### Quickstart: Install and Screenshot

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/bundled-agents/src/web/skill/SKILL.md

Demonstrates how to install agent-browser globally and take a screenshot of a webpage.

```bash
# Install once
npm i -g agent-browser && agent-browser install

# Take a screenshot of a page
agent-browser open https://example.com
agent-browser screenshot home.png
agent-browser close
```

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

### Quickstart: Search, Click, and Capture

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/bundled-agents/src/web/skill/SKILL.md

Guides through searching on a website, clicking a result, and capturing a screenshot of the result page.

```bash
agent-browser open https://duckduckgo.com
agent-browser snapshot -i                      # find the search box ref
agent-browser fill @e1 "agent-browser cli"
agent-browser press Enter
agent-browser wait --load networkidle
agent-browser snapshot -i                      # refs now reflect results
agent-browser click @e5                        # click a result
agent-browser screenshot result.png
```

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

### Atlas CLI Basic Workspace Operations Example

Source: https://github.com/friday-platform/friday-studio/blob/main/apps/atlas-cli/src/claude.md

Demonstrates initializing a new workspace, navigating into its directory, starting the Atlas daemon, and checking the workspace status.

```bash
# Initialize and start a workspace
atlas workspace init my-project
cd my-project
atlas daemon start

# Check status
atlas workspace status
```

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

### Install and Use agent-browser CLI

Source: https://github.com/friday-platform/friday-studio/blob/main/README.md

Install the agent-browser CLI for web scraping and use it to monitor competitor pricing. This example shows the installation command.

```bash
npm i -g agent-browser && agent-browser install
```

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

### Clone and Install Friday Studio

Source: https://github.com/friday-platform/friday-studio/blob/main/README.md

Clones the Friday Studio repository and runs the setup script to bootstrap development environment tools.

```bash
git clone https://github.com/friday-platform/friday-studio
cd friday-studio
bash scripts/setup-dev-env.sh    # bootstraps deno+go+uv+nats+agent-browser, runs deno install, pre-warms Python
```

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

### Start LiteLLM Proxy with API Keys

Source: https://github.com/friday-platform/friday-studio/blob/main/tools/evals/promptfoo/litellm/README.md

Set necessary API keys and run the start script to launch the LiteLLM proxy. The proxy will listen on http://localhost:4000.

```bash
export ANTHROPIC_API_KEY=sk-ant-...
export GROQ_API_KEY=gsk_...
export LITELLM_MASTER_KEY=sk-friday-evals-dev
export LITELLM_API_KEY="$LITELLM_MASTER_KEY"  # runner uses same bearer the proxy validates

./start.sh
# proxy listening on http://localhost:4000
```

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

### Start MCP Servers and Get Tools

Source: https://github.com/friday-platform/friday-studio/blob/main/tools/agent-playground/README.md

Starts MCP servers in-process and returns their tool definitions. These connections are ephemeral.

```APIDOC
## POST /api/mcp/tools

### Description
Start MCP servers in-process, return tool definitions. Connections are ephemeral.

### Method
POST

### Endpoint
/api/mcp/tools
```

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

### Workspace Configuration Example

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/workspace-api/SKILL.md

This example demonstrates various configuration options for permissions, delegation, artifacts, and memory settings within a workspace.yml file. It includes comments explaining defaults, overrides, and best practices.

```yaml
# ── Workspace-level defaults (top of workspace.yml) ────────────
permissions:
  # Bypass tool/skill allowlist enforcement. When true, NO elicitations
  # fire — the elicitation flow and bypass are mutually exclusive.
  # Trusted contexts only. Job-level setting wins; daemon
  # FRIDAY_DANGEROUSLY_SKIP_PERMISSIONS=1 env var is the floor.
  dangerouslySkipAllowlist: false

delegation:
  # Per-field merge with per-job override. Job wins per-field; unset
  # fields fall through to workspace, then to runtime defaults.
  max_depth: 1                # default 1; child cannot itself delegate
  max_steps_per_call: 40
  max_output_tokens: 20000
  max_input_tokens: 100000
  max_wall_time_ms: 120000
  max_cost_usd: null          # reserved; not enforced until cost-tracking lands

artifacts:
  # Workspace-level grace window after job completion before ephemeral
  # artifacts are swept. Default '24h'. Promotion-by-reference
  # (save_memory_entry text containing the id, display_artifact, or
  # aiSummary.keyDetails[].url) keeps an artifact past the window.
  default_grace: 24h

memory:
  own:
    - name: notes
      type: short_term        # ephemeral session-bound
      strategy: narrative
      # ttl: 7d               # OPTIONAL: explicit TTL overrides the
                                  #   type-default. With ttl set, type
                                  #   becomes advisory.
    - name: memory
      type: long_term         # durable across sessions
      strategy: narrative

