### Quick Start Installation

Source: https://github.com/clickhouse/nerve/blob/main/docs/setup.md

This bash script provides the fastest way to get Nerve running by cloning the repository, installing dependencies, building the web UI, and initializing the setup wizard.

```bash
git clone https://github.com/ClickHouse/nerve.git nerve
cd nerve
pip install -e .       # or: uv pip install -e .
cd web && npm install && npm run build && cd ..
nerve init             # Interactive wizard — handles everything
nerve start
```

--------------------------------

### Install Dependencies and Run Dev Server

Source: https://github.com/clickhouse/nerve/blob/main/docs/web-ui.md

Navigate to the web directory, install project dependencies using npm, and start the development server. The dev server proxies API and WebSocket requests to localhost:8900.

```bash
cd web

# Install dependencies
npm install

# Dev server (proxies to backend on :8900)
npm run dev
```

--------------------------------

### Initialize Nerve Setup Wizard

Source: https://context7.com/clickhouse/nerve/llms.txt

Run the first-time setup wizard for Nerve.

```bash
nerve init
```

--------------------------------

### Install Nerve with Curl

Source: https://github.com/clickhouse/nerve/blob/main/README.md

Use this command to download and execute the installation script for Nerve. The script handles dependency installation and launches an interactive setup wizard.

```bash
curl -fsSL https://raw.githubusercontent.com/ClickHouse/nerve/main/install.sh | bash
```

--------------------------------

### Server Installation (Bare Metal)

Source: https://github.com/clickhouse/nerve/blob/main/docs/setup.md

Steps for installing Nerve on a bare metal server, including creating a virtual environment and installing the project and its web UI.

```bash
git clone https://github.com/ClickHouse/nerve.git nerve
cd nerve

# Create virtual environment
uv venv
source .venv/bin/activate

# Install Nerve
uv pip install -e .

# Build web UI
cd web && npm install && npm run build && cd ..

# Run the setup wizard
nerve init
```

--------------------------------

### Manual Nerve Installation Steps

Source: https://github.com/clickhouse/nerve/blob/main/README.md

Follow these steps for a manual installation, which requires Git, Python 3.13+, uv, and Node.js 18+. This includes cloning the repository, setting up a virtual environment, installing dependencies, building the web UI, and initializing Nerve.

```bash
git clone https://github.com/ClickHouse/nerve.git && cd nerve
uv venv --python 3.13 && source .venv/bin/activate
uv pip install -e .
cd web && npm ci && npm run build && cd ..
nerve init
nerve start -f
```

--------------------------------

### Structured TASK.md Example After Onboarding

Source: https://github.com/clickhouse/nerve/blob/main/docs/worker-guide.md

An example of the structured task description in TASK.md after the onboarding process. This format includes Mission, Scope, Triggers, Actions, Approval, and References.

```markdown
# Task

## Mission
Monitor CI pipelines for flaky test failures and propose targeted fixes.

## Scope
- Repository: github.com/org/repo
- CI system: GitHub Actions
- Test framework: pytest

## Triggers
- CI run fails with a test that passed on the previous commit
- Same test fails >2 times in the last 7 days with different commits

## Actions
1. Identify the flaky test from CI logs
2. Check git blame and recent changes to the test and tested code
3. Reproduce locally if possible
4. Propose a fix via plan-approve workflow

## Approval
- Bug fixes: propose plan, wait for approval
- Test-only changes: propose plan, wait for approval
- Infra changes: always escalate via notification

## References
- CI dashboard: https://github.com/org/repo/actions
- Test conventions: docs/testing.md
```

--------------------------------

### Raw TASK.md Example

Source: https://github.com/clickhouse/nerve/blob/main/docs/worker-guide.md

An example of the raw task description in TASK.md before onboarding. This format is a plain English description of the worker's objective.

```markdown
# Task

Monitor the CI pipeline for flaky tests. When a test fails intermittently,
investigate the root cause and propose a fix.
```

--------------------------------

### Clone and Install Nerve

Source: https://github.com/clickhouse/nerve/blob/main/docs/worker-guide.md

Clone the Nerve repository, set up a virtual environment, and install the project dependencies.

```bash
git clone https://github.com/ClickHouse/nerve.git nerve
cd nerve
uv venv && source .venv/bin/activate
uv pip install -e .
```

