs +35 -8
- package/src/session-runtime/runtime-review-api.mjs +3 -1
- package/src/session-runtime/services/agent-tool/close-flow.mjs +6 -1
- package/src/session-runtime/services/agent-tool/execute-resume.mjs +94 -0
- package/src/session-runtime/services/agent-tool/execute-send.mjs +77 -27
- package/src/session-runtime/services/agent-tool/execute-spawn.mjs +28 -7
- package/src/session-runtime/services/agent-tool/execute.mjs +6 -1
- package/src/session-runtime/services/agent-tool/helpers.mjs +11 -0
- package/src/session-runtime/services/agent-tool/job-views/job-render.mjs +13 -10
- package/src/session-runtime/services/agent-tool/render.mjs +2 -0
- package/src/session-runtime/services/agent-tool/send-flow.mjs +1 -0
- package/src/session-runtime/services/agent-tool/spawn-flow/spawn-prep.mjs +6 -0
- package/src/session-runtime/services/agent-tool/spawn-flow/spawn-run.mjs +2 -1
- package/src/session-runtime/services/agent-tool/tag-registry/reaped-sessions.mjs +82 -0
- package/src/session-runtime/services/agent-tool/tag-registry/record-facts-worker.mjs +36 -0
- package/src/session-runtime/services/agent-tool/tag-registry.mjs +4 -0
- package/src/session-runtime/services/agent-tool/terminal-reaper/row-helpers.mjs +29 -0
- package/src/session-runtime/services/agent-tool/terminal-reaper/scan-admission.mjs +11 -3
- package/src/session-runtime/services/agent-tool/terminal-reaper/stale-rows.mjs +19 -8
- package/src/session-runtime/services/agent-tool/terminal-reaper.mjs +5 -2
- package/src/session-runtime/services/agent-tool/worker-index.mjs +1 -0
- package/src/session-runtime/services/daemon-crash-capture/daemon-exit-record.mjs +2 -2
- package/src/session-runtime/session-create.mjs +5 -0
- package/src/session-runtime/setup-tool/settings-contract.mjs +2 -0
- package/src/session-runtime/usage-stats-api.mjs +3 -0
- package/src/session-runtime/workspace/admission-gate.mjs +25 -7
- package/src/session-runtime/workspace/transitions.mjs +4 -0
- package/src/standalone/daemon-log.mjs +28 -4
- package/src/standalone/daemon-owner.mjs +41 -7
- package/src/standalone/daemon-work-gate.mjs +139 -0
- package/src/standalone/daemon.mjs +80 -11
- package/src/standalone/desktop-service-registry.mjs +11 -4
- package/src/standalone/session-client.mjs +57 -5
- package/src/standalone/session-protocol.mjs +5 -0
- package/src/standalone/session-service/agent-tree/agent-cancel.mjs +17 -1
- package/src/standalone/session-service/agent-tree.mjs +7 -0
- package/src/standalone/session-service/session-calls/session-turns.mjs +12 -5
- package/src/standalone/session-transport/call-route.mjs +10 -4
- package/src/standalone/session-transport/routes.mjs +21 -0
- package/src/standalone/session-transport/stack.mjs +3 -1
- package/src/standalone/session-transport.mjs +5 -0
- package/src/tui/dist/index.mjs +495 -477
- package/src/tui/session/session-api/media.mjs +5 -0
- package/src/tui/session/turn.mjs +13 -0
- package/src/runtime/agent/orchestrator/providers/gemini-sdk-stream/cancellation.mjs +0 -71
- package/src/runtime/agent/orchestrator/providers/gemini-sdk-stream/reader.mjs +0 -108
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
name: setup
|
|
3
3
|
description: Inspect or change persisted Mixdog settings through the built-in setup tool.
|
|
4
|
-
when_to_use: 'Settings,
|
|
4
|
+
when_to_use: 'Settings, not code/local installs: routes, workflows, MCP, plugins, skill toggles, style, profile.'
|
|
5
5
|
dependencies:
|
|
6
6
|
tools:
|
|
7
7
|
- type: tool
|
|
@@ -30,13 +30,25 @@ supported `setup` action, and read its result as the change receipt.
|
|
|
30
30
|
`riskAccepted:true`, and remote enablement is refused too. Open `developer`
|
|
31
31
|
instead.
|
|
32
32
|
- A mutation follows the active workflow's approval rules. Status, catalog,
|
|
33
|
-
definition and MCP-config reads do not change persisted settings.
|
|
33
|
+
definition and MCP-config reads do not change persisted settings. Deleting a
|
|
34
|
+
hook rule and forgetting provider authentication are destructive and need
|
|
35
|
+
explicit approval; disabling preserves.
|
|
36
|
+
- Notification permission/subscription, pairing credentials, and authentication
|
|
37
|
+
remain user-operated UI handoffs. Do not imitate them by editing storage.
|
|
34
38
|
- Most changes affect new sessions. Report `appliedToCurrentSession` and
|
|
35
39
|
`appliesTo` exactly when the result supplies them.
|
|
36
|
-
-
|
|
37
|
-
|
|
38
|
-
|
|
39
|
-
|
|
40
|
+
- Ownership: source edits, builds, installation deployment, releases and app
|
|
41
|
+
restarts are not setup's. Managed local-model installation and recovery
|
|
42
|
+
belong to `local-provider` (same setup tool; ordinary route settings stay
|
|
43
|
+
here), skill authoring to `skill-creator`, Core Memory entries to the
|
|
44
|
+
`memory` tool. Read a definition before editing it.
|
|
45
|
+
- Setup changes settings; it does not stop work. Use the owner: `session`
|
|
46
|
+
`cancel` (another session's turn, a routine run, a group room's exchange; a
|
|
47
|
+
Browser/Computer takeover ends with its turn), `agent` `cancel`/`close`,
|
|
48
|
+
`task` (background shell jobs); the user pauses or stops an active Goal from
|
|
49
|
+
its controls. Setup only prevents more: `set_automation_enabled false`,
|
|
50
|
+
`fallbacks: []`, `cancel_quota_resume`, `set_builtin_enabled` browser/computer
|
|
51
|
+
off.
|
|
40
52
|
|
|
41
53
|
## Core loop
|
|
42
54
|
|
|
@@ -56,110 +68,30 @@ supported `setup` action, and read its result as the change receipt.
|
|
|
56
68
|
Completion means the persisted state or UI handoff is evidenced, not merely
|
|
57
69
|
that a call returned without transport error.
