open-multi-agent-kit 0.78.0

package/templates/spec-kit-omk-preset/commands/plan.md

@@ -0,0 +1,21 @@
+# /speckit-plan
+
+Generate an OMK-optimized implementation plan.
+
+Output: `specs/[###-feature]/plan.md`
+
+This plan must align with the current OMK runtime surface:
+
+- Agent roles: `explorer`, `planner`, `architect`, `coder`, `reviewer`, `qa`; use extra local roles only when `.omk/agents/root.yaml` or `chat-agent-harness.json` exposes them.
+- Skills: reference relevant `.kimi/skills` or `.agents/skills` entrypoints by name; do not paste full skill bodies.
+- MCP: prefer project-scoped `omk-project`; all-scope/global MCP is runtime-only and must not leak secrets.
+- Harness: if the run has `chat-agent-harness.json`, use it for active skills/MCP/hooks, worker limits, gates, and authority boundaries.
+
+This plan includes:
+- Agent routing hints for each phase
+- Expected project structure
+- Quality gate commands
+- Complexity check
+- Evidence and replay hooks (`omk verify --json`, run artifacts, screenshots when relevant)
+
+The plan is consumed by `tasks-template.md` to generate DAG-ready task lists.

package/templates/spec-kit-omk-preset/commands/tasks.md

@@ -0,0 +1,21 @@
+# /speckit-tasks
+
+Generate an OMK-optimized task list.
+
+Output: `specs/[###-feature]/tasks.md`
+
+Each task includes OMK Execution Metadata:
+- `role` — exposed agent role that executes the task
+- `deps` — topological dependencies for DAG scheduling
+- `files` — expected output files for evidence gates
+- `verify` — post-task verification command
+- `gate` — evidence gate type (file-exists, command-pass, diff-nonempty, summary-present)
+- `risk` — checkpoint trigger (high = D-Mail/checkpoint before execution)
+
+Runtime rules:
+- Use only roles exposed by `.omk/agents/root.yaml` or `chat-agent-harness.json`.
+- Keep Kimi as writer/merger/final authority unless the harness explicitly delegates otherwise.
+- Use `chat-agent-harness.json` for active MCP/skills/hooks and worker limits.
+- Do not copy global MCP/skill inventories or secret-like values into tasks.
+
+This metadata improves `tasks.md` → DAG conversion accuracy and evidence-gated completion.

package/templates/spec-kit-omk-preset/preset.yml

@@ -0,0 +1,12 @@
+name: omk
+version: 1.1.0
+description: OMK orchestration preset. Generates spec-kit artifacts optimized for OMK DAG execution, evidence gates, worktree isolation, and multi-agent parallel pipelines.
+author: open-multi-agent-kit
+priority: 100
+templates:
+  spec: templates/spec-template.md
+  plan: templates/plan-template.md
+  tasks: templates/tasks-template.md
+commands:
+  plan: commands/plan.md
+  tasks: commands/tasks.md