```

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

### Atlas CLI Session Management Example

Source: https://github.com/friday-platform/friday-studio/blob/main/apps/atlas-cli/src/claude.md

Provides examples for listing all active sessions, retrieving details for a specific session, and following session logs in real-time.

```bash
# Monitor sessions
atlas ps                           # List all sessions
atlas session get sess_123         # Get specific session
atlas logs sess_123 --follow       # Follow session logs
```

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

### Start Svelte library development server

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/ui/README.md

Run 'npm run dev' to start the development server. Use '-- --open' to automatically open the application in a new browser tab.

```sh
npm run dev

# or start the server and open the app in a new browser tab
npm run dev -- --open
```

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

### Snapshot Output Example

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/bundled-agents/src/web/skill/SKILL.md

An example of the structured output format from the `agent-browser snapshot` command, showing element references and types.

```text
Page: Example - Log in
URL: https://example.com/login

@e1 [heading] "Log in"
@e2 [form]
  @e3 [input type="email"] placeholder="Email"
  @e4 [input type="password"] placeholder="Password"
  @e5 [button type="submit"] "Continue"
  @e6 [link] "Forgot password?"
```

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

### Get Provider Details

Source: https://github.com/friday-platform/friday-studio/blob/main/apps/link/README.md

API endpoint to retrieve schema and setup instructions for a specific provider.

```bash
# Get provider details (schema, setup instructions)
GET /v1/providers/:id
```

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

### Example Provider Configuration

Source: https://github.com/friday-platform/friday-studio/blob/main/tools/evals/promptfoo/README.md

An example of a provider entry in `shared/providers.yaml` with a tier tag in its label.

```yaml
- id: openai:chat:friday-sm
  label: 'groq-8b (openai) | tier:small'
```

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

### Basic Profiling Steps

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/bundled-agents/src/web/skill/references/profiling.md

Start profiling, perform actions, and then stop and save the trace to a file.

```bash
# Start profiling
agent-browser profiler start

# Perform actions
agent-browser navigate https://example.com
agent-browser click "#button"
agent-browser wait 1000

# Stop and save
agent-browser profiler stop ./trace.json
```

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

### Start LiteLLM Proxy

Source: https://github.com/friday-platform/friday-studio/blob/main/tools/evals/promptfoo/README.md

Start the LiteLLM proxy once to route requests to real providers. Ensure environment variables are set as per `litellm/README.md`.

```bash
cd tools/evals/promptfoo/litellm && ./start.sh
```

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

### Function Wrapping Suite Setup

Source: https://github.com/friday-platform/friday-studio/blob/main/tools/evals/promptfoo/README.md

Illustrates the setup for a function-wrapping suite where a 'render.ts' file is used to define test cases by spawning a Deno child process.

```typescript
import { spawnSync } from "child_process";

// ... other imports

export default function() {
  const result = spawnSync("deno", ["eval", ...]);
  // ... process result and return TestCase[]
}
```

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

### Start Friday Studio Daemon

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/friday-cli/references/cli.md

Starts the Friday Studio HTTP server. Use --detached to run in the background. Exits with an error if the daemon is already running or the specified port is unavailable.

```bash
atlas daemon start [--port 8080] [--hostname 127.0.0.1] [--detached] [--max-workspaces 10] [--idle-timeout 300] [--log-level info] [--atlas-config <path>]
```

