open-multi-agent-kit 0.94.0 → 0.95.0

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.
Files changed (109) hide show
  1. package/CHANGELOG.md +38 -0
  2. package/README.md +100 -646
  3. package/dist/core/agent-session.d.ts.map +1 -1
  4. package/dist/core/agent-session.js +11 -11
  5. package/dist/core/agent-session.js.map +1 -1
  6. package/dist/core/auth-storage.d.ts +27 -2
  7. package/dist/core/auth-storage.d.ts.map +1 -1
  8. package/dist/core/auth-storage.js +239 -56
  9. package/dist/core/auth-storage.js.map +1 -1
  10. package/dist/core/context-budget-v2-cache-invalidation.d.ts.map +1 -1
  11. package/dist/core/context-budget-v2-cache-invalidation.js +1 -1
  12. package/dist/core/context-budget-v2-cache-invalidation.js.map +1 -1
  13. package/dist/core/context-budget-v2-cache-provider.d.ts +5 -1
  14. package/dist/core/context-budget-v2-cache-provider.d.ts.map +1 -1
  15. package/dist/core/context-budget-v2-cache-provider.js +37 -7
  16. package/dist/core/context-budget-v2-cache-provider.js.map +1 -1
  17. package/dist/core/context-budget-v2-cache-telemetry.d.ts +1 -1
  18. package/dist/core/context-budget-v2-cache-telemetry.d.ts.map +1 -1
  19. package/dist/core/context-budget-v2-cache-telemetry.js +2 -1
  20. package/dist/core/context-budget-v2-cache-telemetry.js.map +1 -1
  21. package/dist/core/context-budget-v2-cache.d.ts.map +1 -1
  22. package/dist/core/context-budget-v2-cache.js +3 -0
  23. package/dist/core/context-budget-v2-cache.js.map +1 -1
  24. package/dist/core/context-budget-v2-representation-cache.d.ts.map +1 -1
  25. package/dist/core/context-budget-v2-representation-cache.js +4 -3
  26. package/dist/core/context-budget-v2-representation-cache.js.map +1 -1
  27. package/dist/core/context-budget-v2-selection.d.ts.map +1 -1
  28. package/dist/core/context-budget-v2-selection.js +4 -0
  29. package/dist/core/context-budget-v2-selection.js.map +1 -1
  30. package/dist/core/context-budget-v2-types.d.ts +3 -0
  31. package/dist/core/context-budget-v2-types.d.ts.map +1 -1
  32. package/dist/core/context-budget-v2-types.js.map +1 -1
  33. package/dist/core/extensions/types.d.ts +2 -0
  34. package/dist/core/extensions/types.d.ts.map +1 -1
  35. package/dist/core/extensions/types.js.map +1 -1
  36. package/dist/core/github-repository.d.ts +3 -0
  37. package/dist/core/github-repository.d.ts.map +1 -0
  38. package/dist/core/github-repository.js +3 -0
  39. package/dist/core/github-repository.js.map +1 -0
  40. package/dist/core/model-registry.d.ts +3 -3
  41. package/dist/core/model-registry.d.ts.map +1 -1
  42. package/dist/core/model-registry.js +9 -8
  43. package/dist/core/model-registry.js.map +1 -1
  44. package/dist/core/provider-usage.d.ts +48 -0
  45. package/dist/core/provider-usage.d.ts.map +1 -0
  46. package/dist/core/provider-usage.js +780 -0
  47. package/dist/core/provider-usage.js.map +1 -0
  48. package/dist/core/sdk.d.ts.map +1 -1
  49. package/dist/core/sdk.js +10 -0
  50. package/dist/core/sdk.js.map +1 -1
  51. package/dist/core/session-termination.d.ts.map +1 -1
  52. package/dist/core/session-termination.js +8 -1
  53. package/dist/core/session-termination.js.map +1 -1
  54. package/dist/core/settings-manager.d.ts +1 -0
  55. package/dist/core/settings-manager.d.ts.map +1 -1
  56. package/dist/core/settings-manager.js +15 -8
  57. package/dist/core/settings-manager.js.map +1 -1
  58. package/dist/core/skills-catalog-cache.d.ts +1 -0
  59. package/dist/core/skills-catalog-cache.d.ts.map +1 -1
  60. package/dist/core/skills-catalog-cache.js +42 -11
  61. package/dist/core/skills-catalog-cache.js.map +1 -1
  62. package/dist/core/slash-commands.d.ts.map +1 -1
  63. package/dist/core/slash-commands.js +1 -0
  64. package/dist/core/slash-commands.js.map +1 -1
  65. package/dist/core/tools/read.d.ts +1 -1
  66. package/dist/core/tools/read.d.ts.map +1 -1
  67. package/dist/core/tools/read.js.map +1 -1
  68. package/dist/modes/interactive/components/footer.d.ts.map +1 -1
  69. package/dist/modes/interactive/components/footer.js.map +1 -1
  70. package/dist/modes/interactive/components/index.d.ts +1 -0
  71. package/dist/modes/interactive/components/index.d.ts.map +1 -1
  72. package/dist/modes/interactive/components/index.js +1 -0
  73. package/dist/modes/interactive/components/index.js.map +1 -1
  74. package/dist/modes/interactive/components/model-selector.d.ts.map +1 -1
  75. package/dist/modes/interactive/components/model-selector.js +5 -3
  76. package/dist/modes/interactive/components/model-selector.js.map +1 -1
  77. package/dist/modes/interactive/components/oauth-account-selector.d.ts +16 -0
  78. package/dist/modes/interactive/components/oauth-account-selector.d.ts.map +1 -0
  79. package/dist/modes/interactive/components/oauth-account-selector.js +61 -0
  80. package/dist/modes/interactive/components/oauth-account-selector.js.map +1 -0
  81. package/dist/modes/interactive/components/oauth-selector.d.ts.map +1 -1
  82. package/dist/modes/interactive/components/oauth-selector.js +25 -4
  83. package/dist/modes/interactive/components/oauth-selector.js.map +1 -1
  84. package/dist/modes/interactive/components/status-sidebar.d.ts +21 -1
  85. package/dist/modes/interactive/components/status-sidebar.d.ts.map +1 -1
  86. package/dist/modes/interactive/components/status-sidebar.js +165 -6
  87. package/dist/modes/interactive/components/status-sidebar.js.map +1 -1
  88. package/dist/modes/interactive/interactive-mode.d.ts +2 -0
  89. package/dist/modes/interactive/interactive-mode.d.ts.map +1 -1
  90. package/dist/modes/interactive/interactive-mode.js +99 -15
  91. package/dist/modes/interactive/interactive-mode.js.map +1 -1
  92. package/dist/utils/image-resize-core.d.ts.map +1 -1
  93. package/dist/utils/image-resize-core.js +2 -2
  94. package/dist/utils/image-resize-core.js.map +1 -1
  95. package/docs/development.md +3 -2
  96. package/docs/providers.md +19 -4
  97. package/docs/settings.md +9 -9
  98. package/docs/usage.md +10 -9
  99. package/examples/extensions/custom-provider-anthropic/package-lock.json +2 -2
  100. package/examples/extensions/custom-provider-anthropic/package.json +1 -1
  101. package/examples/extensions/custom-provider-gitlab-duo/package.json +1 -1
  102. package/examples/extensions/gondolin/package-lock.json +2 -2
  103. package/examples/extensions/gondolin/package.json +1 -1
  104. package/examples/extensions/sandbox/package-lock.json +2 -2
  105. package/examples/extensions/sandbox/package.json +1 -1
  106. package/examples/extensions/with-deps/package-lock.json +2 -2
  107. package/examples/extensions/with-deps/package.json +1 -1
  108. package/npm-shrinkwrap.json +24 -24
  109. package/package.json +6 -6
package/README.md CHANGED
@@ -1,723 +1,177 @@
1
- <p align="center">
2
- <img src="https://raw.githubusercontent.com/dmae97/open-multi-agent-kit/v0.90.4/readmeasset/omk-control.webp" alt="OMK//CONTROL Night City Ops Console for routing agents, evidence gates, telemetry, MCP scope, and operator control" width="100%" />
3
- </p>
1
+ # OMK
4
2
 