--------------------------------

### Docker Deployment

Source: https://github.com/clickhouse/nerve/blob/main/docs/setup.md

Instructions for deploying Nerve using Docker, which involves cloning the repository, installing a host dependency for the wizard, and initializing the setup.

```bash
git clone https://github.com/ClickHouse/nerve.git nerve
cd nerve
pip install -e .   # Needed to run the wizard on the host
nerve init         # Choose "docker" at the deployment step
```

--------------------------------

### Installing Frontend Dependencies

Source: https://github.com/clickhouse/nerve/blob/main/nerve/templates/skills/nerve-dev/SKILL.md

Command to install frontend Node.js dependencies, typically run after changes to package.json.

```bash
cd ~/nerve/web

# Install dependencies (only if package.json changed)
npm install
```

--------------------------------

### Worker Mode Initialization and Start

Source: https://context7.com/clickhouse/nerve/llms.txt

Commands to initialize Nerve in worker mode, providing a task description during setup, and then starting the agent. The first boot triggers an onboarding process.

```bash
# Initialize worker mode
nerve init --mode worker
# → Wizard asks for task description
# → Agent onboards itself on first boot

# Example task description during init:
# "Monitor the CI pipeline for flaky tests. When a test fails intermittently,
#  investigate the root cause and propose a fix."

# Start (first boot triggers onboarding)
nerve start -f
```

--------------------------------

### Task File Structure Example

Source: https://github.com/clickhouse/nerve/blob/main/docs/tasks.md

Illustrates the directory structure for active and completed tasks.

```text
workspace/memory/tasks/
├── active/
│   ├── 2026-02-25-fix-auth.md
│   └── 2026-02-26-review-pr.md
└── done/
    └── 2026-02-20-setup-cron.md
```

--------------------------------

### Non-interactive Docker Setup

Source: https://github.com/clickhouse/nerve/blob/main/docs/setup.md

This command initiates a non-interactive Docker setup for Nerve, suitable for CI/CD pipelines or automated deployments, by configuring environment variables and starting the service.

```bash
cp .env.example .env
# Edit .env with your ANTHROPIC_API_KEY (or set NERVE_USE_PROXY=1)
docker compose up
```

--------------------------------

### Install and Configure mkcert for HTTPS

Source: https://github.com/clickhouse/nerve/blob/main/docs/setup.md

This section covers installing mkcert and generating self-signed SSL certificates for local development. The generated certificates should be referenced in your `config.yaml`.

```bash
sudo apt install libnss3-tools
curl -L https://github.com/FiloSottile/mkcert/releases/download/v1.4.4/mkcert-v1.4.4-linux-arm64 -o mkcert
chmod +x mkcert && sudo mv mkcert /usr/local/bin/
```

```bash
mkdir -p ~/.nerve/certs
mkcert -install
mkcert -cert-file ~/.nerve/certs/cert.pem -key-file ~/.nerve/certs/key.pem \
  localhost 127.0.0.1 "$(hostname)" "$(hostname).local"
```

--------------------------------

### Manual Configuration Placeholder

Source: https://github.com/clickhouse/nerve/blob/main/docs/setup.md

This section indicates where manual configuration steps would be detailed, though the wizard handles most setup automatically.

```bash

```

--------------------------------

### Cron Job Configuration Example

Source: https://context7.com/clickhouse/nerve/llms.txt

Example of defining scheduled AI sessions using YAML format for cron job configuration.

```yaml
schedule: "*/5 * * * *"
action:
  tool: "list_sources"
  args:
    consumer: "inbox"

```

--------------------------------

### List Available Sources

Source: https://github.com/clickhouse/nerve/blob/main/docs/sources.md

Example of the `list_sources` agent tool, showing message counts and consumer cursor status for different sources.

```shell
list_sources(consumer="inbox")
→ - github: 45 messages | consumer "inbox": 3 unread
  - gmail:user@example.com: 120 messages | consumer "inbox": 0 unread
  - telegram: 200 messages | consumer "inbox": 5 unread

