a2acalling 0.6.75 → 0.6.76

package/.claude/worktrees/agent-a3c12538/docs/protocol.md

@@ -0,0 +1,440 @@
+# A2A Protocol v0 Reference
+
+## Overview
+
+A2A enables OpenClaw agents to call each other across instances with scoped permissions and owner notification.
+
+## Token Format
+
+```
+a2a://<hostname>/<token>
+```
+
+Example: `a2a://my-server.example.com/fed_abc123xyz`
+
+Token structure: `fed_<base64url(24 random bytes)>`
+
+## API Endpoints
+
+All endpoints are prefixed with `/api/a2a/`
+
+### GET /status
+
+Check if A2A is enabled.
+
+Response:
+```json
+{
+  "a2a": true,
+  "version": "0.1.0",
+  "capabilities": ["invoke", "multi-turn"],
+  "rate_limits": {
+    "per_minute": 10,
+    "per_hour": 100,
+    "per_day": 1000
+  }
+}
+```
+
+### GET /ping
+
+Health check endpoint.
+
+Response:
+```json
+{
+  "pong": true,
+  "timestamp": "2026-02-11T17:54:00Z"
+}
+```
+
+### POST /invoke
+
+Call the agent.
+
+Headers:
+```
+Authorization: Bearer fed_abc123xyz
+Content-Type: application/json
+x-trace-id: trace_... (optional)
+x-request-id: req_... (optional)
+```
+
+Request body:
+```json
+{
+  "message": "Your question or request",
+  "conversation_id": "optional-for-multi-turn",
+  "caller": {
+    "name": "Alice's Agent",
+    "instance": "alice.example.com",
+    "context": "Why I'm calling"
+  },
+  "timeout_seconds": 60
+}
+```
+
+Success response:
+```json
+{
+  "success": true,
+  "trace_id": "trace_...",
+  "request_id": "req_...",
+  "conversation_id": "conv_123456",
+  "response": "The agent's response text",
+  "can_continue": true,
+  "tokens_remaining": 47
+}
+```
+
+Error responses:
+```json
+{"success": false, "error": "token_expired", "message": "..."}
+{"success": false, "error": "token_revoked", "message": "..."}
+{"success": false, "error": "permission_denied", "message": "..."}
+{"success": false, "error": "rate_limited", "message": "..."}
+{"success": false, "error": "missing_token", "message": "..."}
+{"success": false, "error": "missing_message", "message": "..."}
+```
+
+### POST /end
+
+Explicitly end a conversation and trigger conclusion/summarization.
+
+Headers:
+```
+Authorization: Bearer fed_abc123xyz
+Content-Type: application/json
+```
+
+Request body:
+```json
+{
+  "conversation_id": "conv_123456"
+}
+```
+
+Success response:
+```json
+{
+  "success": true,
+  "trace_id": "trace_...",
+  "request_id": "req_...",
+  "conversation_id": "conv_123456",
+  "status": "concluded",
+  "summary": "Optional summary text"
+}
+```
+
+Error responses:
+```json
+{"success": false, "error": "unauthorized", "message": "..."}
+{"success": false, "error": "missing_conversation_id", "message": "..."}
+{"success": false, "error": "internal_error", "message": "..."}
+```
+
+## Traceability and Log APIs
+
+A2A persists structured runtime logs to `~/.config/openclaw/a2a-logs.db` (or `$A2A_CONFIG_DIR/a2a-logs.db`).
+
+Primary log fields:
+- `trace_id`
+- `conversation_id`
+- `token_id`
+- `error_code`
+- `status_code`
+- `hint`
+
+Dashboard API endpoints:
+- `GET /api/a2a/dashboard/logs`
+- `GET /api/a2a/dashboard/logs/trace/:traceId`
+- `GET /api/a2a/dashboard/logs/stats`
+- `GET /api/a2a/dashboard/debug/call?trace_id=<id>`
+
+Example filters for `/logs`: `trace_id`, `conversation_id`, `token_id`, `error_code`, `status_code`, `component`, `event`, `level`, `search`, `from`, `to`.