5
- <h1 align="center">OMK</h1>
3
+ Provider-neutral terminal coding agent with multi-agent orchestration, scoped tools, replayable sessions, and evidence-backed verification.
6
4
 
7
- <p align="center">
8
- <strong>OMK//CONTROL — provider-neutral multi-agent control plane for coding workflows.</strong>
9
- </p>
5
+ OMK supports interactive terminal use, non-interactive output, RPC integration, and an embeddable TypeScript SDK. It works with API-key and subscription providers without making one provider the control plane.
10
6
 
11
- <p align="center">
12
- Models execute. OMK routes, verifies, measures, and controls.
13
- </p>
7
+ ## Install
14
8
 
15
- <p align="center">
16
- <a href="https://discord.com/invite/3cU7Bz4UPx"><img alt="Discord" src="https://img.shields.io/badge/discord-community-5865F2?style=flat-square&logo=discord&logoColor=white" /></a>
17
- <a href="https://www.npmjs.com/package/open-multi-agent-kit"><img alt="npm" src="https://img.shields.io/npm/v/open-multi-agent-kit?style=flat-square" /></a>
18
- <a href="https://github.com/dmae97/omk/releases/tag/v0.94.0"><img alt="Release" src="https://img.shields.io/badge/release-v0.94.0-00d7ff?style=flat-square" /></a>
19
- </p>
20
-
21
- > New issues and PRs from new contributors are auto-closed by default. Maintainers review auto-closed issues daily. See [CONTRIBUTING.md](https://github.com/dmae97/open-multi-agent-kit/blob/v0.90.4/CONTRIBUTING.md).
22
-
23
- ---
24
-
25
- ## OMK//CONTROL TUI
26
-
27
- <p align="center">
28
- <img src="https://raw.githubusercontent.com/dmae97/open-multi-agent-kit/v0.90.4/readmeasset/omk_tui.png" alt="OMK//CONTROL terminal dashboard — live DAG lanes, provider routing, MCP health, evidence gates, and telemetry in Night City Ops Console style" width="100%" />
29
- </p>
30
-
31
- The OMK//CONTROL startup surface is the default operator view. The header reads `omk v<package.version> · OMK//CONTROL`, using the published `open-multi-agent-kit` package version as the single source of truth.
32
-
33
- The default dark TUI theme uses the `omk-control-grid-dark` Night City palette and keeps the control sidebar focused on route, evidence, loop, MCP, runtime, skills, and context budget state.
34
-
35
- ## Release v0.91.0
36
-
37
- This feature release moves footer resource metrics to whole-machine telemetry, ships the WCAG-verified Aurora theme pair, stabilizes the AdaptOrch Work Packet Loop package as a CLI runtime dependency, and adds two built-in subscription providers (Qwen, Grok).
38
-
39
- | Area | What changed |
40
- |------|--------------|
41
- | Footer metrics | CPU/MEM now reflect the whole machine (`os.cpus()` busy %, `totalmem - freemem`) instead of the OMK process, with percentage-based warning/error thresholds. |
42
- | Themes | `omk-aurora-dark` / `omk-aurora-light`: WCAG-verified contrast (body ≥14:1, semantic ≥4.5:1), stepped thinking-level ramp, aliases `aurora` / `aurora-dark` / `aurora-light`. |
43
- | AdaptOrch | `omk-adaptorch-wpl` promoted to stable as a CLI runtime dependency; opt-in global-only `adaptorchBridge` advisory hints for the auto thinking-level resolver (default off). |
44
- | Providers | New built-in subscription providers under `/login`: Qwen (Qwen Code Subscription) and Grok (xAI OAuth Proxy). |
45
-
46
- GitHub-focused release notes live in [RELEASE_NOTES_v0.94.0.md](https://github.com/dmae97/omk/blob/v0.94.0/.github/RELEASE_NOTES_v0.94.0.md). The GitHub release workflow also extracts the canonical release body from [packages/coding-agent/CHANGELOG.md](https://github.com/dmae97/omk/blob/v0.94.0/packages/coding-agent/CHANGELOG.md).
47
-
48
- ## Release v0.90.8
49
-
50
- This patch release adds the tool-free GPT-5.6 MoA model, ordered path-safe tool-batch waves, global context-budget controls, and evidence-gated computer-use integrations.
51
-
52
- | Area | What changed |
53
- |------|--------------|
54
- | Models | Added `openai-codex/gpt-5.6-moa`: bounded concurrent Sol/Terra analysis with a single Sol synthesis, plus hardened Codex terminal and cancellation handling. |
55
- | Agent loop | Ordered `partitionToolBatchWaves` preserve safe parallel reads while path conflicts and unknown calls remain sequential. |
56
- | Context control | Added global `contextBudget.enabled` and `compaction.model`; planner cache selection stays within the remaining tier budget. |
57
- | Evidence / verification | Correctness Wall fixtureless live OA now requires a bound MCP handler and otherwise stays preview-only; the evidence ledger is tamper-evident. |
58
- | Computer use | Added a project-local Stagehand extension and `omk-computeruse` skill with explicit operator approval and redacted results. |
59
- | Release safety | Nested extension `node_modules` are excluded from release staging while extension source and lockfiles remain versioned. |
60
-
61
- GitHub-focused release notes live in [RELEASE_NOTES_v0.90.8.md](https://github.com/dmae97/omk/blob/v0.90.8/.github/RELEASE_NOTES_v0.90.8.md). The GitHub release workflow also extracts the canonical release body from [packages/coding-agent/CHANGELOG.md](https://github.com/dmae97/omk/blob/v0.90.8/packages/coding-agent/CHANGELOG.md).
62
-
63
- OMK is a minimal terminal coding harness. Adapt omk to your workflows, not the other way around, without having to fork and modify omk internals. Extend it with TypeScript [Extensions](#extensions), [Skills](#skills), [Prompt Templates](#prompt-templates), and [Themes](#themes). Put your extensions, skills, prompt templates, and themes in [OMK Packages](#omk-packages) and share them with others via npm or git.
64
-
65
- OMK ships with powerful defaults but skips features like sub agents and plan mode. Instead, you can ask omk to build what you want or install a third party omk package that matches your workflow.
66
-
67
- OMK runs in four modes: interactive, print or JSON, RPC for process integration, and an SDK for embedding in your own apps. See [openclaw/openclaw](https://github.com/openclaw/openclaw) for a real-world SDK integration.
68
-
69
- ## Share your OSS coding agent sessions
70
-
71
- If you use omk for open source work, please share your coding agent sessions.
72
-
73
- Public OSS session data helps improve models, prompts, tools, and evaluations using real development workflows.
74
-
75
- For the full explanation, see [this post on X](https://x.com/badlogicgames/status/2037811643774652911).
76
-
77
- To publish sessions, share OMK session JSONL files from `~/.omk/agent/sessions/` with a Hugging Face dataset or another public archive. Include the OMK version, provider/model, and repository context when it is safe to publish.
78
-
79
- ## Table of Contents
80
-
81
- - [Quick Start](#quick-start)
82
- - [Providers & Models](#providers--models)
83
- - [Interactive Mode](#interactive-mode)
84
- - [Editor](#editor)
85
- - [Commands](#commands)
86
- - [Keyboard Shortcuts](#keyboard-shortcuts)
87
- - [Message Queue](#message-queue)
88
- - [Sessions](#sessions)
89
- - [Branching](#branching)
90
- - [Compaction](#compaction)
91
- - [Settings](#settings)
92
- - [Context Files](#context-files)
93
- - [Customization](#customization)
94
- - [Prompt Templates](#prompt-templates)
95
- - [Skills](#skills)
96
- - [Extensions](#extensions)
97
- - [Themes](#themes)
98
- - [OMK Packages](#omk-packages)
99
- - [Programmatic Usage](#programmatic-usage)
100
- - [Philosophy](#philosophy)
101
- - [CLI Reference](#cli-reference)
102
-
103
- ---
104
-
105
- ## Quick Start
9
+ Requires Node.js 22.19 or newer.
106
10
 