|
|
58
70
|
|
|
59
|
-
##
|
|
60
|
-
|
|
61
|
-
- Read `references/actions.md` when selecting a mutation or handling a
|
|
62
|
-
domain-specific edge case.
|
|
63
|
-
- Read `references/surfaces.md` only when the user asks where a setting lives,
|
|
64
|
-
a secret must be entered, or no setup mutation exists.
|
|
65
|
-
|
|
66
|
-
## What applies now vs. later
|
|
67
|
-
|
|
68
|
-
- Now: built-in, connector, permission, hook-rule, cache-warming and
|
|
69
|
-
automation changes (each at its next use), and a fallback chain: running
|
|
70
|
-
sessions on that route read it at their next hop.
|
|
71
|
-
- Next turn: an agent/bot route for its open chats, `reset_route_fallback`, a
|
|
72
|
-
workflow switch. A running turn finishes on its current model.
|
|
73
|
-
- New sessions: the Main model (`set_route` model), instruction files, most
|
|
74
|
-
other settings. An established conversation keeps its model.
|
|
75
|
-
- Session actions (`reset_route_fallback`, quota resume, `continue_in_model`)
|
|
76
|
-
answer `scope: "session"`, not a saved setting.
|
|
77
|
-
|
|
78
|
-
## Stopping something already running
|
|
79
|
-
|
|
80
|
-
Setup changes settings; it does not stop work. Use the tool that owns it:
|
|
81
|
-
`session` `cancel` (another session's turn, a routine run in a bot's chat, a
|
|
82
|
-
group room's exchange; a Browser/Computer takeover ends with its turn),
|
|
83
|
-
`agent` `cancel`/`close` (sub-agents), `task` (background shell jobs); an
|
|
84
|
-
active Goal is paused or stopped by the user from its controls. Setup only prevents more: `set_automation_enabled false`
|
|
85
|
-
(future routine runs), `fallbacks: []` (further hops), `cancel_quota_resume`,
|
|
86
|
-
`set_builtin_enabled` browser/computer off.
|
|
87
|
-
|
|
88
|
-
## High-signal distinctions
|
|
71
|
+
## Distinctions to keep in mind
|
|
89
72
|
|
|
90
|
-
- Main model, agent models, and Web Search model are separate routes.
|
|
91
|
-
|
|
92
|
-
|
|
93
|
-
|
|
94
|
-
|
|
95
|
-
-
|
|
96
|
-
|
|
97
|
-
|
|
98
|
-
are read and changed with the `memory` tool, not setup.
|
|
99
|
-
- Built-ins have installation state separate from enabled state; Visualize
|
|
100
|
-
(on by default) and Media generation are toggle-only (no install step). Browser Use,
|
|
101
|
-
Computer Use actions require this conversation open in the local Desktop
|
|
102
|
-
window; voice runs on the runtime and needs no Desktop.
|
|
103
|
-
- Hook rules are before-tool policy, separate from MCP and plugins. Deleting a
|
|
104
|
-
rule is destructive; disabling preserves it. A rule change reaches open
|
|
105
|
-
sessions and their sub-agents on the next tool call.
|
|
106
|
-
- Auto-clear is per provider and can be `unlimited` (`never`). Built-in
|
|
107
|
-
windows: Codex/ChatGPT login unlimited, OpenAI API 30 min, Anthropic 1h.
|
|
108
|
-
Unlimited stops idle clearing only: compaction still runs at the context
|
|
109
|
-
limit, and finished sub-agent sessions are still released, never kept
|
|
110
|
-
forever: after the provider's own window (OpenAI API 30 min, Anthropic 1h,
|
|
111
|
-
DeepSeek 24h), or 1h for Codex/ChatGPT logins; an unlimited override does
|
|
112
|
-
not stop that.
|
|
113
|
-
- Desktop appearance changes affect the Desktop host, not a paired phone.
|
|
114
|
-
Fonts, imported color themes and skin transparency are saved per device.
|
|
115
|
-
Display language is independent of profile response language. A required
|
|
116
|
-
reload is reported, never performed automatically.
|
|
117
|
-
- A routine or webhook run has no user and no approval prompts: it may change
|
|
118
|
-
only its own bot's automations and route, and payment and permanent-delete
|
|
119
|
-
connectors are refused (trash moves are allowed). Bot permission "block"
|
|
120
|
-
still applies. In an attended bot conversation (home or direct chat, group
|
|
121
|
-
room, or a sub-agent/task session working for one) a payment or permanent
|
|
122
|
-
delete connector call always asks first: allow once or deny, never "항상
|
|
123
|
-
허용", whatever was stored before. Coding sessions are not asked.
|
|
124
|
-
- Prompt-affecting changes land without rewriting the cached prompt: a
|
|
125
|
-
workflow switch arrives as a reminder on the next turn; a bot's
|
|
126
|
-
instruction, persona or name applies on 적용 (one cache break); connector
|
|
127
|
-
and built-in feature changes take effect at once (tool availability and
|
|
128
|
-
call-time refusal), and only the prompt text about them waits for the cache
|
|
129
|
-
to expire.
|
|
130
|
-
- Read definitions before editing them; use skill-creator when authoring a
|
|
131
|
-
skill. Setup persists the supported definition fields. Workflow and agent
|
|
132
|
-
ids are stable identities, distinct from their display names.
|
|
133
|
-
- MCP servers and plugins are machine-global integrations whose visibility may
|
|
134
|
-
be scoped to selected Projects.
|
|
135
|
-
- User skills are machine-global. Enabled plugin skills and built-in skills are
|
|
136
|
-
also discoverable; project-local `.mixdog/skills` directories are ignored.
|
|
137
|
-
- `set_disabled_skills` replaces the complete disabled list; inspect the
|
|
138
|
-
current list before changing one entry.
|
|
73
|
+
- Main model, agent models, and Web Search model are separate routes. An
|
|
74
|
+
explicitly empty agent or Web Search provider restores Main inheritance;
|
|
75
|
+
omission never resets it. Disabling an agent preserves its route.
|
|
76
|
+
- Built-ins have installation state separate from enabled state; Visualize and
|
|
77
|
+
Media generation are toggle-only (`set_builtin_enabled`, no install step).
|
|
78
|
+
- Unknown keys and retired configuration are diagnostics, not permission for
|
|
79
|
+
direct cleanup. Do not expose internal-only configuration as a supported
|
|
80
|
+
user option.