```

--------------------------------

### Docker Subsequent Starts and Logs

Source: https://github.com/clickhouse/nerve/blob/main/docs/setup.md

Common Docker Compose commands for managing the Nerve container, including starting, stopping, and viewing logs.

```bash
docker compose up        # Start
docker compose up -d     # Start in background
docker compose down      # Stop
docker compose logs -f   # Follow logs
```

--------------------------------

### Get Plan Details

Source: https://context7.com/clickhouse/nerve/llms.txt

Retrieves the detailed content and metadata for a specific implementation plan. Requires the plan ID and an authorization token.

```bash
curl http://localhost:8900/api/plans/plan-abc123 \
  -H "Authorization: Bearer $TOKEN"
```

--------------------------------

### Start Nerve Server in Daemon or Foreground Mode

Source: https://context7.com/clickhouse/nerve/llms.txt

Start the Nerve server. Use '-f' for foreground mode, which is useful for debugging.

```bash
nerve start           # Daemon mode (background)
nerve start -f        # Foreground mode (for debugging)
```

--------------------------------

### Get Source Overview with Stats

Source: https://context7.com/clickhouse/nerve/llms.txt

Fetches an overview of all configured sources along with their message counts and last run times.

```bash
curl http://localhost:8900/api/sources/overview \
  -H "Authorization: Bearer $TOKEN"
```

--------------------------------

### Interactive Tool Handler Example

Source: https://github.com/clickhouse/nerve/blob/main/docs/architecture.md

This example demonstrates the interactive tool handler which pauses the agent for tools needing user input. Answers are routed via WebSocket.

```python
def can_use_tool(self, tool_name: str, tool_args: dict) -> bool:
    if tool_name in ("AskUserQuestion", "ExitPlanMode", "EnterPlanMode"):
        # Pause agent for user input
        return True
    # Auto-approve all other tools
    return False
```

--------------------------------

### Start Nerve in Foreground

Source: https://github.com/clickhouse/nerve/blob/main/docs/worker-guide.md

Start the Nerve worker in the foreground. The first boot will trigger the onboarding process if not already completed.

```bash
nerve start -f
```

--------------------------------

### Ask User Tool Example

Source: https://github.com/clickhouse/nerve/blob/main/docs/architecture.md

This tool facilitates questions with predefined options or free-text input, supporting both blocking and asynchronous modes.

```python
async def ask_user(session_id: str, question: str, options: list[str] | None = None, wait: bool = False):
    # Logic to ask the user a question and handle the response
    pass
```

--------------------------------

### Initialize and Start Multiple Nerve Workers

Source: https://github.com/clickhouse/nerve/blob/main/docs/worker-guide.md

Deploy multiple workers by cloning the Nerve repository into separate directories and initializing each with the worker mode. Start each worker independently.

```bash
cd ~/nerve-ci-monitor
nerve init --mode worker
nerve start
```

```bash
cd ~/nerve-pr-reviewer
nerve init --mode worker
nerve start
```

--------------------------------

### Installing/Updating Backend Dependencies

Source: https://github.com/clickhouse/nerve/blob/main/nerve/templates/skills/nerve-dev/SKILL.md

Command to install or update backend Python dependencies after changes to pyproject.toml, using uv.

```bash
cd ~/nerve

# Install/update dependencies after pyproject.toml changes
.venv/bin/uv pip install -e .
```

--------------------------------

### Nerve Agent Configuration Example

Source: https://github.com/clickhouse/nerve/blob/main/docs/sources.md

Example configuration for Nerve agent sources in YAML format. Includes settings for message TTL, consumer cursor TTL, and specific configurations for Telegram, Gmail, GitHub, and GitHub Events.

```yaml
sync:
  message_ttl_days: 7               # How long to keep inbox messages
  consumer_cursor_ttl_days: 2       # Consumer cursors expire after N days of inactivity

  telegram:
    enabled: true
    api_id: 12345678              # From my.telegram.org
    api_hash: "abc123..."         # From my.telegram.org
    schedule: "*/5 * * * *"       # Every 5 minutes
    batch_size: 50                # Max records per fetch
    condense: false
    exclude_chats: []             # Chat IDs to skip
    monitored_folders: []         # Telegram folder names to filter

  gmail:
    enabled: true
    accounts:                     # One source per account, each with own cursor
      - personal@gmail.com
      - work@company.com
    keyring_password: "..."       # gog keyring password (in config.local.yaml)
    schedule: "*/15 * * * *"
    batch_size: 20
    condense: true                # Strip boilerplate + Haiku extraction for long emails

  github:
    enabled: true
    schedule: "*/15 * * * *"
    batch_size: 30
    condense: true                # Haiku extraction for long notifications

  github_events:
    enabled: true
    schedule: "*/15 * * * *"
    repos: []                     # Empty = all repos. Example: ["owner/repo"]
    username: ""                  # Auto-detect from gh auth
    batch_size: 50
    condense: false               # Events are naturally short