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

### Start Link Service

Source: https://github.com/friday-platform/friday-studio/blob/main/apps/link/CLAUDE.md

Use `deno task start` to run the Link service on port 3100. Other development tasks include `dev` for watch mode, `test` for running tests, and `check` for type checking.

```bash
deno task start
```

```bash
deno task dev
```

```bash
deno task test
```

```bash
deno task test tests/oauth.test.ts
```

```bash
deno task check
```

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

### daemon start

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/friday-cli/references/cli.md

Starts the HTTP server. Can run in the foreground or detached. Exits with an error if the daemon is already running or the port is unavailable.

```APIDOC
## daemon start

### Description
Starts the HTTP server. `--detached` forks and returns; otherwise runs in foreground. Exits 1 if already running or port unavailable. With `FRIDAY_KEY` set, fetches credentials and re-execs with OTEL env.

### Method
`daemon start [--port 8080] [--hostname 127.0.0.1] [--detached] [--max-workspaces 10] [--idle-timeout 300] [--log-level info] [--atlas-config <path>]`

### Aliases
`daemon run`
```

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

### Declare Script Dependencies

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/authoring-skills/references/scripts.md

List all required packages in SKILL.md and provide installation instructions. The execution environment may not have network access to install packages at runtime.

```bash
pip install pypdf
```

```bash
python scripts/extract_pdf.py input.pdf
```

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

### Atlas CLI Agent Management Example

Source: https://github.com/friday-platform/friday-studio/blob/main/apps/atlas-cli/src/claude.md

Illustrates how to discover agents, view their details, and test their connectivity.

```bash
# Discover and test agents
atlas agent list
atlas agent describe k8s-agent
atlas agent test k8s-agent
```

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

### Example SDK Version Pin Drift Detection

Source: https://github.com/friday-platform/friday-studio/blob/main/UPDATING-AGENT-SDK.md

This is an example of the output when a drift in SDK version pins is detected across different files. It highlights which files have differing versions.

```text
✗ SDK version pin drift detected:
  launcher (paths.go): X.Y.Z
  Dockerfile: A.B.C
  installer (prewarm_agent_sdk.rs): X.Y.Z
```

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

### File Output (JSON) Example

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/logger/README.md

Example of a log entry formatted as JSON, suitable for disk storage and machine parsing. Includes timestamp, level, message, process ID, hostname, and context.

```json
{
  "timestamp": "2025-07-31T18:29:50.330Z",
  "level": "info",
  "message": "User authenticated",
  "pid": 1234,
  "hostname": "hostname",
  "context": { "userId": "123", "sessionId": "abc" }
}
```

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

### Good Description Example for Spreadsheet Skill

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/authoring-skills/references/frontmatter.md

This example demonstrates how to describe a skill that analyzes spreadsheet data, specifying file extensions and use cases.

```yaml
description: Analyzes Excel spreadsheets, creates pivot tables, generates charts. Use when analyzing .xlsx files, spreadsheets, or tabular data.
```

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

### Start Agent Playground

Source: https://github.com/friday-platform/friday-studio/blob/main/tools/agent-playground/README.md

Run the Agent Playground from the monorepo root. Access the UI at http://localhost:5200.

```bash
deno task playground     # from monorepo root — starts on http://localhost:5200
```

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

### Example of Debugging Empty Output

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/debugging-broken-jobs/SKILL.md

This example demonstrates the expected interaction flow when a job tool returns an empty output signature. The assistant recognizes the symptom and loads the 'debugging-empty-output' skill, then proceeds to diagnose the agent's behavior based on the session's tool calls and contract adherence.

```text
USER: run stale
ASSISTANT [calls workspace job tool]
JOB TOOL: { success: true, status: completed, summary: "", artifactIds: [] }
ASSISTANT [recognizes empty-output signature]
ASSISTANT [load_skill: debugging-empty-output]
ASSISTANT [calls describe_session(sessionId)]
ASSISTANT [reads agentBlocks[0].toolCalls]
ASSISTANT [observes complete was not called; agent emitted text only]
ASSISTANT: The job ran successfully but the agent emitted prose without
calling `complete`. Per the `@friday/agent-action-handshake`
contract, an `outputTo` action requires the agent to call complete
with the final text. Fix the agent's prompt to instruct it explicitly.
```