|
|
139
81
|
|
|
140
82
|
## Failure handling
|
|
141
83
|
|
|
142
84
|
- A headless `open` returns `opened:false`; relay its guidance instead of
|
|
143
85
|
pretending a window opened.
|
|
144
|
-
- For an MCP failure, inspect the reported transport error before changing
|
|
145
|
-
command, arguments, directory, environment, URL, headers, or enabled state.
|
|
146
|
-
- MCP reads intentionally omit raw connection values that may contain secrets.
|
|
147
|
-
- An HTTP/SSE MCP server that needs OAuth waits for sign-in instead of
|
|
148
|
-
failing; setup rows then show it enabled, not connected, with no `error`.
|
|
149
|
-
Setup cannot sign in, and `reconnect_mcp` does not log in: hand off to
|
|
150
|
-
`open mcp` (Connect/Reconnect) or the TUI `/mcp login`.
|
|
151
|
-
- For an unavailable built-in, distinguish not installed, installed but
|
|
152
|
-
disabled, and bridge inactive.
|
|
153
86
|
- Desktop requests require a single live claimant and a confirmed receipt.
|
|
154
87
|
A timeout or cancelled request is not success. If execution already started,
|
|
155
88
|
inspect state before retrying; do not resubmit a possibly completed mutation.
|
|
156
|
-
- Notification permission/subscription, pairing credentials, and authentication
|
|
157
|
-
remain user-operated UI handoffs. Do not imitate them by editing storage.
|
|
158
|
-
- Unknown keys and retired configuration are diagnostics, not permission for
|
|
159
|
-
direct cleanup.
|
|
160
|
-
|
|
161
|
-
## Finish
|
|
162
|
-
|
|
163
|
-
- Do not expose internal-only configuration as a supported user option.
|
|
164
89
|
- Do not commit, deploy, restart, or update the app unless separately requested
|
|
165
90
|
and approved.
|
|
91
|
+
|
|
92
|
+
## Resources
|
|
93
|
+
|
|
94
|
+
- `references/actions.md` — read when selecting a mutation, explaining domain
|
|
95
|
+
behavior or change timing, or diagnosing MCP, plugin, or built-in failures.
|
|
96
|
+
- `references/surfaces.md` — read when the user asks where a setting lives, a
|
|
97
|
+
secret must be entered, or no setup mutation exists.
|
|
@@ -1,6 +1,7 @@
|
|
|
1
1
|
# Setup actions and domain notes
|
|
2
2
|
|
|
3
|
-
Read this when choosing a `setup` mutation
|
|
3
|
+
Read this when choosing a `setup` mutation, explaining domain behavior or
|
|
4
|
+
change timing, or diagnosing MCP, plugin, or built-in failures.
|
|
4
5
|
|
|
5
6
|
## Read and navigate
|
|
6
7
|
|
|
@@ -84,7 +85,7 @@ Read this when choosing a `setup` mutation.
|
|
|
84
85
|
| Enable/disable (pause/resume) a schedule/webhook | `set_automation_enabled` |
|
|
85
86
|
| Run a schedule once now, or re-run a webhook on its latest received payload; the result posts into the owning agent's thread | `run_automation` |
|
|
86
87
|
| Webhook listener enabled state, port, or domain | `set_webhook_config` |
|
|
87
|
-
| Desktop keep-awake, run in background after window close,
|
|
88
|
+
| Desktop keep-awake, run in background after window close, notification volume (Windows), usage pin, usage rail short + long bars, expanded usage accounts, usage rail slots (three / all), Computer observe-only, Browser auto sign-in, saving logins entered in chat, browser HID device connections, session auto-archive / auto-delete days, or the representative bot overlay (`desktop.botOverlay`) | `set_desktop_settings` |
|
|
88
89
|
| Desktop theme, imported color theme, UI/code font, UI scale, code size and ligatures, chat text size, skin transparency (outer or main panels), display language, time format, side panels and their position, markdown default view, link previews, or tool activity expansion | `set_appearance` |
|
|
89
90
|
| Add a before-tool hook rule | `add_hook_rule` |
|
|
90
91
|
| Enable or disable a hook rule by its `status hooks` index | `set_hook_rule_enabled` |
|
|
@@ -107,6 +108,21 @@ Read this when choosing a `setup` mutation.
|
|
|
107
108
|
| Bring back an archived bot or agent (`definitionKind: "agent"`) | `restore_definition` |
|
|
108
109
|
| Bot display order (Bots page, Sessions bot section; `status agents` `botOrder`) | `set_bot_order` |
|
|
109
110
|
|
|
111
|
+
## When a change applies
|
|
112
|
+
|
|
113
|
+
Report the result's `appliesTo` and `appliedToCurrentSession` when it has them;
|
|
114
|
+
this table is for when it does not.
|
|
115
|
+
|
|
116
|
+
- Now (each at its next use): built-in, connector, permission, hook-rule,
|
|
117
|
+
cache-warming and automation changes, and a fallback chain: running sessions
|
|
118
|
+
on that route read it at their next hop.
|
|
119
|
+
- Next turn: an agent/bot route for its open chats, `reset_route_fallback`, a
|
|
120
|
+
workflow switch. A running turn finishes on its current model.
|
|
121
|
+
- New sessions: the Main model (`set_route` model), instruction files, most
|
|
122
|
+
other settings. An established conversation keeps its model.
|
|
123
|
+
- Session actions (`reset_route_fallback`, quota resume, `continue_in_model`)
|
|
124
|
+
answer `scope: "session"`, not a saved setting.
|
|
125
|
+
|
|
110
126
|
## Routes and workflow
|
|
111
127
|
|
|
112
128
|
Inspect `status model`, `status agents`, `status websearch`, or `status
|
|
@@ -231,14 +247,13 @@ key's source.
|
|
|
231
247
|
built-in window. A reset and a duration are mutually exclusive; provider
|
|
232
248
|
reset requires a provider.
|
|
233
249
|
- `duration: "unlimited"` (or `"never"`) stops idle clearing for a provider;
|
|
234
|
-
without `provider` it applies to the current route's provider.