+
+## Permission Tiers
+
+| Tier | Default capabilities | Default `allowed_tools` |
+|------|----------------------|--------------------------|
+| `public` | `context-read` | `Read`, `Grep`, `Glob` |
+| `friends` | `context-read`, `calendar.read`, `email.read`, `search` | `Bash(readonly)`, `Read`, `Grep`, `Glob`, `WebSearch`, `WebFetch` |
+| `family` | `context-read`, `calendar`, `email`, `search`, `tools`, `memory` | `Bash`, `Read`, `Grep`, `Glob`, `WebSearch`, `WebFetch` |
+
+## Disclosure Levels
+
+| Level | Behavior |
+|-------|----------|
+| `public` | Agent may share any non-private information |
+| `minimal` | Agent gives direct answers only, no context about owner |
+| `none` | Agent confirms capability only, provides no actual information |
+
+## Token Storage Schema
+
+Stored in `~/.config/openclaw/a2a.json`:
+
+```json
+{
+  "tokens": [
+    {
+      "id": "tok_abc123xyz789",
+      "token_hash": "sha256...",
+      "name": "Alice's agent",
+      "tier": "public",
+      "capabilities": ["context-read"],
+      "allowed_topics": ["chat"],
+      "allowed_goals": [],
+      "allowed_tools": ["Read", "Grep", "Glob"],
+      "tier_settings": {},
+      "disclosure": "minimal",
+      "notify": "all",
+      "max_calls": null,
+      "calls_made": 5,
+      "created_at": "2026-02-11T17:54:00Z",
+      "expires_at": "2026-02-18T17:54:00Z",
+      "last_used": "2026-02-12T10:30:00Z",
+      "revoked": false
+    }
+  ],
+  "contacts": [
+    {
+      "id": "contact_xyz",
+      "name": "Bob's agent",
+      "owner": "Bob",
+      "host": "bob.example.com",
+      "token_hash": "sha256...",
+      "token_enc": "base64...",
+      "server_name": "Bob's server",
+      "notes": "Met via A2A",
+      "tags": ["collaborator"],
+      "fields": { "email": "bob@example.com" },
+      "linked_token_id": "tok_abc123xyz789",
+      "added_at": "2026-02-11T18:00:00Z"
+    }
+  ]
+}
+```
+
+## Rate Limits
+
+Default per-token limits:
+- 10 requests per minute
+- 100 requests per hour
+- 1000 requests per day
+
+Limits reset on natural boundaries (minute, hour, day UTC).
+
+## Security Considerations
+
+1. **Token hashing**: Tokens stored as SHA-256 hashes server-side
+2. **TLS required**: All A2A calls should use HTTPS
+3. **No credential forwarding**: Tokens are never forwarded to other agents
+4. **Audit logging**: All invocations are logged with caller info
+5. **Auto-revocation**: Tokens may auto-revoke after repeated errors
+
+## Multi-turn Conversations
+
+To continue a conversation, include `conversation_id` from the previous response:
+
+```json
+{
+  "message": "Follow-up question",
+  "conversation_id": "conv_123456"
+}
+```
+
+When finished, either:
+- Call `POST /end` with the same `conversation_id` for explicit conclusion, or
+- Let the receiver auto-conclude on idle timeout/max duration (if enabled).
+
+Conversations expire after 1 hour of inactivity.
+
+## Owner Notifications
+
+When `notify: all`:
+```
+🤝 A2A call received
+
+From: Alice's Agent (alice.example.com)
+Token: "Work collab" (expires 2026-02-18)
+
+---
+Alice's Agent: Does Ben have time this week?
+You: Ben is available Thursday 2-4pm.
+---
+
+📊 5 of unlimited calls | Token expires in 6d
+```
+
+Owner can reply to inject into the conversation.
+
+## OpenClaw Integration
+
+### Gateway Proxy (Recommended)
+
+Run A2A Calling as its own server (separate process), and let the OpenClaw gateway proxy to it:
+
+- A2A backend (separate): `a2a server --port 3001`
+- OpenClaw gateway path(s):
+  - `/a2a` proxies to the A2A dashboard UI (`/dashboard`)
+  - `/api/a2a/*` proxies to the A2A backend API
+
+This keeps the A2A server decoupled from OpenClaw's gateway runtime while still allowing the gateway to be the single public entry point.
+
+### Agent Context
+
+When handling an A2A call, inject context:
+```json
+{
+  "a2a": {
+    "active": true,
+    "caller": "Alice's Agent",
+    "tier": "public",
+    "disclosure": "minimal"
+  }
+}
+```
+
+### New Tool: a2a_call
+
+```typescript
+a2a_call({
+  endpoint: string,    // a2a:// URL
+  message: string,     // Message to send
+  conversation_id?: string  // For multi-turn
+}): {
+  success: boolean,
+  response?: string,
+  conversation_id?: string,
+  error?: string
+}
+```
+
+## E2E Testing
+
+### Architecture
+
+E2E tests run in fully isolated environments. Each test gets its own temp directory, config directory (`A2A_CONFIG_DIR`), and ephemeral port. No shared state between tests; no pollution of the host system.