package/templates/spec-kit-omk-preset/templates/plan-template.md

@@ -0,0 +1,75 @@
+---
+description: "OMK Implementation Plan template with current agent routing and evidence gates"
+---
+
+# Implementation Plan: [FEATURE]
+
+**Branch**: `[###-feature-name]` | **Date**: [DATE] | **Spec**: [link]
+**OMK Preset**: `omk`
+
+## Summary
+
+[One-paragraph summary of approach]
+
+## Runtime Inventory
+
+- **Harness**: [Path to `.omk/runs/<runId>/chat-agent-harness.json` if present, else "not present"]
+- **MCP Scope**: [project | all | none]
+- **Skills**: [Relevant skill names only]
+- **Authority**: Kimi is final writer/merger unless harness says otherwise.
+
+## Agent Routing
+
+<!--
+  Use roles exposed by .omk/agents/root.yaml or chat-agent-harness.json.
+  Match phase names to tasks.md phases for automatic routing.
+-->
+
+| Phase | Primary Role | Secondary Roles | Evidence Gate |
+|-------|--------------|-----------------|---------------|
+| Bootstrap | explorer | qa | file-exists |
+| Design | architect | planner | file-exists |
+| Core | coder | reviewer | command-pass |
+| QA | qa | reviewer | command-pass |
+| Integration | reviewer | qa | command-pass |
+
+## Project Structure
+
+```text
+src/
+├── models/         # Data models (Phase 3)
+├── services/       # Business logic (Phase 3)
+├── routes/         # API / CLI / UI (Phase 3)
+└── contracts.ts    # Types and interfaces (Phase 2)
+
+tests/
+├── unit/           # Unit tests (Phase 4)
+├── integration/    # Integration tests (Phase 4)
+└── contract/       # Contract tests (Phase 4)
+
+.omk/runs/{runId}/
+├── chat-agent-harness.json  # Active inventory/gates when chat agent mode is used
+├── plan.md                 # This file
+├── explore-summary.md      # Phase 1 output
+├── security-review.md      # Phase 4 output when security review is needed
+└── merge-summary.md        # Phase 5 output
+```
+
+## Complexity Check
+
+| Concern | Decision | Rationale |
+|---------|----------|-----------|
+| New dependencies | [List or "none"] | [Why needed] |
+| Breaking changes | [Yes/No] | [Migration plan if yes] |
+| Parallel tasks | [Count] | [Which phases can run in parallel] |
+| MCP/secret exposure | [None/Scoped] | [How secrets stay out of artifacts] |
+
+## Quality Gates
+
+- **YAML**: `npm run yaml:check` — template/config syntax
+- **Lint**: `npm run lint` — static analysis
+- **TypeCheck**: `npm run check` — TypeScript check
+- **Secrets**: `npm run secret:scan` — no secret leakage
+- **Build**: `npm run build:clean` — clean dist build
+- **Tests**: `npm test` — project test harness
+- **Evidence**: `omk verify --json` or run-specific evidence command when available

package/templates/spec-kit-omk-preset/templates/spec-template.md

@@ -0,0 +1,80 @@
+---
+description: "OMK Feature Specification template with agent-oriented requirements"
+---
+
+# Feature Specification: [FEATURE NAME]
+
+**Feature Branch**: `[###-feature-name]`
+**Created**: [DATE]
+**Status**: Draft
+**Input**: User description: "$ARGUMENTS"
+**OMK Preset**: `omk` (DAG-optimized, parallel-agent ready)
+
+## Agent-Oriented Requirements
+
+<!--
+  Write requirements as tasks an agent can execute.
+  Each requirement should be verifiable by an evidence gate.
+  Use roles exposed by .omk/agents/root.yaml or chat-agent-harness.json.
+-->
+
+### Requirement 1 - [Brief Title] (Priority: P1)
+
+**Agent**: coder / architect
+**Skills**: [Relevant skill names, e.g. omk-typescript-strict, omk-security-review]
+**MCP**: [omk-project or specific configured server, if needed]
+**Evidence Gate**: file-exists + command-pass
+**Risk**: medium
+
+**What**: [Describe what the agent should build]
+**Verify**: [How OMK checks completion — exact command or file path]
+
+**Acceptance**:
+1. File `src/.../xxx.ts` exists and exports `yyy`
+2. Running `npm test -- --match xxx` passes
+3. Running `npm run lint` reports no errors in new files
+
+---
+
+### Requirement 2 - [Brief Title] (Priority: P2)
+
+**Agent**: coder
+**Skills**: [Relevant skill names]
+**Evidence Gate**: command-pass
+**Risk**: low
+
+**What**: [Describe]
+**Verify**: [How OMK checks]
+
+---
+
+## Expected Files
+
+<!--
+  List all files the agent is expected to create or modify.
+  OMK uses these for evidence gates.
+-->
+
+- `src/[module]/[file].ts` — [purpose]
+- `test/[module].test.mjs` — [test coverage]
+- `docs/[feature].md` — [documentation]
+
+## Verification Commands
+
+<!--
+  Keep commands fast, deterministic, and safe for agents.
+-->
+
+- `npm run yaml:check` — YAML/template validation
+- `npm run lint` — static analysis
+- `npm run check` — TypeScript check
+- `npm run secret:scan` — no secret leakage
+- `npm run build:clean` — clean build
+- `npm test` — test harness
+- `omk verify --json` — evidence summary when available
+
+## Assumptions
+
+- [Assumption about environment]
+- [Assumption about existing code]
+- [Assumption about MCP/skills scope]