```

--------------------------------

### Task ID Generation Example

Source: https://github.com/clickhouse/nerve/blob/main/docs/tasks.md

Shows how a task ID is generated from the date and a slugified title.

```text
2026-02-25-fix-auth-token-expiry
```

--------------------------------

### Get Skill Instructions

Source: https://github.com/clickhouse/nerve/blob/main/nerve/templates/personal/AGENTS.md

Use `skill_get` to load detailed instructions for a specific skill. This provides information on how to use the skill and its parameters.

```python
skill_get(skill_name="web_search")
```

--------------------------------

### Nerve Template Configuration File

Source: https://context7.com/clickhouse/nerve/llms.txt

Example of the main `config.yaml` file for template settings, including workspace, timezone, agent models, gateway, and sync schedules.

```yaml
# config.yaml — Template settings (committed to repo)
workspace: ~/nerve-workspace
timezone: America/New_York

agent:
  model: claude-opus-4-7         # Primary model for conversations
  cron_model: claude-sonnet-4-6  # Cheaper model for cron jobs
  max_turns: 50
  max_concurrent: 4

gateway:
  host: 0.0.0.0
  port: 8900
  ssl:
    cert: ~/.nerve/certs/cert.pem
    key: ~/.nerve/certs/key.pem

telegram:
  enabled: true
  dm_policy: pairing    # "open" or "pairing"

sync:
  telegram:
    enabled: true
    schedule: "*/5 * * * *"
  gmail:
    enabled: true
    schedule: "*/15 * * * *"
  github:
    enabled: true
    schedule: "*/15 * * * *"

memory:
  chat_model: claude-sonnet-4-6
  categories:
    - name: work
      description: "Work projects, PRs, reviews"
    - name: personal
      description: "Personal tasks, finances"
```

--------------------------------

### Markdown Task File Format Example

Source: https://github.com/clickhouse/nerve/blob/main/docs/tasks.md

Defines the structure and metadata for a task markdown file.

```markdown
# Fix Auth Token Expiry

**Source:** https://github.com/...
**Deadline:** 2026-02-28

Context and details here...

## Updates
- 2026-02-25: Created
- 2026-02-26: Started investigation
- 2026-02-27: DONE — Fixed in PR #123
```

--------------------------------

### CLI Command Example

Source: https://github.com/clickhouse/nerve/blob/main/nerve/templates/worker/TOOLS.md

The 'gh pr list' command defaults to returning a maximum of 30 items. Use the --limit flag to retrieve more.

```bash
gh pr list --limit 100
```

--------------------------------

### Diff Computation Example

Source: https://github.com/clickhouse/nerve/blob/main/docs/architecture.md

Computes structured unified diffs using difflib from stored snapshots versus the current file content.

```python
import difflib

def compute_diff(original_content: str, current_content: str, fromfile='a', tofile='b'):
    return list(difflib.unified_diff(
        original_content.splitlines(keepends=True),
        current_content.splitlines(keepends=True),
        fromfile=fromfile,
        tofile=tofile,
    ))
```

--------------------------------

### Nerve Local Configuration File for Secrets

Source: https://context7.com/clickhouse/nerve/llms.txt

Example of the `config.local.yaml` file for secrets and overrides, which should be gitignored. Includes API keys and authentication details.

```yaml
# config.local.yaml — Secrets and overrides (gitignored)
anthropic_api_key: sk-ant-...
openai_api_key: sk-...  # Optional: enables vector embeddings

telegram:
  bot_token: "123456:ABC..."

auth:
  password_hash: "$2b$12$.."
  jwt_secret: "your-secret-here"
```