|
|
250
|
+
without `provider` it applies to the current route's provider. Unlimited
|
|
251
|
+
stops idle clearing only: compaction still runs at the context limit. `status
|
|
235
252
|
autoclear` reports `unlimited`, `label` and per-provider `providerDefaults`
|
|
236
253
|
(`builtInUnlimited` marks a built-in unlimited window); a provider reset
|
|
237
254
|
restores its built-in window. Built-ins follow each provider's cache: Codex/
|
|
238
255
|
ChatGPT login unlimited, OpenAI API 30 min, Anthropic and Gemini 1h. The
|
|
239
|
-
OpenAI API
|
|
240
|
-
moving tail for 30 min (Lead, bot chats, bot task sessions, routine and
|
|
241
|
-
webhook runs), so the 30 min window clears only a cold tail. Finished
|
|
256
|
+
OpenAI API cache lasts 30 min, so that window clears only a cold cache. Finished
|
|
242
257
|
sub-agent sessions are always released on a finite timer (the provider's
|
|
243
258
|
window, e.g. DeepSeek 24h; 1h for Codex/ChatGPT logins); unlimited does not
|
|
244
259
|
disable it.
|
|
@@ -248,13 +263,18 @@ key's source.
|
|
|
248
263
|
`summaryMode`: `smart`, `maintainer`, or `self` (Conversation model).
|
|
249
264
|
`status compaction` returns them plus `aiSessionTitles`
|
|
250
265
|
(`set_ai_session_titles`).
|
|
266
|
+
- Memory master state (`set_memory_enabled`) controls the memory tools and
|
|
267
|
+
background cycles; Core Memory injection stays on regardless.
|
|
268
|
+
`set_recap_enabled` changes only the background cycles. Core Memory entries
|
|
269
|
+
are read and changed with the `memory` tool, not setup.
|
|
251
270
|
- If no supported action exposes a requested Memory interval, report it as
|
|
252
271
|
unavailable rather than editing configuration.
|
|
253
272
|
|
|
254
273
|
## Built-in features
|
|
255
274
|
|
|
256
275
|
Read `status features` and distinguish installation, enabled state, and live
|
|
257
|
-
bridge state
|
|
276
|
+
bridge state; for an unavailable built-in, say which applies: not installed,
|
|
277
|
+
installed but disabled, or bridge inactive. `status features` covers Git, Memory, Office, Code Tidy, Local
|
|
258
278
|
Provider, Auto Effort and Web Search plus Browser/Computer/Terminal bridge
|
|
259
279
|
activity and voice state (enabled, runtime installed, missing whisper/model/
|
|
260
280
|
ffmpeg components); the Browser Use / Computer Use installed and enabled
|
|
@@ -290,6 +310,7 @@ markers are read with `status desktop`.
|
|
|
290
310
|
|
|
291
311
|
## Hooks and media
|
|
292
312
|
|
|
313
|
+
- Hook rules are before-tool policy, separate from MCP and plugins.
|
|
293
314
|
- `status hooks` lists before-tool rules with their `index`, plus configured
|
|
294
315
|
lifecycle events, config sources and errors. Argument edits are reported
|
|
295
316
|
only as `hasArgumentEdit`; their contents are never returned. Indexes shift
|
|
@@ -311,10 +332,22 @@ markers are read with `status desktop`.
|
|
|
311
332
|
## Skills, MCP, and plugins
|
|
312
333
|
|
|
313
334
|
- `status skills` plus `set_disabled_skills` manage activation.
|
|
314
|
-
|
|
335
|
+
`set_disabled_skills` replaces the complete disabled list; inspect the
|
|
336
|
+
current list before changing one entry.
|
|
337
|
+
- User skills are machine-global. Enabled plugin skills and built-in skills are
|
|
338
|
+
also discoverable; project-local `.mixdog/skills` directories are ignored.
|
|
339
|
+
- MCP servers and plugins are machine-global integrations whose visibility may
|
|
340
|
+
be scoped to selected Projects: `set_extension_scope` limits a global item to
|
|
341
|
+
selected Project roots.
|
|
315
342
|
- Plugin scope is inherited by the skills and MCP integrations it contributes.
|
|
316
|
-
- MCP mutation results include reconnection state.
|
|
317
|
-
before
|
|
343
|
+
- MCP mutation results include reconnection state. For a failure, inspect the
|
|
344
|
+
reported transport error before changing command, arguments, directory,
|
|
345
|
+
environment, URL, headers, or enabled state, and before making another
|
|
346
|
+
change. MCP reads omit raw connection values that may contain secrets.
|
|
347
|
+
- An HTTP/SSE MCP server that needs OAuth waits for sign-in instead of
|
|
348
|
+
failing; setup rows then show it enabled, not connected, with no `error`.
|
|
349
|
+
Setup cannot sign in, and `reconnect_mcp` does not log in: hand off to
|
|
350
|
+
`open mcp` (Connect/Reconnect) or the TUI `/mcp login`.
|
|
318
351
|
- MCP add requires name plus command or URL. Save uses `originalName` (defaults
|
|
319
352
|
to name), so changing `name` with `originalName` renames the existing server.
|
|
320
353
|
Omitted fields remain unchanged. Switching transport requires a complete new
|
|
@@ -341,7 +374,10 @@ markers are read with `status desktop`.
|
|
|
341
374
|
("Slack 연결해 줘"): it leaves a connect card in the conversation whose
|
|
342
375
|
press runs the service's sign-in, key sheet or (Gmail: app password only)
|
|
343
376
|
dialog in Desktop; nothing connects until the user presses it, so tell them to press the card and read
|
|
344
|
-
`status mcp` afterwards.
|
|
377
|
+
`status mcp` afterwards. For each of Gmail, Calendar and Drive, multiple
|
|
378
|
+
accounts share that connector's server and its tools select an account, so
|
|
379
|
+
a second account is another card, not another server. A Google sign-in that leaves a requested access box
|
|
380
|
+
unchecked stores no token: connect again and tick every box. `add_mcp_server` with the service URL is not the
|
|
345
381
|
sign-in flow; never use it for a catalog service.
|
|
346
382
|
- Agent permissions and connector switches change only toward less: `block`
|
|
347
383
|
or `ask`, a connector off, a tool to `disabled`. An action permission
|
|
@@ -365,9 +401,12 @@ markers are read with `status desktop`.
|
|
|
365
401
|
- `status capabilities` lists supported actions and deliberate handoffs.
|
|
366
402
|
- Desktop-only operations require the owning conversation visible in the local
|
|
367
403
|
Desktop app. Never claim a persisted change from merely delivering a request.
|
|
404
|
+
- Desktop appearance changes affect the Desktop host, not a paired phone.