+
+Core components:
+
+| Component | File | Purpose |
+|-----------|------|---------|
+| Environment | `test/e2e/env.js` | Creates isolated temp dir, config dir, port finder, cleanup |
+| Two-server harness | `test/e2e/two-server.js` | Spins up two independent A2A servers for cross-agent tests |
+| CLI runner | `test/e2e/cli-runner.js` | Wraps `bin/cli.js` as child process with structured output |
+| Agent prompt | `docs/prompts/e2e-test-agent.md` | 9-step prompt sequence for Claude subagent validation |
+
+### Test Categories
+
+- **Infrastructure** (`env.test.js`) -- Isolated environments, port allocation, cleanup
+- **CLI integration** (`cli-runner.test.js`) -- Command execution, onboarding, exit codes, timeouts
+- **Cross-agent flow** (`two-server.test.js`) -- Two servers, token isolation, ping/invoke across instances
+- **Error handling** -- Bad tokens, missing auth, revoked tokens, malformed requests
+- **Summary validation** -- Prompt correctness across `server.js`, `openclaw-integration.js`, `claude-subagent.js` paths
+
+### Entry Points
+
+```bash
+# Run all tests (unit + integration + E2E)
+node test/run.js
+
+# Run E2E tests only
+node test/run.js --e2e
+
+# Standalone orchestrator with verbose or JSON output
+node test/e2e/orchestrate.js [--verbose] [--json]
+
+# Default: excludes E2E for faster feedback
+node test/run.js
+```
+
+The `--e2e` flag filters to files under `test/e2e/`. The standalone orchestrator runs the full 9-step sequence from `docs/prompts/e2e-test-agent.md` and outputs structured JSON results.
+
+### File Structure
+
+```
+test/e2e/
+  env.js                  # createE2EEnv() — isolated temp + config + port
+  env.test.js             # Tests for env isolation and port allocation
+  cli-runner.js           # CLIRunner class — child-process CLI wrapper
+  cli-runner.test.js      # Tests for CLI execution, onboarding, timeouts
+  two-server.js           # TwoServerHarness — two independent A2A instances
+  two-server.test.js      # Tests for cross-agent token isolation and ping
+  orchestrate.js           # Standalone E2E orchestrator (planned)
+```
+
+### Adding New E2E Tests
+
+1. Create `test/e2e/<name>.test.js` exporting `function(test, assert, helpers)`.
+2. Use `createE2EEnv()` for isolation. Always call `env.cleanup()` in a finally block.
+3. Use `TwoServerHarness` if you need two agents. Call `harness.teardown()` after.
+4. Use `CLIRunner` for CLI interactions. It handles `A2A_CONFIG_DIR` automatically.
+5. The test runner discovers all `*.test.js` files recursively -- no registration needed.
+
+Example skeleton:
+
+```js
+module.exports = function (test, assert, helpers) {
+  const { createE2EEnv } = require('./env');
+
+  test('my new E2E scenario', async () => {
+    const env = createE2EEnv('my-test');
+    try {
+      // ... test logic using env.configDir, env.findAvailablePort()
+    } finally {
+      env.cleanup();
+    }
+  });
+};
+```
+
+## CLI Skills (Claude Code & Codex)
+
+A2A ships with slash commands for Claude Code and agent instructions for Codex CLI.
+
+### Installation
+
+```bash
+a2a skills              # Install into current project
+a2a skills --check      # See what would be installed
+a2a skills --force      # Overwrite existing files
+```
+
+Skills are also installed automatically on `npm install -g a2acalling`.
+
+### Claude Code Commands
+
+| Command | Description |
+|---------|-------------|
+| `/a2a-call <contact> <msg>` | Call another agent (multi-turn) |
+| `/a2a-invite [name] [--tier]` | Create invite token |
+| `/a2a-contacts [add\|show\|ping\|rm]` | Manage contacts |
+| `/a2a-status` | Server and agent health dashboard |
+| `/a2a-setup` | First-time setup and onboarding |
+
+Files installed to: `.claude/commands/a2a-*.md`
+
+### Codex CLI
+
+A2A agent instructions are installed to `.codex/AGENTS.md`. Codex reads this file automatically to understand available A2A commands, permission tiers, and workflows.