--------------------------------

### Upgrade Nerve Installation

Source: https://context7.com/clickhouse/nerve/llms.txt

Perform an upgrade of the Nerve installation, including fetching the latest code, installing dependencies, and rebuilding the frontend.

```bash
nerve upgrade         # Git pull, install deps, rebuild frontend
```

--------------------------------

### Building Frontend for Production

Source: https://github.com/clickhouse/nerve/blob/main/nerve/templates/skills/nerve-dev/SKILL.md

Command to build the frontend for production, including TypeScript checks and Vite bundling.

```bash
cd ~/nerve/web

# Build for production (TypeScript check + Vite bundle → dist/)
npm run build
```

--------------------------------

### Create a New Task

Source: https://github.com/clickhouse/nerve/blob/main/nerve/templates/personal/AGENTS.md

Use `task_create` to initiate a new task. Provide a title and description for the task.

```python
task_create(title="Refactor authentication module", description="Improve security and performance.")
```

--------------------------------

### Clone and Set Up Nerve Repository

Source: https://github.com/clickhouse/nerve/blob/main/nerve/templates/skills/nerve-dev/SKILL.md

Steps to clone the Nerve repository, add the upstream remote, and set up Python and frontend environments. Ensure you have Python 3 and npm installed.

```bash
# 1. Fork on GitHub, then clone your fork
git clone git@github.com:<your-username>/nerve.git
cd nerve

# 2. Add upstream remote
git remote add upstream git@github.com:ClickHouse/nerve.git

# 3. Set up Python environment
python3 -m venv .venv
source .venv/bin/activate
pip install -e .

# 4. Set up frontend
cd web && npm install && cd ..

# 5. Verify
.venv/bin/pytest tests/ -v
cd web && npm run build
```

--------------------------------

### Build Production Bundle

Source: https://github.com/clickhouse/nerve/blob/main/docs/web-ui.md

Build the production-ready bundle for the web application. The output is placed in the web/dist/ directory, which is automatically served by the FastAPI backend.

```bash
cd web && npx vite build
```

--------------------------------

### Worker Task Description Example

Source: https://context7.com/clickhouse/nerve/llms.txt

An example Markdown file defining the mission, scope, triggers, and actions for a Nerve agent operating in worker mode. This file is generated after the agent's onboarding.

```markdown
# Task

## Mission
Monitor CI pipelines for flaky test failures and propose targeted fixes.

## Scope
- Repository: github.com/org/repo
- CI system: GitHub Actions
- Test framework: pytest

## Triggers
- CI run fails with a test that passed on the previous commit
- Same test fails >2 times in the last 7 days

## Actions
1. Identify the flaky test from CI logs
2. Check git blame and recent changes
3. Reproduce locally if possible
4. Propose a fix via plan-approve workflow

## Approval
- Bug fixes: propose plan, wait for approval
- Infra changes: always escalate via notification
```

--------------------------------

### Create a New Task

Source: https://context7.com/clickhouse/nerve/llms.txt

Creates a new task with a title, content, deadline, and optional source URL. Requires an authorization token.

```bash
curl -X POST http://localhost:8900/api/tasks \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "title": "Review PR #123",
    "content": "Review the authentication changes in PR #123",
    "deadline": "2026-03-15",
    "source_url": "https://github.com/org/repo/pull/123"
  }'
```

--------------------------------

### Get Plans for Task

Source: https://github.com/clickhouse/nerve/blob/main/docs/plans.md

Retrieves all plans associated with a specific task.

```APIDOC
## GET /api/tasks/:id/plans

### Description
Retrieves a list of all plans associated with a specific task.

### Method
GET

### Endpoint
/api/tasks/:id/plans

### Path Parameters
- **id** (string) - Required - The unique identifier of the task.
```

--------------------------------

### Initialize Theme from Local Storage

Source: https://github.com/clickhouse/nerve/blob/main/web/index.html

This script initializes the theme of the application by checking local storage for a saved theme preference ('light' or 'dark') and applying it to the document's root element. It runs immediately upon loading.

```javascript
Nerve (function() { var p = localStorage.getItem('nerve-theme'); if (p === 'light' || p === 'dark') { document.documentElement.setAttribute('data-theme', p); } })();
```