107
11
  ```bash
108
- npm install -g --ignore-scripts open-multi-agent-kit
12
+ npm install -g open-multi-agent-kit --ignore-scripts
13
+ omk --version
14
+ omk
109
15
  ```
110
16
 
111
- `--ignore-scripts` disables dependency lifecycle scripts during install. OMK does not require install scripts for normal npm installs.
112
-
113
- Installer alternative:
17
+ Run without a global install:
114
18
 
115
19
  ```bash
116
- curl -fsSL https://omk.dev/install.sh | sh
20
+ npx --ignore-scripts open-multi-agent-kit
117
21
  ```
118
22
 
119
- Authenticate with an API key or an existing subscription:
23
+ The published package name remains `open-multi-agent-kit`; the command is `omk`.
24
+
25
+ ## Quick Start
26
+
27
+ Start OMK in a project directory:
120
28
 
121
29
  ```bash
30
+ cd your-project
122
31
  omk
123
- /login # Then select provider and sign in with OAuth or save an API key
124
32
  ```
125
33
 
126
- Then just talk to omk. By default, omk gives the model four tools: `read`, `write`, `edit`, and `bash`. The model uses these to fulfill your requests. Add capabilities via [skills](#skills), [prompt templates](#prompt-templates), [extensions](#extensions), or [omk packages](#omk-packages).
34
+ Then:
127
35
 
128
- **Platform notes:** [Windows](docs/windows.md) | [Termux (Android)](docs/termux.md) | [tmux](docs/tmux.md) | [Terminal setup](docs/terminal-setup.md) | [Shell aliases](docs/shell-aliases.md)
36
+ 1. Run `/login` to configure an OAuth subscription or API key.
37
+ 2. Run `/model` to choose an available model.
38
+ 3. Describe the change and the evidence required for completion.
39
+ 4. Review tool calls, diffs, diagnostics, and test results before accepting the work.
129
40
 
130
- ---
41
+ OMK stores user configuration under `~/.omk/agent/` and project configuration under `.omk/`.
131
42
 
132
43
  ## Providers & Models
133
44
 
134
- For each built-in provider, omk maintains a list of tool-capable models, updated with every release. Authenticate via subscription (`/login`) or API key, then select any model from that provider via `/model` (or Ctrl+L).
135
-
136
- **Subscriptions:**
137
- - Anthropic Claude Pro/Max
138
- - OpenAI ChatGPT Plus/Pro (Codex)
139
- - GitHub Copilot
140
-
141
- **API keys:**
142
- - Anthropic
143
- - Ant Ling
144
- - OpenAI
145
- - Azure OpenAI
146
- - DeepSeek
147
- - NVIDIA NIM
148
- - Google Gemini
149
- - Google Vertex
150
- - Amazon Bedrock
151
- - Mistral
152
- - Groq
153
- - Cerebras
154
- - Cloudflare AI Gateway
155
- - Cloudflare Workers AI
156
- - xAI
157
- - OpenRouter
158
- - Vercel AI Gateway
159
- - ZAI
160
- - ZAI Coding Plan (China)
161
- - OpenCode Zen
162
- - OpenCode Go
163
- - Hugging Face
164
- - Fireworks
165
- - Together AI
166
- - Kimi For Coding
167
- - MiniMax
168
- - Xiaomi MiMo
169
- - Xiaomi MiMo Token Plan (China)
170
- - Xiaomi MiMo Token Plan (Amsterdam)
171
- - Xiaomi MiMo Token Plan (Singapore)
172
- - Zyloo
173
-
174
- See [docs/providers.md](docs/providers.md) for detailed setup instructions.
175
-
176
- **Custom providers & models:** Add providers via `~/.omk/agent/models.json` if they speak a supported API (OpenAI, Anthropic, Google). For custom APIs or OAuth, use extensions. See [docs/models.md](docs/models.md) and [docs/custom-provider.md](docs/custom-provider.md).
177
-
178
- ---
179
-
180
- ## Interactive Mode
45
+ OMK includes provider adapters for Anthropic, OpenAI Codex, Google, OpenCode, Kimi, Qwen, ZAI/GLM, Grok, and other compatible services. Availability depends on the credentials and endpoints configured on the current machine.
181
46
 
182
- <p align="center"><img src="https://raw.githubusercontent.com/dmae97/open-multi-agent-kit/v0.90.4/packages/coding-agent/docs/images/interactive-mode.png" alt="Interactive Mode" width="600"></p>
47
+ - `/login` adds or selects credentials.
48
+ - `/logout` removes credentials for a selected provider.
49
+ - `/model` lists models available to the current credential set.
50
+ - The optional status rail shows independent provider quota windows when an official API or passive response signal is available.
51
+ - Missing quota values are shown as unavailable rather than estimated.
183
52
 