package/templates/spec-kit-omk-preset/templates/tasks-template.md

@@ -0,0 +1,160 @@
+---
+description: "OMK Task list template with execution metadata for DAG conversion"
+---
+
+# Tasks: [FEATURE NAME]
+
+**Input**: Design documents from `/specs/[###-feature-name]/`
+**Prerequisites**: plan.md (required), spec.md (required)
+**Output**: OMK DAG-ready task list with execution metadata
+
+## Format
+
+Each task MUST include OMK Execution Metadata:
+
+```txt
+- [ ] T001 [P] Description with exact file paths in `backticks`
+  > role: [explorer | planner | architect | coder | reviewer | qa]
+  > deps: [comma-separated task IDs, or 'none']
+  > files: [`expected/file/path.ts`, `test/file.test.mjs`]
+  > verify: `npm test` or `npm run check`
+  > gate: [file-exists | command-pass | diff-nonempty | summary-present]
+  > risk: [low | medium | high]
+```
+
+### Legend
+- **[P]**: Can run in parallel (different files, no dependencies)
+- **role**: OMK agent role exposed by `.omk/agents/root.yaml` or `chat-agent-harness.json`
+- **deps**: Explicit dependencies. OMK scheduler uses these for topological ordering.
+- **files**: Expected output files. OMK evidence gate checks these after execution.
+- **verify**: Command to run after task completion. OMK runs this as a command-pass gate.
+- **gate**: Evidence gate type that proves task completion.
+- **risk**: If high, OMK adds a D-Mail/checkpoint before this task.
+
+---
+
+## Phase 1: Bootstrap
+
+**Purpose**: Project scaffolding and agent context loading
+
+- [ ] T001 [P] Load project context, current harness, and existing codebase structure
+  > role: explorer
+  > deps: none
+  > files: [`.omk/runs/{runId}/explore-summary.md`]
+  > verify: `test -f .omk/runs/{runId}/explore-summary.md`
+  > gate: file-exists
+  > risk: low
+
+- [ ] T002 [P] Verify toolchain scripts and runtime scopes
+  > role: qa
+  > deps: none
+  > files: [`.omk/runs/{runId}/toolchain-summary.md`]
+  > verify: `npm run yaml:check && npm run check`
+  > gate: command-pass
+  > risk: low
+
+---
+
+## Phase 2: Design & Planning
+
+**Purpose**: Architecture decisions before any code change
+
+- [ ] T003 Produce implementation plan with file-level breakdown
+  > role: planner
+  > deps: T001
+  > files: [`.omk/runs/{runId}/plan.md`]
+  > verify: `grep -c "## Phase" .omk/runs/{runId}/plan.md`
+  > gate: file-exists
+  > risk: low
+
+- [ ] T004 Define public API contracts and types
+  > role: architect
+  > deps: T003
+  > files: [`src/contracts.ts`]
+  > verify: `npm run check`
+  > gate: file-exists
+  > risk: medium
+
+---
+
+## Phase 3: Core Implementation
+
+**Purpose**: Build the feature slice by slice
+
+- [ ] T005 [P] Implement domain models and data layer
+  > role: coder
+  > deps: T004
+  > files: [`src/models/*.ts`, `src/repositories/*.ts`]
+  > verify: `npm test -- --match models`
+  > gate: command-pass
+  > risk: medium
+
+- [ ] T006 [P] Implement service / business logic layer
+  > role: coder
+  > deps: T005
+  > files: [`src/services/*.ts`]
+  > verify: `npm test -- --match services`
+  > gate: command-pass
+  > risk: medium
+
+- [ ] T007 Implement API / CLI / UI presentation layer
+  > role: coder
+  > deps: T006
+  > files: [`src/routes/*.ts`, `src/commands/*.ts`]
+  > verify: `npm run lint && npm run check`
+  > gate: command-pass
+  > risk: high
+
+---
+
+## Phase 4: Quality Assurance
+
+**Purpose**: Parallel verification pipeline
+
+- [ ] T008 [P] Write or update regression tests for all new modules
+  > role: coder
+  > deps: T005, T006, T007
+  > files: [`test/**/*.test.mjs`]
+  > verify: `npm test -- --match [feature]`
+  > gate: command-pass
+  > risk: low
+
+- [ ] T009 [P] Run static analysis, type checking, and secret scan
+  > role: qa
+  > deps: T007
+  > files: []
+  > verify: `npm run lint && npm run secret:scan && npm run check`
+  > gate: command-pass
+  > risk: low
+
+- [ ] T010 Review security and trust-boundary impact
+  > role: reviewer
+  > deps: T007
+  > files: [`.omk/runs/{runId}/security-review.md`]
+  > verify: `grep -Ei "risk|vuln|exposure|secret|none" .omk/runs/{runId}/security-review.md`
+  > gate: file-exists
+  > risk: high
+
+---
+
+## Phase 5: Integration & Delivery
+
+**Purpose**: Merge-ready validation
+
+- [ ] T011 Run full quality gate and evidence verification
+  > role: qa
+  > deps: T008, T009, T010
+  > files: []
+  > verify: `npm run build:clean && npm test`
+  > gate: command-pass
+  > risk: medium
+
+- [ ] T012 Produce merge summary and evidence handoff
+  > role: reviewer
+  > deps: T011
+  > files: [`.omk/runs/{runId}/merge-summary.md`]
+  > verify: `grep -c "## Summary" .omk/runs/{runId}/merge-summary.md`
+  > gate: file-exists
+  > risk: low
+
+---