|
|
405
|
+
Fonts, imported color themes and skin transparency are saved per device.
|
|
406
|
+
Display language is independent of the profile response language.
|
|
368
407
|
- A display-language change returns `requiresReload`; so does a changed
|
|
369
408
|
`timeFormat` (already-shown times update after the window reloads, which is
|
|
370
|
-
never automatic).
|
|
409
|
+
never automatic). A required reload is reported, never performed.
|
|
371
410
|
- `set_appearance` also takes `uiFont` / `codeFont` (a family name, `""` =
|
|
372
411
|
built-in; text it lacks falls back to the built-in fonts), skin transparency
|
|
373
412
|
(0–100, defaults outer 20 / main 40; higher shows more of the skin picture):
|
|
@@ -389,6 +428,25 @@ markers are read with `status desktop`.
|
|
|
389
428
|
`autoClose` and `scale` (50–200, step 10). Enabling needs the
|
|
390
429
|
representative bot; `status desktop` returns `botOverlay`. Reset position
|
|
391
430
|
stays in the UI.
|
|
431
|
+
- Saved logins: `browserAutoSignIn` (the agent signs in by itself),
|
|
432
|
+
`saveChatLogins` (keep logins the user types in chat) and `browserHidDevices`
|
|
433
|
+
(sites may offer a HID device the person picks) are only switches. The logins
|
|
434
|
+
themselves (passwords, authenticator keys, imports, sign-out, delete, clear
|
|
435
|
+
all) are managed by the user in Settings → Authentication; never ask for or
|
|
436
|
+
type them in chat.
|
|
437
|
+
- `sessionAutoDeleteDays` permanently deletes archived sessions (never
|
|
438
|
+
restorable): ask for explicit approval before setting or shortening it, and
|
|
439
|
+
prefer `sessionAutoArchiveDays` (restorable) when the user only wants a
|
|
440
|
+
tidier list. Favorite and bot sessions, running or open sessions,
|
|
441
|
+
scheduled-task targets and sessions with unread replies are never archived.
|
|
442
|
+
- `set_worktree_settings`: `maxWorktrees` / `unusedDays` of `0` disable that
|
|
443
|
+
cleanup criterion; setting a positive value where it was `0`, or lowering a
|
|
444
|
+
positive one, makes cleanup remove more, and `deleteBranch:true` also
|
|
445
|
+
deletes the cleaned-up worktrees' branches: confirm with the user first. The
|
|
446
|
+
size warning is bytes in the tool (the page shows GB); `0` never asks.
|
|
447
|
+
- `set_cache_warming_enabled` covers Anthropic models only (their cache
|
|
448
|
+
expires): it warms a conversation's cache only once its own return times show
|
|
449
|
+
it pays, and adds a small usage charge, so say that when offering it.
|
|
392
450
|
- A Project removal unregisters it only.
|
|
393
451
|
- Inspect schedules/webhooks before editing. `overwrite:true` updates an
|
|
394
452
|
existing named entry and preserves omitted fields; a new schedule needs
|
|
@@ -407,7 +465,8 @@ markers are read with `status desktop`.
|
|
|
407
465
|
answers "accepted" at once and a delivery that arrives while its run is
|
|
408
466
|
active is skipped. In an attended bot conversation (chat, room, or work a
|
|
409
467
|
bot started) the same payment and permanent-delete connectors ask first,
|
|
410
|
-
allow once or deny only
|
|
468
|
+
allow once or deny only, never "항상 허용" whatever was stored before; a
|
|
469
|
+
coding session is not asked.
|
|
411
470
|
- `entry.executorAgentId` (schedule or webhook) is an enabled workflow
|
|
412
471
|
sub-agent id (Worker, …; never a bot or Mixy; unknown or turned-off ids are
|
|
413
472
|
refused). With it, each run is first carried out by that sub-agent on its own
|
|
@@ -33,13 +33,26 @@ An attached Desktop or TUI may navigate and return `opened:true`.
|
|
|
33
33
|
|
|
34
34
|
## Desktop map
|
|
35
35
|
|
|
36
|
-
- Settings → General: profile
|
|
37
|
-
|
|
38
|
-
|
|
39
|
-
|
|
40
|
-
markdown default view.
|
|
36
|
+
- Settings → General: profile, the Memory master, Voice transcription,
|
|
37
|
+
session retention (auto-archive / auto-delete), and import conversations
|
|
38
|
+
from other apps (UI-only).
|
|
39
|
+
- Settings → Display: display language, time format, link preview, confirm
|
|
40
|
+
question answers, markdown default view, side panels and their position.
|
|
41
|
+
- Settings → Work style: Web search, Load instruction files, Auto reasoning
|
|
42
|
+
base.
|
|
43
|
+
- Settings → Notifications: per-device choices (what to notify, conversation
|
|
44
|
+
scope, schedule and webhook results, quiet hours, test notification, taskbar
|
|
45
|
+
flash, skip if seen on this PC) are UI-only; only the Windows notification
|
|
46
|
+
volume has a setup action.
|
|
47
|
+
- Settings → Worktrees: worktree defaults, automatic cleanup and the managed
|
|
48
|
+
worktree list (`set_worktree_settings` covers the defaults and cleanup).
|
|
49
|
+
- Settings → Delegation mode: the instruction text of each orchestration mode
|
|
50
|
+
(UI-only free text; the mode itself is `set_orchestration_mode`).
|
|
51
|
+
- Settings → Authentication (desktop only): every saved login Browser Use may
|
|
52
|
+
use, with imports and the auto sign-in, save-chat-logins and HID switches.
|
|
53
|
+
Passwords and authenticator keys are entered only here.
|
|
41
54
|
- Settings → Context: Let AI rename sessions, auto-compact (also on model
|
|
42
|
-
change, summary model)
|
|
55
|
+
change, summary model), auto-clear and Anthropic cache warming.
|
|
43
56
|
- Settings → Model management: "Show in /model" and favorites per model.
|
|
44
57
|
- Settings → Output style: output style selection, plus the desktop-only
|
|
45
58
|
"Expand tool activity" choice (collapsed / commands and edits / everything),
|
|
@@ -49,7 +62,9 @@ An attached Desktop or TUI may navigate and return `opened:true`.
|
|
|
49
62
|
themes (about 26 color roles; import/export/remove are UI-only), installed
|
|
50
63
|
fonts, font sizes and chat text size. Saved on this device only. First-run
|
|
51
64
|
setup has a Theme step with the same System / Dark / Light / skin choice.