184
- The interface from top to bottom:
53
+ OAuth credentials are stored locally in `~/.omk/agent/auth.json`. Multi-account providers keep accounts separate and send requests only through the explicitly selected account.
185
54
 
186
- - **Startup header** - Shows shortcuts (`/hotkeys` for all), loaded AGENTS.md files, prompt templates, skills, and extensions
187
- - **Messages** - Your messages, assistant responses, tool calls and results, notifications, errors, and extension UI
188
- - **Editor** - Where you type; border color indicates thinking level
189
- - **Footer** - Working directory, session name, total token/cache usage (`↑` input, `↓` output, `R` cache read, `W` cache write, `CH` latest cache hit rate), cost, context usage, current model
55
+ See [Provider setup](https://github.com/dmae97/omk/blob/main/packages/coding-agent/docs/providers.md) for provider-specific authentication, quota behavior, and environment variables.
190
56
 
191
- The editor can be temporarily replaced by other UI, like built-in `/settings` or custom UI from extensions (e.g., a Q&A tool that lets the user answer model questions in a structured format). [Extensions](#extensions) can also replace the editor, add widgets above/below it, a status line, custom footer, or overlays.
57
+ ### Kimi For Coding
192
58
 
193
- ### Editor
59
+ Configure Kimi through `/login` or the documented Kimi API-key environment variable. Requests use the fixed provider endpoint selected by the provider adapter.
194
60
 
195
- | Feature | How |
196
- |---------|-----|
197
- | File reference | Type `@` to fuzzy-search project files |
198
- | Path completion | Tab to complete paths |
199
- | Multi-line | Shift+Enter (or Ctrl+Enter on Windows Terminal) |
200
- | Images | Ctrl+V to paste (Alt+V on Windows), or drag onto terminal |
201
- | Skills/bash launcher | `!` opens skill autocomplete, `!skill:name prompt` invokes a skill, `! command` runs bash with context, `!! command` runs bash without context |
61
+ ### Hugging Face
202
62
 
203
- Standard editing keybindings for delete word, undo, etc. See [docs/keybindings.md](docs/keybindings.md).
63
+ Set the documented Hugging Face token and select a compatible model. See the provider guide for current model and endpoint requirements.
204
64
 
205
- ### Commands
65
+ ## Interactive Mode
206
66
 
207
- Type `/` in the editor to trigger commands. [Extensions](#extensions) can register custom commands, [skills](#skills) are available as `/skill:name`, `!skill:name`, or `!name` for known skills, and [prompt templates](#prompt-templates) expand via `/templatename`.
67
+ Type `/` in the editor to open command completion.
208
68
 
209
69
  | Command | Description |
210
- |---------|-------------|
211
- | `/login`, `/logout` | OAuth authentication |
212
- | `/model` | Switch models |
213
- | `/scoped-models` | Enable/disable models for Ctrl+P cycling |
214
- | `/settings` | Thinking level, theme, message delivery, transport |
215
- | `/resume` | Pick from previous sessions |
70
+ | --- | --- |
71
+ | `/login`, `/logout` | Manage OAuth or API-key credentials |
72
+ | `/model` | Select a model |
73
+ | `/think` | Select a thinking level |
74
+ | `/settings` | Change interactive settings |
75
+ | `/resume` | Open a previous session |
216
76
  | `/new` | Start a new session |
217
- | `/name <name>` | Set session display name |
218
- | `/session` | Show session info (file, ID, messages, tokens, cost) |
219
- | `/tree` | Jump to any point in the session and continue from there |
220
- | `/trust` | Save project trust decision for future sessions (restart required) |
221
- | `/fork` | Create a new session from a previous user message |
222
- | `/clone` | Duplicate the current active branch into a new session |
223
- | `/compact [prompt]` | Manually compact context, optional custom instructions |
224
- | `/copy` | Copy last assistant message to clipboard |
225
- | `/export [file]` | Export session to HTML file |
226
- | `/share` | Upload as private GitHub gist with shareable HTML link |
227
- | `/reload` | Reload keybindings, extensions, skills, prompts, and context files (themes hot-reload automatically) |
228
- | `/hotkeys` | Show all keyboard shortcuts |
229
- | `/changelog` | Display version history |
230
- | `/quit` | Quit omk |
231
-
232
- ### Keyboard Shortcuts
233
-
234
- See `/hotkeys` for the full list. Customize via `~/.omk/agent/keybindings.json`. See [docs/keybindings.md](docs/keybindings.md).
235
-
236
- **Commonly used:**
237
-
238
- | Key | Action |
239
- |-----|--------|
240
- | Ctrl+C | Clear editor |
241
- | Ctrl+C twice | Quit |
242
- | Escape | Cancel/abort |
243
- | Escape twice | Open `/tree` |
244
- | Ctrl+L | Open model selector |
245
- | Ctrl+P / Shift+Ctrl+P | Cycle scoped models forward/backward |
246
- | Shift+Tab | Cycle thinking level |
247
- | Ctrl+O | Collapse/expand tool output |
248
- | Ctrl+T | Collapse/expand thinking blocks |
249
-
250
- ### Message Queue
251
-
252
- Submit messages while the agent is working:
253
-
254
- - **Enter** queues a *steering* message, delivered after the current assistant turn finishes executing its tool calls
255
- - **Alt+Enter** queues a *follow-up* message, delivered only after the agent finishes all work
256
- - **Escape** aborts and restores queued messages to editor
257
- - **Alt+Up** retrieves queued messages back to editor
258
-
259
- On Windows Terminal, `Alt+Enter` is fullscreen by default. Remap it in [docs/terminal-setup.md](docs/terminal-setup.md) so omk can receive the follow-up shortcut.
260
-
261
- Configure delivery in [settings](docs/settings.md): `steeringMode` and `followUpMode` can be `"one-at-a-time"` (default, waits for response) or `"all"` (delivers all queued at once). `transport` selects provider transport preference (`"sse"`, `"websocket"`, or `"auto"`) for providers that support multiple transports.
262
-
263
- ---
77
+ | `/session` | Show session path, messages, tokens, and cost |
78
+ | `/tree` | Navigate the current session tree |
79
+ | `/compact [prompt]` | Compact context, optionally with custom instructions |
80
+ | `/copy` | Copy the last assistant message |
81
+ | `/export [file]` | Export the session |
82
+ | `/reload` | Reload project and user resources |
83
+ | `/hotkeys` | Show keyboard shortcuts |
84
+ | `/star` | Open the OMK GitHub repository |
85
+ | `/quit` | Exit OMK |
86
+
87
+ See [Usage](https://github.com/dmae97/omk/blob/main/packages/coding-agent/docs/usage.md) and [Keybindings](https://github.com/dmae97/omk/blob/main/packages/coding-agent/docs/keybindings.md) for the complete interactive reference.
264
88
 