package/templates/web-bridge/chrome-extension/README.md

@@ -0,0 +1,18 @@
+# OMK Web Bridge Chrome Extension
+
+Manifest V3 companion extension for `omk web-bridge`.
+
+## Install
+
+1. Run `omk web-bridge install-host` for native-host instructions.
+2. Open `chrome://extensions`, enable Developer mode, and choose **Load unpacked**.
+3. Select this `templates/web-bridge/chrome-extension` directory.
+4. Copy the extension ID and run `omk web-bridge install-host --extension-id <id> --write`.
+5. Verify with `omk web-bridge doctor --json`.
+
+## Security defaults
+
+- Uses `activeTab`, `scripting`, and `nativeMessaging` only.
+- Does not request cookies, history, bookmarks, downloads, webRequest, or broad host permissions.
+- Does not read cookies, passwords, localStorage, sessionStorage, indexedDB, or request headers.
+- Browser mutations such as clicking, form fill, downloads, uploads, posting, or navigation are out of scope for v1 and require explicit approval in future profiles.

package/templates/web-bridge/chrome-extension/background.js

@@ -0,0 +1,57 @@
+const HOST_NAME = "io.ohmykimi.web_bridge";
+const SCHEMA_VERSION = 1;
+
+chrome.runtime.onMessage.addListener((message, sender, sendResponse) => {
+  if (!message || message.type !== "OMK_WEB_BRIDGE_CAPTURE") return false;
+  captureActiveTab(Boolean(message.includeScreenshot))
+    .then((snapshot) => sendToNativeHost(snapshot))
+    .then((response) => sendResponse({ ok: true, response }))
+    .catch((error) => sendResponse({ ok: false, error: String(error && error.message ? error.message : error) }));
+  return true;
+});
+
+async function captureActiveTab(includeScreenshot) {
+  const [tab] = await chrome.tabs.query({ active: true, currentWindow: true });
+  if (!tab || tab.id === undefined) throw new Error("No active tab is available");
+  await chrome.scripting.executeScript({ target: { tabId: tab.id }, files: ["content-script.js"] });
+  const [page] = await chrome.tabs.sendMessage(tab.id, { type: "OMK_WEB_BRIDGE_COLLECT_PAGE" });
+  let screenshot;
+  if (includeScreenshot) {
+    const bytes = await chrome.tabs.captureVisibleTab(tab.windowId, { format: "png" });
+    screenshot = { mimeType: "image/png", bytes, redacted: false };
+  }
+  return sanitizeSnapshot({
+    tab: { id: tab.id, windowId: tab.windowId, url: tab.url, title: tab.title, active: tab.active },
+    tabs: [{ id: tab.id, windowId: tab.windowId, url: tab.url, title: tab.title, active: true }],
+    ...page,
+    screenshot,
+    metadata: {
+      ...page.metadata,
+      url: tab.url,
+      title: tab.title,
+      capturedAt: new Date().toISOString(),
+      source: "chrome-extension"
+    }
+  });
+}
+
+function sendToNativeHost(snapshot) {
+  return chrome.runtime.sendNativeMessage(HOST_NAME, {
+    schemaVersion: SCHEMA_VERSION,
+    requestId: `extension-${Date.now()}`,
+    method: "orchestration.context.write",
+    params: { snapshot }
+  });
+}
+
+function sanitizeSnapshot(snapshot) {
+  const forbidden = /(cookie|authorization|password|token|secret|localStorage|sessionStorage|indexedDB)/gi;
+  const scrubText = (value) => typeof value === "string" ? value.replace(forbidden, "[redacted]").slice(0, 120000) : value;
+  return JSON.parse(JSON.stringify(snapshot, (key, value) => {
+    forbidden.lastIndex = 0;
+    if (forbidden.test(key)) return "[redacted]";
+    forbidden.lastIndex = 0;
+    if (typeof value === "string") return scrubText(value);
+    return value;
+  }));
+}