+
+### Manual Installation
+
+If the automatic install didn't work, copy the files manually:
+
+```bash
+# Claude Code commands
+cp node_modules/a2acalling/.claude/commands/a2a-*.md .claude/commands/
+
+# Codex instructions
+cp node_modules/a2acalling/.codex/AGENTS.md .codex/AGENTS.md
+```
+
+## Future Protocol Extensions (v1+)
+
+- **Capability advertisement**: Agents declare what they can help with
+- **Cryptographic identity**: Ed25519 signatures for caller verification
+- **Streaming responses**: SSE for long-running operations
+- **Webhooks**: Push notifications instead of polling
+- **Payments**: Token-gated access with usage billing

package/.claude/worktrees/agent-a3c12538/docs/signing-setup.md

@@ -0,0 +1,49 @@
+# macOS Signing & Notarization Setup (A2A-94)
+
+This document describes the required Apple credentials for signed, notarized macOS release artifacts.
+
+## Prerequisites
+
+- Apple Developer Program membership (active)
+- Developer ID Application certificate for your team
+- Access to GitHub repository secrets for this project
+
+## 1) Create a Developer ID Application certificate
+
+1. Open Apple Developer portal → Certificates, IDs & Profiles.
+2. Create a new **Developer ID Application** certificate.
+3. Export the certificate + private key from Keychain as `.p12`.
+4. Protect the `.p12` export with a strong password.
+
+## 2) Create GitHub Actions secrets
+
+Upload the following repository secrets:
+
+- `APPLE_CERTIFICATE`: base64-encoded `.p12` archive
+- `APPLE_CERTIFICATE_PASSWORD`: password used for `.p12` export
+- `APPLE_SIGNING_IDENTITY`: exact certificate identity string
+- `APPLE_ID`: Apple account email used for notarization
+- `APPLE_PASSWORD`: app-specific password for notarization
+- `APPLE_TEAM_ID`: Apple Developer Team ID
+
+Example identity format:
+
+```text
+Developer ID Application: Your Name (TEAMID)
+```
+
+## 3) Validate the release output
+
+The macOS build workflow verifies all three checks before publishing artifacts:
+
+- `codesign --verify --deep --strict` on the `.app`
+- `spctl --assess --type execute` on the `.app`
+- `xcrun stapler validate` on the `.dmg`
+
+A release is considered valid only when all checks pass.
+
+## Security notes
+
+- Never commit `.p12`, `.cer`, or private key material.
+- Keep certificate exports in a local, encrypted store only.
+- Rotate Apple app-specific passwords if they are exposed.

package/.claude/worktrees/agent-a3c12538/docs/style.css

@@ -0,0 +1,209 @@
+*, *::before, *::after { box-sizing: border-box; margin: 0; padding: 0; }
+
+:root {
+  --bg: #0B0E14;
+  --panel: #151B28;
+  --card-bg: #1E2738;
+  --ink: #e2e8f0;
+  --muted: #94a3b8;
+  --accent: #3B82F6;
+  --accent-hover: #2563EB;
+  --radius: 12px;
+  --max-w: 960px;
+}
+
+html { scroll-behavior: smooth; }
+body {
+  font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, sans-serif;
+  background: var(--bg);
+  color: var(--ink);
+  line-height: 1.6;
+}
+
+/* Nav */
+.nav {
+  position: sticky;
+  top: 0;
+  z-index: 100;
+  background: rgba(11, 14, 20, 0.85);
+  backdrop-filter: blur(12px);
+  border-bottom: 1px solid rgba(255,255,255,0.06);
+}
+.nav-inner {
+  max-width: var(--max-w);
+  margin: 0 auto;
+  padding: 0.75rem 1.5rem;
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+}
+.nav-brand {
+  display: flex;
+  align-items: center;
+  gap: 0.5rem;
+  text-decoration: none;
+  color: var(--ink);
+  font-weight: 600;
+  font-size: 1.1rem;
+}
+.nav-links { display: flex; gap: 1.5rem; }
+.nav-links a {
+  color: var(--muted);
+  text-decoration: none;
+  font-size: 0.9rem;
+  transition: color 0.2s;
+}
+.nav-links a:hover { color: var(--ink); }
+
+/* Hero */
+.hero {
+  max-width: var(--max-w);
+  margin: 0 auto;
+  padding: 6rem 1.5rem 4rem;
+  text-align: center;
+}