265
89
  ## Sessions
266
90
 
267
- Sessions are stored as JSONL files with a tree structure. Each entry has an `id` and `parentId`, enabling in-place branching without creating new files. See [docs/session-format.md](docs/session-format.md) for file format.
268
-
269
- ### Management
270
-
271
- Sessions auto-save to `~/.omk/agent/sessions/` organized by working directory.
272
-
273
- ```bash
274
- omk -c # Continue most recent session
275
- omk -r # Browse and select from past sessions
276
- omk --no-session # Ephemeral mode (don't save)
277
- omk --name "my task" # Set session display name at startup
278
- omk --session <path|id> # Use specific session file or ID
279
- omk --fork <path|id> # Fork specific session file or ID into a new session
280
- ```
281
-
282
- Use `/session` in interactive mode to see the current session ID before reusing it with `--session <id>` or `--fork <id>`.
283
-
284
- ### Branching
285
-
286
- **`/tree`** - Navigate the session tree in-place. Select any previous point, continue from there, and switch between branches. All history preserved in a single file.
287
-
288
- <p align="center"><img src="https://raw.githubusercontent.com/dmae97/open-multi-agent-kit/v0.90.4/packages/coding-agent/docs/images/tree-view.png" alt="Tree View" width="600"></p>
289
-
290
- - Search by typing, fold/unfold and jump between branches with Ctrl+←/Ctrl+→ or Alt+←/Alt+→, page with ←/→
291
- - Filter modes (Ctrl+O): default → no-tools → user-only → labeled-only → all
292
- - Press Shift+L to label entries as bookmarks and Shift+T to toggle label timestamps
293
-
294
- **`/fork`** - Create a new session file from a previous user message on the active branch. Opens a selector, copies the active path up to that point, and places the selected prompt in the editor for modification.
295
-
296
- **`/clone`** - Duplicate the current active branch into a new session file at the current position. The new session keeps the full active-path history and opens with an empty editor.
297
-
298
- **`--fork <path|id>`** - Fork an existing session file or partial session UUID directly from the CLI. This copies the full source session into a new session file in the current project.
299
-
300
- ### Compaction
301
-
302
- Long sessions can exhaust context windows. Compaction summarizes older messages while keeping recent ones.
303
-
304
- **Manual:** `/compact` or `/compact <custom instructions>`
305
-
306
- **Automatic:** Enabled by default. Triggers on context overflow (recovers and retries) or when approaching the limit (proactive). Configure via `/settings` or `settings.json`.
307
-
308
- Compaction is lossy. The full history remains in the JSONL file; use `/tree` to revisit. Customize compaction behavior via [extensions](#extensions). See [docs/compaction.md](docs/compaction.md) for internals.
309
-
310
- ---
311
-
312
- ## Settings
313
-
314
- Use `/settings` to modify common options, or edit JSON files directly:
315
-
316
- | Location | Scope |
317
- |----------|-------|
318
- | `~/.omk/agent/settings.json` | Global (all projects) |
319
- | `.omk/settings.json` | Project (overrides global) |
91
+ Sessions are append-only JSONL transcripts stored under `~/.omk/agent/sessions/`. OMK can resume, branch, export, and repair sessions without sending the transcript to a separate orchestration service.
320
92
 
321
- See [docs/settings.md](docs/settings.md) for all options.
93
+ Do not publish session files until you have reviewed them for source code, credentials, personal information, and private tool output.
322
94
 
323
- ### Project Trust
95
+ ### Context Compaction
324
96
 
325
- On interactive startup, omk asks before trusting a project folder that contains project-local settings, resources, or project `.agents/skills` and has no saved decision for the folder or a parent folder in `~/.omk/agent/trust.json`. Trusting a project allows omk to load `.omk/settings.json` and `.omk` resources, install missing project packages, and execute project extensions.
97
+ Automatic and manual compaction reduce older context while retaining recent conversation state. Cancellation restores queued user input without committing a partial summary.
326
98
 
327
- Before the trust decision, omk loads only context files, user/global extensions, and CLI `-e` extensions so they can handle the `project_trust` event. Project-local extensions, project package-managed extensions, and project settings are loaded only after the project is trusted. This split also applies when switching to a session from a different cwd whose trust has not been resolved in the current process.
99
+ See [Compaction](https://github.com/dmae97/omk/blob/main/packages/coding-agent/docs/compaction.md).
328
100
 
329
- Non-interactive modes (`-p`, `--mode json`, and `--mode rpc`) do not show a trust prompt. Without an applicable saved trust decision, they use `defaultProjectTrust` from global settings: `ask` (default) and `never` ignore those project resources, while `always` trusts them. Pass `--approve`/`-a` or `--no-approve`/`-na` to override project trust for one run.
101
+ ## Configuration
330
102
 
331
- If no extension or saved decision applies, `defaultProjectTrust` controls the fallback behavior. Set it to `"ask"`, `"always"`, or `"never"` in `~/.omk/agent/settings.json`, or change it with `/settings`.
103
+ ### Context Files
332
104
 
333
- `omk config` and package commands use the same project trust flow, except `omk update` never prompts. Pass `--approve` to trust project-local settings for one command or `--no-approve` to ignore them.
334
-
335
- Use `/trust` in interactive mode to save a project trust decision for future sessions, including trust for the immediate parent folder. It writes `~/.omk/agent/trust.json` only; the current session is not reloaded, so restart omk for changes to take effect.
336
-
337
- ### Telemetry and update checks
338
-
339
- OMK has two separate startup features:
340
-
341
- - **Update check:** fetches `https://omk.dev/api/latest-version` to check whether a newer OMK version exists. Disable it with `OMK_SKIP_VERSION_CHECK=1`. Disabling update checks only turns off this check.
342
- - **Install/update telemetry:** after first install or a changelog-detected update, sends an anonymous version ping to `https://omk.dev/api/report-install`. This setting also controls optional provider attribution headers for OpenRouter, Cloudflare, and direct NVIDIA NIM requests. Opt out by setting `enableInstallTelemetry` to `false` in `settings.json`, or by setting `OMK_TELEMETRY=0`. This does not disable update checks; OMK may still contact `omk.dev` for the latest version unless update checks are disabled or offline mode is enabled.
343
-
344
- Use `--offline` or `OMK_OFFLINE=1` to disable all startup network operations described here, including update checks, package update checks, and install/update telemetry.
345
-
346
- ---
347
-
348
- ## Context Files
349
-
350
- OMK loads `AGENTS.md` (or `CLAUDE.md`) at startup from:
351
- - `~/.omk/agent/AGENTS.md` (global)
352
- - Parent directories (walking up from cwd)
353
- - Current directory
354
-
355
- Override context files are also discovered in the same locations:
356
- - `AGENTS.override.md`
357
- - `AGENTS.GODMODE.md`
358
- - `GODMODE.md`
359
- - `ENI.override.md`
360
-
361
- These override files are loaded in addition to `AGENTS.md`/`CLAUDE.md` and are concatenated with them. Use them for project-specific rules, conventions, and common commands that must coexist with or extend base instructions.
362
-
363
- Disable context file loading with `--no-context-files` (or `-nc`).
364
-
365
- ### System Prompt
366
-
367
- Replace the default system prompt with `.omk/SYSTEM.md` (project) or `~/.omk/agent/SYSTEM.md` (global). Append without replacing via `APPEND_SYSTEM.md`.
368
-
369
- ---
370
-
371
- ## Customization
372
-
373
- ### Prompt Templates
374
-
375
- Reusable prompts as Markdown files. Type `/name` to expand.
376
-
377
- ```markdown
378
- <!-- ~/.omk/agent/prompts/review.md -->
379
- Review this code for bugs, security issues, and performance problems.
380
- Focus on: {{focus}}
381
- ```
105
+ OMK discovers project instructions such as `AGENTS.md` from the current directory and parent directories. Use `--no-context-files` when a clean session must not load project context.
382
106
 