|
|
52
|
-
- Settings → Bots
|
|
65
|
+
- Settings → Bots: Regular check-ins and their interval, Reactions
|
|
66
|
+
(`set_heartbeat_enabled`, `set_heartbeat_interval`, `set_reactions_enabled`,
|
|
67
|
+
every surface). Desktop only: the representative bot overlay, a small
|
|
53
68
|
window with the representative bot that opens a quick message to it. On
|
|
54
69
|
by default; options: enable, Position (4-way: pinned above the message box
|
|
55
70
|
(default), pinned here, free in the app window, free on screen), keep above
|
|
@@ -76,7 +91,7 @@ An attached Desktop or TUI may navigate and return `opened:true`.
|
|
|
76
91
|
key saved there is kept in a private file (owner-only) when the server has
|
|
77
92
|
no keychain, as on Linux. Changing the model login restarts the server's
|
|
78
93
|
Mixdog after one confirmation and stops its running sessions.
|
|
79
|
-
- Settings → System: update, keep-awake, run in background after the window
|
|
94
|
+
- Settings → System: update and auto-update, keep-awake, run in background after the window
|
|
80
95
|
closes, and Doctor.
|
|
81
96
|
- Settings → Developer options: data-driven sections (approvals, risk-gated
|
|
82
97
|
providers); read them with `status developer`.
|
|
@@ -26,8 +26,9 @@ names or dotted internal names. End a step with one question, not a list of opti
|
|
|
26
26
|
|
|
27
27
|
- Setup walk-through: the user accepted setup help ("기본 설정부터 같이 할까요?"),
|
|
28
28
|
or asks how to start or what to do first.
|
|
29
|
-
- Concept question: answer
|
|
30
|
-
only when asked, and offer the matching setup step only if it is
|
|
29
|
+
- Concept question: read `references/glossary.md`, answer from it, then stop.
|
|
30
|
+
Go deeper only when asked, and offer the matching setup step only if it is
|
|
31
|
+
not done.
|
|
31
32
|
|
|
32
33
|
## Opening
|
|
33
34
|
|
|
@@ -49,29 +50,10 @@ Tier 2, recommended: Memory (`memory`), Computer Use (`computer`), Browser Use
|
|
|
49
50
|
Tier 3, optional: Voice transcription (`voice`), Local Provider
|
|
50
51
|
(`localProvider`), Code Tidy (`tidy`). Offer the missing ones in one short line.
|
|
51
52
|
|
|
52
|
-
Install only what the user accepts.
|
|
53
|
-
|
|
54
|
-
|
|
55
|
-
|
|
56
|
-
Desktop when this chat is open there. Auto reasoning can be unsupported on
|
|
57
|
-
a machine: report that instead of retrying. Computer Use and Browser Use
|
|
58
|
-
need the Desktop, and the user approves OS permission dialogs. Local
|
|
59
|
-
Provider only prepares the runtime; models belong to `local-provider`.
|
|
60
|
-
- Memory is install-only; its later on/off is `set_memory_enabled`.
|
|
61
|
-
- Visualize has no install: `set_builtin_enabled` turns it on.
|
|
62
|
-
- Media generation has no setup action: `open plugins` and let the user turn
|
|
63
|
-
it on there, then re-check `status features`.
|
|
64
|
-
|
|
65
|
-
Connectors come after the tiers. First explain in a sentence or two: a
|
|
66
|
-
connector links an outside service so bots and agents can read and act in it,
|
|
67
|
-
and anything that sends or posts asks first through an approval card. Then
|
|
68
|
-
suggest by what the user does, only these: Gmail (with a Google app password;
|
|
69
|
-
it needs 2-Step Verification on the Google account, and the connect window
|
|
70
|
-
opens the app-password page in the browser), Google Calendar, Google Drive
|
|
71
|
-
(both with the Google sign-in), Notion, Linear or Jira. GitHub is
|
|
72
|
-
not a connector; it is the signed-in GitHub CLI of Git & GitHub. Connect one with
|
|
73
|
-
`connect_connector`; it only leaves a card, so tell the user to press it and
|
|
74
|
-
confirm with `status mcp`.
|
|
53
|
+
Install only what the user accepts. Visualize and Media generation have no
|
|
54
|
+
install: once accepted, turn them on with `set_builtin_enabled` (`name`
|
|
55
|
+
`visualize` or `media`, `enabled` true). Read `references/features.md` before
|
|
56
|
+
offering add-ons or explaining, recommending, or connecting connectors.
|
|
75
57
|
|
|
76
58
|
## Walk-through
|
|
77
59
|
|
|
@@ -87,104 +69,35 @@ done. Take the first open step, explain it in a sentence or two, ask, then act.
|
|
|
87
69
|
2. project: done when a Project exists or `onboarding` lists `project` as
|
|
88
70
|
visited. A folder is optional. Register one with `save_project` for a path
|
|
89
71
|
the user names.
|
|
90
|
-
3. workflow: done when `onboarding` lists `workflow` as visited. Explain
|
|
91
|
-
|
|
92
|
-
|
|
93
|
-
|
|
94
|
-
|
|
95
|
-
|
|
96
|
-
of any length, so recommend the first one that qualifies and otherwise take
|
|
97
|
-
the next. A subscription sign-in always comes first: when the user has one,
|
|
98
|
-
take the first qualifying pick it offers before any pick that would bill an
|
|
99
|
-
API key, and only fall back to API-key providers when no subscription
|
|
100
|
-
offers one. If `asOf` is older than `staleAfterDays`, say the list may be out of
|
|
101
|
-
date. Never print ids or file names. Roles with
|
|
102
|
-
`preferOtherProviderThan` (Reviewer, Advisor) work best with the opposite
|
|
103
|
-
tendency from the model doing the work: pick the first qualifying model whose
|
|
104
|
-
provider differs from the chosen Lead's and Heavy Worker's providers, and
|
|
105
|
-
fall back to the plain first pick only when the user has a single signed-in
|
|
106
|
-
provider. Give the reason in one line: a different model family catches what
|
|
107
|
-
the main one misses. Suggest turning on the Maintainer with
|
|
108
|
-
a cheap model at the lowest effort; it also decides turns in group rooms.
|
|
109
|
-
Offer Explore the same way, from its own picks: a cheap model that
|
|
110
|
-
searches code for the Lead in separate parallel sessions.