--------------------------------

### Get Plan Detail

Source: https://github.com/clickhouse/nerve/blob/main/docs/plans.md

Retrieves the details of a specific plan by its ID.

```APIDOC
## GET /api/plans/:id

### Description
Retrieves the detailed information for a specific plan.

### Method
GET

### Endpoint
/api/plans/:id

### Path Parameters
- **id** (string) - Required - The unique identifier of the plan.
```

--------------------------------

### Create a New Source Adapter

Source: https://github.com/clickhouse/nerve/blob/main/docs/sources.md

Implement a new data source by creating an adapter class inheriting from `nerve.sources.base.Source`. The `fetch` method is crucial for retrieving data, and `preprocess` can be used for data cleanup.

```python
from nerve.sources.base import Source
from nerve.sources.models import FetchResult, SourceRecord

class MySource(Source):
    source_name = "mysource"

    def __init__(self, config: dict):
        self._config = config

    async def fetch(self, cursor: str | None, limit: int = 100) -> FetchResult:
        # Fetch records from your service
        # cursor is opaque — you define the semantics (timestamp, ID, page token, etc.)
        records = []
        next_cursor = cursor

        # ... fetch logic here ...
        # Build SourceRecord objects:
        # SourceRecord(id, source, record_type, summary, content, timestamp, metadata)

        return FetchResult(
            records=records,
            next_cursor=next_cursor,
            has_more=False,
        )

    async def preprocess(self, records: list[SourceRecord]) -> list[SourceRecord]:
        # Optional: source-specific content cleanup before storage.
        # Default (inherited from Source): returns records unchanged.
        return records

    async def close(self) -> None:
        # Optional cleanup (disconnect clients, etc.)
        pass
```

--------------------------------

### Notify Tool Example

Source: https://github.com/clickhouse/nerve/blob/main/docs/architecture.md

This tool is used for fire-and-forget notifications, such as status updates, alerts, and reminders.

```python
async def notify(session_id: str, message: str):
    # Logic to send a notification
    pass
```

--------------------------------

### Task Search FTS5 Query Example

Source: https://github.com/clickhouse/nerve/blob/main/docs/tasks.md

Demonstrates the AND semantics used in `task_search` for precise matching.

```text
Note: `task_search` (user-facing) uses strict AND matching for precision. Duplicate detection (internal) uses fuzzy OR matching for recall — these are intentionally different trade-offs.
```

--------------------------------

### Approve Plan and Spawn Implementation Session

Source: https://context7.com/clickhouse/nerve/llms.txt

Approves a specific implementation plan and initiates a new session for its implementation. Requires the plan ID and an authorization token.

```bash
curl -X POST http://localhost:8900/api/plans/plan-abc123/approve \
  -H "Authorization: Bearer $TOKEN"
```

--------------------------------

### Task Planner Workflow

Source: https://github.com/clickhouse/nerve/blob/main/docs/plans.md

Illustrates the workflow of the proactive task planner, from cron job triggering to user review and plan execution.

```text
Cron (every 4h) → persistent "task-planner" job
  → agent browses tasks, checks memory, picks one worth planning
  → explores codebase with Read, Glob, Grep, Bash, etc.
  → calls plan_propose(task_id, content) to store the proposal
  → plan appears in /plans UI for review

User reviews (via /plans UI or chat tools)
  → approve → spawns implementation session (visible in Chat)
  → decline → marks plan declined
  → request revision → sends feedback to same persistent planner session
```

--------------------------------

### Manage Skills with Skill Tools

Source: https://context7.com/clickhouse/nerve/llms.txt

Use these tools to load, read, run, create, and update skills. Ensure the skill name and path are correct for operations.

```python
skill_get(name="deploy-app")
```

```python
skill_read_reference(name="deploy-app", path="deployment-checklist.md")
```

```python
skill_run_script(name="deploy-app", path="pre-deploy-check.sh", args="--env staging")
```

```python
skill_create(
    name="database-migration",
    description="This skill should be used when running database migrations...",
    content="## Steps\n1. Create backup\n2. Run migration script\n3. Verify data..."
)
```

```python
skill_update(
    name="database-migration",
    content="---\nname: database-migration\ndescription: ...\n---\n\n# Updated Instructions..."
)
```