383
- Place in `~/.omk/agent/prompts/`, `.omk/prompts/`, or a [omk package](#omk-packages) to share with others. See [docs/prompt-templates.md](docs/prompt-templates.md).
107
+ ### Prompt Caching
384
108
 
385
- ### Skills
109
+ Provider prompt caching follows provider capabilities and configured retention. Cache metadata is local and bounded; provider-side retention remains subject to the provider's policy.
386
110
 
387
- On-demand capability packages following the [Agent Skills standard](https://agentskills.io). Invoke via `/skill:name`, `!skill:name`, or `!name` when the name is unambiguous, or let the agent load them automatically.
388
-
389
- ```markdown
390
- <!-- ~/.omk/agent/skills/my-skill/SKILL.md -->
391
- # My Skill
392
- Use this skill when the user asks about X.
393
-
394
- ## Steps
395
- 1. Do this
396
- 2. Then that
397
- ```
398
-
399
- Place in `~/.omk/agent/skills/`, `~/.agents/skills/`, `.omk/skills/`, or `.agents/skills/` (from `cwd` up through parent directories) or a [omk package](#omk-packages) to share with others. See [docs/skills.md](docs/skills.md).
400
-
401
- ### Extensions
402
-
403
- <p align="center"><img src="https://raw.githubusercontent.com/dmae97/open-multi-agent-kit/v0.90.4/packages/coding-agent/docs/images/doom-extension.png" alt="Doom Extension" width="600"></p>
404
-
405
- TypeScript modules that extend omk with custom tools, commands, keyboard shortcuts, event handlers, and UI components.
406
-
407
- ```typescript
408
- export default function (omk: ExtensionAPI) {
409
- omk.registerTool({ name: "deploy", ... });
410
- omk.registerCommand("stats", { ... });
411
- omk.on("tool_call", async (event, ctx) => { ... });
412
- }
413
- ```
414
-
415
- The default export can also be `async`. omk waits for async extension factories before startup continues, which is useful for one-time initialization such as fetching remote model lists before calling `omk.registerProvider()`.
416
-
417
- **What's possible:**
418
- - Custom tools (or replace built-in tools entirely)
419
- - Sub-agents and plan mode
420
- - Custom compaction and summarization
421
- - Permission gates and path protection
422
- - Custom editors and UI components
423
- - Status lines, headers, footers
424
- - Git checkpointing and auto-commit
425
- - SSH and sandbox execution
426
- - MCP server integration
427
- - Make omk look like Claude Code
428
- - Games while waiting (yes, Doom runs)
429
- - ...anything you can dream up
430
-
431
- Place in `~/.omk/agent/extensions/`, `.omk/extensions/`, or a [omk package](#omk-packages) to share with others. See [docs/extensions.md](docs/extensions.md) and [examples/extensions/](examples/extensions/).
432
-
433
- ### Themes
434
-
435
- Built-in: `dark`, `light`. Themes hot-reload: modify the active theme file and omk immediately applies changes.
436
-
437
- Place in `~/.omk/agent/themes/`, `.omk/themes/`, or a [omk package](#omk-packages) to share with others. See [docs/themes.md](docs/themes.md).
438
-
439
- ### OMK Packages
440
-
441
- Bundle and share extensions, skills, prompts, and themes via npm or git. Find packages on [npmjs.com](https://www.npmjs.com/search?q=keywords%3Aomk-package) or [Discord](https://discord.com/channels/1456806362351669492/1457744485428629628).
442
-
443
- > **Security:** OMK packages run with full system access. Extensions execute arbitrary code, and skills can instruct the model to perform any action including running executables. Review source code before installing third-party packages.
444
-
445
- ```bash
446
- omk install npm:@foo/omk-tools
447
- omk install npm:@foo/omk-tools@1.2.3 # pinned version
448
- omk install git:github.com/user/repo
449
- omk install git:github.com/user/repo@v1 # tag or commit
450
- omk install git:git@github.com:user/repo
451
- omk install git:git@github.com:user/repo@v1 # tag or commit
452
- omk install https://github.com/user/repo
453
- omk install https://github.com/user/repo@v1 # tag or commit
454
- omk install ssh://git@github.com/user/repo
455
- omk install ssh://git@github.com/user/repo@v1 # tag or commit
456
- omk remove npm:@foo/omk-tools
457
- omk uninstall npm:@foo/omk-tools # alias for remove
458
- omk list
459
- omk update # update omk and packages (skips pinned packages)
460
- omk update --extensions # update packages only
461
- omk update --self # update omk only
462
- omk update --self --force # reinstall omk even if current
463
- omk update npm:@foo/omk-tools # update one package
464
- omk config # enable/disable extensions, skills, prompts, themes
465
- ```
466
-
467
- Packages install to `~/.omk/agent/git/` (git) or `~/.omk/agent/npm/` (npm). Use `-l` for project-local installs (`.omk/git/`, `.omk/npm/`). Git `@ref` values are pinned tags or commits; pinned packages are skipped by `omk update`, so use `omk install git:host/user/repo@new-ref` to move an existing package to a new ref. Git packages install dependencies with `npm install --omit=dev` by default, so runtime deps must be listed under `dependencies`; when `npmCommand` is configured, git packages use plain `install` for compatibility with wrappers. If you use a Node version manager and want package installs to reuse a stable npm context, set `npmCommand` in `settings.json`, for example `["mise", "exec", "node@20", "--", "npm"]`.
468
-
469
- Create a package by adding a `omk` key to `package.json`:
470
-
471
- ```json
472
- {
473
- "name": "my-omk-package",
474
- "keywords": ["omk-package"],
475
- "omk": {
476
- "extensions": ["./extensions"],
477
- "skills": ["./skills"],
478
- "prompts": ["./prompts"],
479
- "themes": ["./themes"]
480
- }
481
- }
482
- ```
483
-
484
- Without a `omk` manifest, omk auto-discovers from conventional directories (`extensions/`, `skills/`, `prompts/`, `themes/`).
485
-
486
- See [docs/packages.md](docs/packages.md).
487
-
488
- ---
489
-
490
- ## Programmatic Usage
491
-
492
- ### SDK
493
-
494
- ```typescript
495
- import { AuthStorage, createAgentSession, ModelRegistry, SessionManager } from "open-multi-agent-kit";
496
-
497
- const authStorage = AuthStorage.create();
498
- const modelRegistry = ModelRegistry.create(authStorage);
499
- const { session } = await createAgentSession({
500
- sessionManager: SessionManager.inMemory(),
501
- authStorage,
502
- modelRegistry,
503
- });
504
-
505
- await session.prompt("What files are in the current directory?");
506
- ```
507
-
508
- For advanced multi-session runtime replacement, use `createAgentSessionRuntime()` and `AgentSessionRuntime`.
509
-
510
- See [docs/sdk.md](docs/sdk.md) and [examples/sdk/](examples/sdk/).
511
-
512
- ### RPC Mode
513
-
514
- For non-Node.js integrations, use RPC mode over stdin/stdout:
515
-
516
- ```bash
517
- omk --mode rpc
518
- ```
519
-
520
- RPC mode uses strict LF-delimited JSONL framing. Clients must split records on `\n` only. Do not use generic line readers like Node `readline`, which also split on Unicode separators inside JSON payloads.
521
-
522
- See [docs/rpc.md](docs/rpc.md) for the protocol.
523
-
524
- ---
525
-
526
- ## Philosophy
527
-
528
- OMK is aggressively extensible so it doesn't have to dictate your workflow. Features that other tools bake in can be built with [extensions](#extensions), [skills](#skills), or installed from third-party [omk packages](#omk-packages). This keeps the core minimal while letting you shape omk to fit how you work.
529
-
530
- **No MCP.** Build CLI tools with READMEs (see [Skills](#skills)), or build an extension that adds MCP support. [Why?](https://mariozechner.at/posts/2025-11-02-what-if-you-dont-need-mcp/)
531
-
532
- **No sub-agents.** There's many ways to do this. Spawn omk instances via tmux, or build your own with [extensions](#extensions), or install a package that does it your way.
533
-
534
- **No permission popups.** Run in a container, or build your own confirmation flow with [extensions](#extensions) inline with your environment and security requirements.
535
-
536
- **No plan mode.** Write plans to files, or build it with [extensions](#extensions), or install a package.
537
-
538
- **No built-in to-dos.** They confuse models. Use a TODO.md file, or build your own with [extensions](#extensions).
539
-
540
- **No background bash.** Use tmux. Full observability, direct interaction.
541
-
542
- The goal is a small core with workflow-specific behavior supplied by user code and packages.
111
+ ### Environment Variables
543
112
 