|
|
111
|
-
A pick may carry `plans`: when the Lead provider's plan (`plan` in
|
|
112
|
-
`status providers`) is listed there, recommend that model instead. A null or
|
|
113
|
-
unlisted plan keeps the pick as is; never fail on an unknown plan.
|
|
114
|
-
Delegation and roster come from `delegation` in the data file. First ask
|
|
115
|
-
once how the user likes to work: one agent doing everything, or splitting
|
|
116
|
-
work across agents. Say plainly that Solo is a good choice at any plan: it
|
|
117
|
-
is the most efficient, and with a strong model its quality is often not much
|
|
118
|
-
lower; delegation trades more usage for speed and parallel work. Then suggest
|
|
119
|
-
the mode whose `plans` lists the Lead provider's plan (the default
|
|
120
|
-
suggestion), but the user's preference overrides it. Ask once; set it with
|
|
121
|
-
`set_orchestration_mode`, and say it can be changed later in settings. The
|
|
122
|
-
roster follows the chosen mode: turn on only `baseRoster` plus that mode's
|
|
123
|
-
`agents`, using `set_agent_route` with `route.disabled` set false for what
|
|
124
|
-
the user accepts and true for the rest; do not enable others. Offer
|
|
125
|
-
`leanReview` for Solo or focused as a cost saver: the Lead reviews its own
|
|
126
|
-
work, Reviewer stays off, Advisor is on but used rarely (only when stuck).
|
|
127
|
-
Advisor in balanced or swarm is a light, occasional second opinion from a
|
|
128
|
-
different model family at key decision points, not on every task. Say in
|
|
129
|
-
one line that more agents can be turned on later on the Workflows page.
|
|
130
|
-
Right after the models are picked, explain Auto reasoning: it adjusts
|
|
131
|
-
reasoning depth per message around a medium baseline, so hard work may
|
|
132
|
-
come out slightly lower in quality but cost is much lower. The Lead runs
|
|
133
|
-
the main coding work, where stability matters most, so also recommend
|
|
134
|
-
fixing the Lead's effort at High or above in the model picker (choosing an
|
|
135
|
-
effort there turns Auto off for that model). Ask whether to turn Auto on
|
|
136
|
-
(`install_builtin` `autoEffort`). If declined, say in one line that it can
|
|
137
|
-
be turned on later by choosing Auto at the top of the model picker's
|
|
138
|
-
reasoning effort list.
|
|
72
|
+
3. workflow: done when `onboarding` lists `workflow` as visited. Explain that a
|
|
73
|
+
workflow sets how work gets done and agents are helpers behind a task, then
|
|
74
|
+
`open workflows` so the user can turn agents on and pick models
|
|
75
|
+
(`set_agent_route` once they name a model). Read
|
|
76
|
+
`references/workflow-models.md` before recommending any model, mode, roster
|
|
77
|
+
or Auto reasoning.
|
|
139
78
|
4. remote: done when a device is paired (`status connection`) or `remote` is
|
|
140
79
|
visited. Pairing is user-operated: `open connection`.
|
|
141
|
-
5. extensions: done when `extensions` is visited. Explain plugins
|
|
142
|
-
offer only built-ins not already covered
|
|
80
|
+
5. extensions: done when `extensions` is visited. Explain that plugins add
|
|
81
|
+
tools and skills teach how to work; offer only built-ins not already covered
|
|
82
|
+
by the tiers above.
|
|
143
83
|
6. mixy: done once Mixy exists, which is already true when this skill runs in
|
|
144
84
|
Mixy's chat. Offer to make another bot when the user wants one.
|
|
145
85
|
|
|
146
86
|
After a change, report it in a sentence and move on. The user may stop at any
|
|
147
87
|
step; do not insist.
|
|
148
88
|
|
|
149
|
-
## Glossary
|
|
150
|
-
|
|
151
|
-
- Bot (봇): a contact you chat with in bot mode (봇 모드). Mixy is the representative
|
|
152
|
-
bot; the user can have it make others, such as a GitHub Watcher.
|
|
153
|
-
- Agent (에이전트): a workflow sub-agent such as Worker or Reviewer that
|
|
154
|
-
does part of a task. Users add their own on the Workflows page with the
|
|
155
|
-
"에이전트 +" button. Bots are people-facing; agents work behind a task.
|
|
156
|
-
- Default agents (기본 에이전트): built-in helpers listed on Workflows, such as
|
|
157
|
-
Web Search (runs the web search tool), the Maintainer (memory upkeep and
|
|
158
|
-
session titles) and Explore (code search for the Lead). They need a model
|
|
159
|
-
route to work.
|
|
160
|
-
- Agent mode (에이전트 모드): a coding work session run by Lead and agents.
|
|
161
|
-
A session a bot starts for a job is an agent task session (에이전트 작업
|
|
162
|
-
세션); others are just sessions (세션).
|
|
163
|
-
- Sessions and projects: an agent-mode session runs in a Project, a folder whose
|
|
164
|
-
instructions and memories it loads.
|
|
165
|
-
- Workflows and work rules: a workflow sets how a request gets done; work
|
|
166
|
-
rules are the instructions it applies.
|
|
167
|
-
- Connectors and extensions: connectors are MCP services such as Notion or
|
|
168
|
-
Linear (`connect_connector`, the user presses the card); extensions are
|
|
169
|
-
plugins that add tools and skills that teach how to work.
|
|
170
|
-
- Group rooms (그룹 대화방): one chat with several bots. `@name` or "봇 부르기"
|
|
171
|
-
addresses one bot; otherwise the Maintainer decides who answers next.
|
|
172
|
-
- Schedules and webhooks: a timed or incoming-event trigger (`save_automation`).
|
|
173
|
-
Each run is one turn in its bot's own chat, with that bot's persona, rules,
|
|
174
|
-
permissions, memory and model. Without a bot they cannot be used. A routine
|
|
175
|
-
can also name an executor sub-agent (`executorAgentId`, e.g. Worker): that
|
|
176
|
-
sub-agent does each run on its own model and the bot only relays its report.
|
|
177
|
-
In the Schedules/Webhooks editor this is the "Executor" field (Korean UI:
|
|
178
|
-
"실행 에이전트"); its default "The bot itself" ("직접 처리") means the bot
|
|
179
|
-
does the run itself.
|
|
180
|
-
- Approval cards: when a tool needs permission it waits in the chat until the
|
|
181
|
-
user approves or declines.