--------------------------------

### OpenClaw Cron Job JSON Format

Source: https://github.com/clickhouse/nerve/blob/main/docs/migration.md

Example of a cron job configuration in the legacy OpenClaw JSON format.

```json
{
  "id": "morning-telegram",
  "cron": "30 11 * * *",
  "payload": "...",
  "model": "claude-sonnet-4-6"
}
```

--------------------------------

### List Configured Sync Sources

Source: https://github.com/clickhouse/nerve/blob/main/nerve/templates/personal/AGENTS.md

Use `list_sources` to check the status and configuration of integrated external services. This helps in understanding which sources are active and healthy.

```python
list_sources()
```

--------------------------------

### Nerve API Health Check Endpoint

Source: https://context7.com/clickhouse/nerve/llms.txt

Example of a health check request to the Nerve API. This endpoint does not require authentication.

```bash
# Health check (no auth required)
curl http://localhost:8900/health
# Response: {"status": "ok", "version": "0.1.0"}
```

--------------------------------

### Get System Diagnostics

Source: https://context7.com/clickhouse/nerve/llms.txt

Retrieves system-level diagnostic information, including memory usage, disk space, and task statuses.

```bash
curl http://localhost:8900/api/diagnostics \
  -H "Authorization: Bearer $TOKEN"
```

--------------------------------

### List Pending Plans

Source: https://context7.com/clickhouse/nerve/llms.txt

Retrieves a list of implementation plans that are currently in a pending state. Requires an authorization token.

```bash
curl "http://localhost:8900/api/plans?status=pending" \
  -H "Authorization: Bearer $TOKEN"
```

--------------------------------

### Get memU Semantic Memory API Response

Source: https://github.com/clickhouse/nerve/blob/main/docs/api.md

Response structure for retrieving memU categories, items, and indexed resources.

```json
Response: {
  "available": true,
  "categories": [{ "id": "...", "name": "preferences", "description": "...", "summary": "..." }],
  "items": [{ "id": "...", "memory_type": "profile", "summary": "User works at Acme Corp", "resource_id": "...", "created_at": "...", "happened_at": "..." }],
  "resources": [{ "id": "...", "url": "/path/to/file.md", "modality": "document", "caption": "...", "created_at": "..." }],
  "category_items": { "category_id": ["item_id_1", "item_id_2"] }
}
```

--------------------------------

### Get Task Details API Response

Source: https://github.com/clickhouse/nerve/blob/main/docs/api.md

Response structure for retrieving task details, including its full markdown content.

```json
Response: { "id": "2026-03-01-fix-bug", "title": "Fix bug", "status": "pending", "content": "# Fix bug\n\n...", ... }
```

--------------------------------

### Update Task API Request

Source: https://github.com/clickhouse/nerve/blob/main/docs/api.md

Examples of request payloads for updating a task. Fields like status, note, and content are optional.

```json
Request: { "status": "done", "note": "Fixed in PR #123" }
```

```json
Request: { "content": "# Updated Title\n\n**Deadline:** 2026-03-15\n\nNew details..." }
```

--------------------------------

### Propose an Implementation Plan

Source: https://github.com/clickhouse/nerve/blob/main/nerve/templates/personal/AGENTS.md

Use `plan_propose` to submit an implementation plan for asynchronous approval. This is used for outlining steps for autonomous work or background tasks.

```python
plan_propose(plan_description="Automate daily report generation.", steps=["Fetch data", "Generate report", "Send notification"])
```

--------------------------------

### Agent Task Tools Overview

Source: https://github.com/clickhouse/nerve/blob/main/docs/tasks.md

Lists the available agent tools for managing tasks, including their purpose.

```text
| Tool |
|------|-------------|
| `task_create` | Create task with duplicate detection (writes .md + inserts SQLite row) |
| `task_search` | Full-text search on title + content (FTS5) |
| `task_list` | List tasks with status filter (queries SQLite) |
| `task_update` | Update status/deadline/notes (updates both) |
| `task_read` | Read full task content |
| `task_done` | Mark complete, move to `done/` |
```

--------------------------------

### Get Source Run History

Source: https://context7.com/clickhouse/nerve/llms.txt

Retrieves the history of synchronization runs for all sources, including timestamps, records fetched, and any errors encountered.

```bash
curl http://localhost:8900/api/sources/runs \
  -H "Authorization: Bearer $TOKEN"
```