544
- ---
113
+ Credential, network, model, and runtime variables are documented in [Environment variables](https://github.com/dmae97/omk/blob/main/packages/coding-agent/docs/environment-variables.md). Prefer the documented fixed provider endpoints for authenticated requests.
545
114
 
546
115
  ## CLI Reference
547
116
 
548
117
  ```bash
549
- omk [options] [@files...] [messages...]
118
+ omk [options] [prompt]
119
+ omk --help
550
120
  ```
551
121
 
552
- ### Package Commands
122
+ Common modes:
553
123
 
554
124
  ```bash
555
- omk install <source> [-l] # Install package, -l for project-local
556
- omk remove <source> [-l] # Remove package
557
- omk uninstall <source> [-l] # Alias for remove
558
- omk update [source|self|omk] # Update omk and packages (skips pinned packages)
559
- omk update --extensions # Update packages only
560
- omk update --self # Update omk only
561
- omk update --self --force # Reinstall omk even if current
562
- omk update --extension <src> # Update one package
563
- omk list # List installed packages
564
- omk config # Enable/disable package resources
125
+ omk # interactive TUI
126
+ omk -p "summarize changes" # print mode
127
+ omk --mode json "task" # structured event stream
128
+ omk --rpc # JSONL RPC mode
565
129
  ```
566
130
 
567
- `omk config` and project package commands accept `--approve`/`--no-approve` to trust or ignore project-local settings for one command. `omk update` never prompts for project trust.
568
-
569
- ### Modes
570
-
571
- | Flag | Description |
572
- |------|-------------|
573
- | (default) | Interactive mode |
574
- | `-p`, `--print` | Print response and exit |
575
- | `--mode json` | Output all events as JSON lines (see [docs/json.md](docs/json.md)) |
576
- | `--mode rpc` | RPC mode for process integration (see [docs/rpc.md](docs/rpc.md)) |
577
- | `--export <in> [out]` | Export session to HTML |
578
-
579
- In print mode, omk also reads piped stdin and merges it into the initial prompt:
580
-
581
- ```bash
582
- cat README.md | omk -p "Summarize this text"
583
- ```
584
-
585
- ### Model Options
586
-
587
- | Option | Description |
588
- |--------|-------------|
589
- | `--provider <name>` | Provider (anthropic, openai, google, etc.) |
590
- | `--model <pattern>` | Model pattern or ID (supports `provider/id` and optional `:<thinking>`) |
591
- | `--api-key <key>` | API key (overrides env vars) |
592
- | `--thinking <level>` | `off`, `minimal`, `low`, `medium`, `high`, `xhigh` |
593
- | `--models <patterns>` | Comma-separated patterns for Ctrl+P cycling |
594
- | `--list-models [search]` | List available models |
595
-
596
- ### Session Options
597
-
598
- | Option | Description |
599
- |--------|-------------|
600
- | `-c`, `--continue` | Continue most recent session |
601
- | `-r`, `--resume` | Browse and select session |
602
- | `--session <path\|id>` | Use specific session file or partial UUID |
603
- | `--fork <path\|id>` | Fork specific session file or partial UUID into a new session |
604
- | `--session-dir <dir>` | Custom session storage directory |
605
- | `--no-session` | Ephemeral mode (don't save) |
606
- | `--name <name>`, `-n <name>` | Set session display name at startup |
607
-
608
- ### Tool Options
609
-
610
- | Option | Description |
611
- |--------|-------------|
612
- | `--tools <list>`, `-t <list>` | Allowlist specific tool names across built-in, extension, and custom tools |
613
- | `--exclude-tools <list>`, `-xt <list>` | Disable specific tool names across built-in, extension, and custom tools |
614
- | `--no-builtin-tools`, `-nbt` | Disable built-in tools by default but keep extension/custom tools enabled |
615
- | `--no-tools`, `-nt` | Disable all tools by default |
616
-
617
- Available built-in tools: `read`, `bash`, `edit`, `write`, `grep`, `find`, `ls`
618
-
619
- ### Resource Options
620
-
621
- | Option | Description |
622
- |--------|-------------|
623
- | `-e`, `--extension <source>` | Load extension from path, npm, or git (repeatable) |
624
- | `--no-extensions` | Disable extension discovery |
625
- | `--skill <path>` | Load skill (repeatable) |
626
- | `--no-skills` | Disable skill discovery |
627
- | `--prompt-template <path>` | Load prompt template (repeatable) |
628
- | `--no-prompt-templates` | Disable prompt template discovery |
629
- | `--theme <path>` | Load theme (repeatable) |
630
- | `--no-themes` | Disable theme discovery |
631
- | `--no-context-files`, `-nc` | Disable AGENTS.md and CLAUDE.md context file discovery |
632
-
633
- Combine `--no-*` with explicit flags to load exactly what you need, ignoring settings.json (e.g., `--no-extensions -e ./my-ext.ts`).
634
-
635
131
  ### Other Options
636
132
 