package/templates/web-bridge/chrome-extension/content-script.js

@@ -0,0 +1,46 @@
+chrome.runtime.onMessage.addListener((message, _sender, sendResponse) => {
+  if (!message || message.type !== "OMK_WEB_BRIDGE_COLLECT_PAGE") return false;
+  sendResponse(collectPageContext());
+  return false;
+});
+
+function collectPageContext() {
+  const clone = document.documentElement.cloneNode(true);
+  scrubDom(clone);
+  const selectedText = String(window.getSelection ? window.getSelection() : "");
+  const description = document.querySelector('meta[name="description"]')?.getAttribute("content") || undefined;
+  return {
+    metadata: {
+      url: location.href,
+      title: document.title,
+      description,
+      language: document.documentElement.lang || undefined,
+      contentType: document.contentType,
+      source: "chrome-extension"
+    },
+    text: safeText(document.body?.innerText || ""),
+    selectedText: safeText(selectedText),
+    dom: safeText(clone.outerHTML)
+  };
+}
+
+function scrubDom(root) {
+  root.querySelectorAll("script, style, noscript").forEach((node) => node.remove());
+  root.querySelectorAll("input, textarea, select").forEach((node) => {
+    node.removeAttribute("value");
+    node.removeAttribute("checked");
+    node.removeAttribute("selected");
+    if (String(node.getAttribute("type") || "").toLowerCase() === "password") {
+      node.setAttribute("data-omk-redacted", "password");
+    }
+  });
+  root.querySelectorAll("[data-token], [data-secret], [data-password]").forEach((node) => {
+    node.setAttribute("data-omk-redacted", "secret-like-attribute");
+  });
+}
+
+function safeText(value) {
+  return String(value)
+    .replace(/\b(cookie|set-cookie|authorization|password|token|secret|localStorage|sessionStorage|indexedDB)\b\s*[:=]?\s*[^\n\r;]*/gi, "$1: [redacted]")
+    .slice(0, 120000);
+}

package/templates/web-bridge/chrome-extension/manifest.json

@@ -0,0 +1,16 @@
+{
+  "manifest_version": 3,
+  "name": "OMK Web Bridge",
+  "version": "0.1.0",
+  "description": "Local-first read-only page context bridge for open-multi-agent-kit.",
+  "permissions": ["activeTab", "scripting", "nativeMessaging"],
+  "host_permissions": [],
+  "background": {
+    "service_worker": "background.js",
+    "type": "module"
+  },
+  "action": {
+    "default_title": "Send sanitized page context to OMK",
+    "default_popup": "popup.html"
+  }
+}

package/templates/web-bridge/chrome-extension/popup.html

@@ -0,0 +1,21 @@
+<!doctype html>
+<html lang="en">
+<head>
+  <meta charset="utf-8">
+  <title>OMK Web Bridge</title>
+  <style>
+    body { min-width: 260px; font: 13px system-ui, sans-serif; margin: 12px; }
+    button { width: 100%; padding: 8px; }
+    label { display: block; margin: 8px 0; }
+    #status { white-space: pre-wrap; color: #333; }
+  </style>
+</head>
+<body>
+  <strong>OMK Web Bridge</strong>
+  <p>Send sanitized active-tab text, selection, DOM metadata, and optional screenshot to local OMK.</p>
+  <label><input id="screenshot" type="checkbox"> Include screenshot</label>
+  <button id="capture">Send current page</button>
+  <p id="status"></p>
+  <script src="popup.js"></script>
+</body>
+</html>

package/templates/web-bridge/chrome-extension/popup.js

@@ -0,0 +1,9 @@
+document.getElementById("capture").addEventListener("click", async () => {
+  const status = document.getElementById("status");
+  status.textContent = "Capturing active tab...";
+  const includeScreenshot = document.getElementById("screenshot").checked;
+  const response = await chrome.runtime.sendMessage({ type: "OMK_WEB_BRIDGE_CAPTURE", includeScreenshot });
+  status.textContent = response && response.ok
+    ? "Sent sanitized page context to OMK."
+    : `Bridge unavailable: ${response && response.error ? response.error : "unknown error"}`;
+});