|
|
182
|
-
- Phone remote access: pair a phone under Settings → Connection to follow and
|
|
183
|
-
steer tasks away from the PC.
|
|
184
|
-
|
|
185
89
|
## Pitfalls
|
|
186
90
|
|
|
187
|
-
- Do not claim a step is done from memory;
|
|
188
|
-
-
|
|
189
|
-
|
|
91
|
+
- Do not claim a step is done from memory; read state before the step.
|
|
92
|
+
- Trust a receipt that proves the change. Re-check `status` only after a UI
|
|
93
|
+
handoff (`open`: ask the user to confirm first), a pending install, or a
|
|
94
|
+
receipt that does not show the persisted value.
|
|
190
95
|
- Do not run several steps in one message or queue the next question early.
|
|
96
|
+
|
|
97
|
+
## Resources
|
|
98
|
+
|
|
99
|
+
- `references/glossary.md` — read only for a concept question.
|
|
100
|
+
- `references/features.md` — read before offering add-ons or explaining,
|
|
101
|
+
recommending, or connecting connectors.
|
|
102
|
+
- `references/workflow-models.md` — read at the workflow step before
|
|
103
|
+
recommending models, roster or Auto reasoning.
|
|
@@ -0,0 +1,36 @@
|
|
|
1
|
+
# Feature installs and connectors
|
|
2
|
+
|
|
3
|
+
Read this before offering add-ons or explaining, recommending, or connecting
|
|
4
|
+
connectors. Visualize and Media generation are toggles and are handled in
|
|
5
|
+
`SKILL.md`.
|
|
6
|
+
|
|
7
|
+
## Installing add-ons
|
|
8
|
+
|
|
9
|
+
Use `install_builtin` with the name for git, memory, autoEffort, computer,
|
|
10
|
+
browser, voice, localProvider and tidy.
|
|
11
|
+
|
|
12
|
+
- Git installs system Git through the Desktop when this chat is open there.
|
|
13
|
+
- Auto reasoning can be unsupported on a machine: report that instead of
|
|
14
|
+
retrying.
|
|
15
|
+
- Computer Use and Browser Use need the Desktop, and the user approves OS
|
|
16
|
+
permission dialogs.
|
|
17
|
+
- Local Provider only prepares the runtime; models belong to `local-provider`.
|
|
18
|
+
- Memory is install-only; its later on/off is `set_memory_enabled`.
|
|
19
|
+
|
|
20
|
+
## Connectors
|
|
21
|
+
|
|
22
|
+
Explain first, in a sentence or two: a connector links an outside service so
|
|
23
|
+
bots and agents can read and act in it, and anything that sends or posts asks
|
|
24
|
+
first through an approval card. Then suggest by what the user does, only
|
|
25
|
+
these:
|
|
26
|
+
|
|
27
|
+
- Gmail: needs a Google app password, which needs 2-Step Verification on the
|
|
28
|
+
Google account; the connect window opens the app-password page in the
|
|
29
|
+
browser.
|
|
30
|
+
- Google Calendar and Google Drive: both use the Google sign-in.
|
|
31
|
+
- Notion, Linear or Jira.
|
|
32
|
+
|
|
33
|
+
GitHub is not a connector; it is the signed-in GitHub CLI of Git & GitHub.
|
|
34
|
+
|
|
35
|
+
Connect one with `connect_connector`. It only leaves a card: tell the user to
|
|
36
|
+
press it, then confirm with `status mcp`.
|
|
@@ -0,0 +1,38 @@
|
|
|
1
|
+
# Glossary
|
|
2
|
+
|
|
3
|
+
Read this to answer a concept question. Answer in a sentence or two from the
|
|
4
|
+
matching entry, then stop.
|
|
5
|
+
|
|
6
|
+
- Bot (봇): a contact you chat with in bot mode (봇 모드). Mixy is the representative
|
|
7
|
+
bot; the user can have it make others, such as a GitHub Watcher.
|
|
8
|
+
- Agent (에이전트): a workflow sub-agent such as Worker or Reviewer that
|
|
9
|
+
does part of a task. Users add their own on the Workflows page with the
|
|
10
|
+
"에이전트 +" button. Bots are people-facing; agents work behind a task.
|
|
11
|
+
- Default agents (기본 에이전트): built-in helpers listed on Workflows, such as
|
|
12
|
+
Web Search (runs the web search tool), the Maintainer (memory upkeep and
|
|
13
|
+
session titles) and Explore (code search for the Lead). They need a model
|
|
14
|
+
route to work.
|
|
15
|
+
- Agent mode (에이전트 모드): a coding work session run by Lead and agents.
|
|
16
|
+
A session a bot starts for a job is an agent task session (에이전트 작업
|
|
17
|
+
세션); others are just sessions (세션).
|
|
18
|
+
- Sessions and projects: an agent-mode session runs in a Project, a folder whose
|
|
19
|
+
instructions and memories it loads.
|
|
20
|
+
- Workflows and work rules: a workflow sets how a request gets done; work
|
|
21
|
+
rules are the instructions it applies.
|
|
22
|
+
- Connectors and extensions: connectors are MCP services such as Notion or
|
|
23
|
+
Linear (`connect_connector`, the user presses the card); extensions are
|
|
24
|
+
plugins that add tools and skills that teach how to work.
|
|
25
|
+
- Group rooms (그룹 대화방): one chat with several bots. `@name` or "봇 부르기"
|
|
26
|
+
addresses one bot; otherwise the Maintainer decides who answers next.
|
|
27
|
+
- Schedules and webhooks: a timed or incoming-event trigger (`save_automation`).
|
|
28
|
+
Each run is one turn in its bot's own chat, with that bot's persona, rules,
|
|
29
|
+
permissions, memory and model. Without a bot they cannot be used. A routine
|
|
30
|
+
can also name an executor sub-agent (`executorAgentId`, e.g. Worker): that
|
|
31
|
+
sub-agent does each run on its own model and the bot only relays its report.
|
|
32
|
+
In the Schedules/Webhooks editor this is the "Executor" field (Korean UI:
|
|
33
|
+
"실행 에이전트"); its default "The bot itself" ("직접 처리") means the bot
|
|
34
|
+
does the run itself.
|
|
35
|
+
- Approval cards: when a tool needs permission it waits in the chat until the
|
|
36
|
+
user approves or declines.
|
|
37
|
+
- Phone remote access: pair a phone under Settings → Connection to follow and
|
|
38
|
+
steer tasks away from the PC.
|