637
- | Option | Description |
638
- |--------|-------------|
639
- | `--system-prompt <text>` | Replace default prompt (context files and skills still appended) |
640
- | `--append-system-prompt <text>` | Append to system prompt |
641
- | `--verbose` | Force verbose startup |
642
- | `-a`, `--approve` | Trust project-local files for this run |
643
- | `-na`, `--no-approve` | Ignore project-local files for this run |
644
- | `-h`, `--help` | Show help |
645
- | `-v`, `--version` | Show version |
646
-
647
- ### File Arguments
648
-
649
- Prefix files with `@` to include in the message:
650
-
651
- ```bash
652
- omk @prompt.md "Answer this"
653
- omk -p @screenshot.png "What's in this image?"
654
- omk @code.ts @test.ts "Review these files"
655
- ```
656
-
657
- ### Examples
133
+ Use `omk --help` for the authoritative option list for the installed version. Avoid copying options from an unreleased branch into automation.
658
134
 
659
- ```bash
660
- # Interactive with initial prompt
661
- omk "List all .ts files in src/"
135
+ ## RPC Mode
662
136
 
663
- # Non-interactive
664
- omk -p "Summarize this codebase"
137
+ RPC mode reads JSON commands from standard input and emits JSON events to standard output. Use it when another process owns the UI or lifecycle.
665
138
 
666
- # Non-interactive with piped stdin
667
- cat README.md | omk -p "Summarize this text"
139
+ See [RPC](https://github.com/dmae97/omk/blob/main/packages/coding-agent/docs/rpc.md).
668
140
 
669
- # Named one-shot session
670
- omk --name "release audit" -p "Audit this repository"
141
+ ## SDK and Extensions
671
142
 
672
- # Different model
673
- omk --provider openai --model gpt-4o "Help me refactor"
143
+ The TypeScript SDK exposes session creation, provider configuration, tools, and event streams. Extensions can add commands, tools, UI components, themes, and resource loaders without patching OMK core.
674
144
 
675
- # Model with provider prefix (no --provider needed)
676
- omk --model openai/gpt-4o "Help me refactor"
145
+ - [SDK](https://github.com/dmae97/omk/blob/main/packages/coding-agent/docs/sdk.md)
146
+ - [Extensions](https://github.com/dmae97/omk/blob/main/packages/coding-agent/docs/extensions.md)
147
+ - [Skills](https://github.com/dmae97/omk/blob/main/packages/coding-agent/docs/skills.md)
148
+ - [Themes](https://github.com/dmae97/omk/blob/main/packages/coding-agent/docs/themes.md)
677
149
 
678
- # Model with thinking level shorthand
679
- omk --model sonnet:high "Solve this complex problem"
150
+ ## Security and Privacy
680
151
 
681
- # Limit model cycling
682
- omk --models "claude-*,gpt-4o"
152
+ - Tool access is explicit and can be scoped by path or runtime policy.
153
+ - Default tests scrub provider credentials; live provider tests require `LIVE_E2E=1`.
154
+ - Subscription quota requests use fixed HTTPS provider origins.
155
+ - Passive quota observers do not block model streams.
156
+ - Logs and issue reports must not contain API keys, OAuth tokens, cookies, or full private transcripts.
683
157
 
684
- # Read-only mode
685
- omk --tools read,grep,find,ls -p "Review the code"
158
+ Report vulnerabilities privately through the repository's GitHub security contact rather than a public issue.
686
159
 
687
- # Disable one extension or built-in tool while keeping the rest available
688
- omk --exclude-tools ask_question
160
+ ## Development
689
161
 
690
- # High thinking level
691
- omk --thinking high "Solve this complex problem"
162
+ ```bash
163
+ git clone https://github.com/dmae97/omk.git
164
+ cd omk
165
+ npm ci --ignore-scripts
166
+ npm run build
167
+ npm run check
168
+ npm test
692
169
  ```
693
170
 
694
- ### Environment Variables
695
-
696
- | Variable | Description |
697
- |----------|-------------|
698
- | `OMK_CODING_AGENT_DIR` | Override config directory (default: `~/.omk/agent`) |
699
- | `OMK_CODING_AGENT_SESSION_DIR` | Override session storage directory (overridden by `--session-dir`) |
700
- | `OMK_PACKAGE_DIR` | Override package directory (useful for Nix/Guix where store paths tokenize poorly) |
701
- | `OMK_OFFLINE` | Disable startup network operations, including update checks, package update checks, and install/update telemetry |
702
- | `OMK_SKIP_VERSION_CHECK` | Skip the OMK version update check at startup. This prevents the `omk.dev` latest-version request |
703
- | `OMK_TELEMETRY` | Override install/update telemetry and provider attribution headers. Use `1`/`true`/`yes` to enable or `0`/`false`/`no` to disable. This does not disable update checks |
704
- | `OMK_CACHE_RETENTION` | Set to `long` for extended prompt cache (Anthropic: 1h, OpenAI: 24h) |
705
- | `VISUAL`, `EDITOR` | External editor for Ctrl+G |
706
-
707
- ---
171
+ See [Development](https://github.com/dmae97/omk/blob/main/packages/coding-agent/docs/development.md) and [CONTRIBUTING.md](https://github.com/dmae97/omk/blob/main/CONTRIBUTING.md).
708
172
 
709
- ## Contributing & Development
710
-
711
- See [CONTRIBUTING.md](https://github.com/dmae97/open-multi-agent-kit/blob/v0.90.4/CONTRIBUTING.md) for guidelines and [docs/development.md](docs/development.md) for setup, forking, and debugging.
712
-
713
- ---
173
+ Release notes: [v0.95.0](https://github.com/dmae97/omk/blob/main/.github/RELEASE_NOTES_v0.95.0.md).
714
174
 
715
175
  ## License
716
176
 
717
- MIT
718
-
719
- ## See Also
720
-
721
- - [omk-ai](https://www.npmjs.com/package/omk-ai): Core LLM toolkit
722
- - [omk-agent-core](https://www.npmjs.com/package/omk-agent-core): Agent framework
723
- - [omk-tui](https://www.npmjs.com/package/omk-tui): Terminal UI components
177
+ MIT. See [LICENSE](https://github.com/dmae97/omk/blob/main/LICENSE).