+.hero h1 {
+  font-size: clamp(2.2rem, 5vw, 3.5rem);
+  font-weight: 700;
+  line-height: 1.15;
+  margin-bottom: 1rem;
+}
+.hero-sub {
+  color: var(--muted);
+  font-size: 1.15rem;
+  max-width: 520px;
+  margin: 0 auto 2rem;
+}
+.hero-actions {
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  gap: 1.5rem;
+  flex-wrap: wrap;
+}
+.btn {
+  display: inline-block;
+  padding: 0.7rem 1.6rem;
+  border-radius: 8px;
+  text-decoration: none;
+  font-weight: 600;
+  font-size: 0.95rem;
+  transition: background 0.2s, transform 0.1s;
+}
+.btn:active { transform: scale(0.98); }
+.btn-primary {
+  background: var(--accent);
+  color: #fff;
+}
+.btn-primary:hover { background: var(--accent-hover); }
+.install-cmd {
+  background: var(--panel);
+  padding: 0.6rem 1rem;
+  border-radius: 8px;
+  font-size: 0.9rem;
+  color: var(--muted);
+  border: 1px solid rgba(255,255,255,0.08);
+}
+
+/* Features */
+.features, .get-started, .cli-section {
+  max-width: var(--max-w);
+  margin: 0 auto;
+  padding: 4rem 1.5rem;
+}
+.features h2, .get-started h2, .cli-section h2 {
+  text-align: center;
+  font-size: 1.8rem;
+  margin-bottom: 2.5rem;
+}
+.feature-grid {
+  display: grid;
+  grid-template-columns: repeat(auto-fit, minmax(200px, 1fr));
+  gap: 1.25rem;
+}
+.feature-card {
+  background: var(--card-bg);
+  padding: 1.5rem;
+  border-radius: var(--radius);
+  border: 1px solid rgba(255,255,255,0.06);
+}
+.feature-icon { font-size: 1.8rem; display: block; margin-bottom: 0.75rem; }
+.feature-card h3 { font-size: 1.05rem; margin-bottom: 0.4rem; }
+.feature-card p { color: var(--muted); font-size: 0.9rem; }
+
+/* Steps */
+.steps { display: flex; flex-direction: column; gap: 1.5rem; max-width: 480px; margin: 0 auto; }
+.step {
+  display: flex;
+  align-items: flex-start;
+  gap: 1rem;
+  background: var(--card-bg);
+  padding: 1.25rem;
+  border-radius: var(--radius);
+  border: 1px solid rgba(255,255,255,0.06);
+}
+.step-num {
+  background: var(--accent);
+  color: #fff;
+  width: 2rem;
+  height: 2rem;
+  border-radius: 50%;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  font-weight: 700;
+  font-size: 0.9rem;
+  flex-shrink: 0;
+}
+.step h3 { font-size: 1rem; margin-bottom: 0.3rem; }
+.step code {
+  display: block;
+  background: var(--panel);
+  padding: 0.4rem 0.7rem;
+  border-radius: 6px;
+  font-size: 0.85rem;
+  color: var(--muted);
+}
+
+/* CLI */
+.cli-grid {
+  display: grid;
+  grid-template-columns: repeat(auto-fit, minmax(250px, 1fr));
+  gap: 1.25rem;
+}
+.cli-group {
+  background: var(--card-bg);
+  padding: 1.25rem;
+  border-radius: var(--radius);
+  border: 1px solid rgba(255,255,255,0.06);
+}
+.cli-group h3 { font-size: 1rem; margin-bottom: 0.75rem; color: var(--accent); }
+.cli-group code {
+  display: block;
+  background: var(--panel);
+  padding: 0.4rem 0.7rem;
+  border-radius: 6px;
+  font-size: 0.82rem;
+  color: var(--muted);
+  margin-bottom: 0.5rem;
+  white-space: nowrap;
+  overflow-x: auto;
+}
+
+/* Footer */
+.footer {
+  text-align: center;
+  padding: 3rem 1.5rem;
+  color: var(--muted);
+  font-size: 0.85rem;
+  border-top: 1px solid rgba(255,255,255,0.06);
+}
+.footer a { color: var(--accent); text-decoration: none; }
+.footer a:hover { text-decoration: underline; }
+
+/* Responsive */
+@media (max-width: 600px) {
+  .hero { padding: 4rem 1rem 2.5rem; }
+  .nav-links { gap: 1rem; }
+  .hero-actions { flex-direction: column; }
+}

package/.claude/worktrees/agent-a3c12538/eslint.config.js

@@ -0,0 +1,16 @@
+const sonarjs = require("eslint-plugin-sonarjs");
+
+module.exports = [
+  {
+    files: ["src/**/*.js"],
+    plugins: {
+      sonarjs,
+    },
+    rules: {
+      "sonarjs/cognitive-complexity": ["warn", 15],
+    },
+  },
+  {
+    ignores: ["node_modules/", "native/"],
+  },
+];