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

### Console Log Output Example

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/logger/README.md

Example of how log messages appear in the console, including timestamps, log levels, component names, and context. Colors are automatically applied based on log level and TTY detection.

```text
[18:29:50.330] INFO (atlas): User authenticated {"userId":"123","sessionId":"abc"}
[18:29:50.331] ERROR (atlas): Database connection failed {"error":"timeout","retries":3}
[18:29:50.332] INFO (atlas:worker): Task completed {"workerId":"worker-1","workspaceId":"ws-123"}
```

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

### Run Link Service and Tests

Source: https://github.com/friday-platform/friday-studio/blob/main/apps/link/README.md

Commands to start the Link service and run its tests using Deno tasks.

```bash
# Run the service
deno task start
```

```bash
# Run tests
deno task test
```

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

### Initialize and Start Agent Server

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/core/src/agent-server/README.md

Instantiate and start the AtlasAgentsMCPServer. It requires an agent registry, logger, port, and daemon URL. The server exposes agents as MCP tools.

```typescript
import { AtlasAgentsMCPServer } from "@atlas/agent-server";
import { MyAgentRegistry } from "./my-registry.ts";
import { AtlasLogger } from "@atlas/core";
import { getAtlasDaemonUrl } from "@atlas/oapi-client";

// Create the MCP server. getAtlasDaemonUrl() picks the right scheme/port
// from FRIDAYD_URL + FRIDAY_TLS_CERT so the server works on plain-HTTP and
// TLS-enabled installs without per-env config.
const server = new AtlasAgentsMCPServer({
  agentRegistry: new MyAgentRegistry(),
  logger: AtlasLogger.getInstance(),
  port: 8082,
  daemonUrl: getAtlasDaemonUrl(),
});

// Start the server
await server.start();

// Server exposes agents as MCP tools
// Clients can call agents like: client.callTool("github", { prompt: "scan my repo" })

```

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

### Profiler Commands: Start and Stop

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/bundled-agents/src/web/skill/references/profiling.md

Initiate profiling with default or custom categories, and stop to save the trace data.

```bash
# Start profiling with default categories
agent-browser profiler start

# Start with custom trace categories
agent-browser profiler start --categories "devtools.timeline,v8.execute,blink.user_timing"

# Stop profiling and save to file
agent-browser profiler stop ./trace.json
```

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

### Basic Logging Example

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/logger/README.md

Log messages with different levels and associated context. Ensure you have imported the logger instance.

```typescript
import { logger } from "@atlas/logger";

logger.info("User authenticated", { userId: "123", sessionId: "abc" });
logger.error("Database connection failed", { error: "timeout", retries: 3 });
logger.debug("Processing request", { requestId: "req-456" });
```

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

### GET Request with Path Parameters

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/client/README.md

Example of performing a GET request to the chat storage endpoint, including a path parameter for the stream ID. Use parseResult for error handling.

```typescript
const result = await parseResult(
  client.chatStorage[":streamId"].$get({
    param: { streamId: "abc-123" },
  }),
);
```

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

### Local Development Setup

Source: https://github.com/friday-platform/friday-studio/blob/main/tools/pty-server/README.md

Steps to set up the local development environment, including compiling the atlas binary and ensuring it's in the PATH.

```bash
deno task compile          # produces ./bin/atlas
export PATH="$PWD/bin:$PATH"
deno task playground       # cheatsheet terminal will pick it up
```

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

### Generating Documentation Videos

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/bundled-agents/src/web/skill/references/video-recording.md

Create video documentation of user workflows. This example demonstrates recording a login process with pauses to make the steps clear for viewers.

```bash
#!/bin/bash
# Record workflow for documentation

agent-browser record start ./docs/how-to-login.webm

agent-browser open https://app.example.com/login
agent-browser wait 1000  # Pause for visibility

agent-browser snapshot -i
agent-browser fill @e1 "demo@example.com"
agent-browser wait 500

agent-browser fill @e2 "password"
agent-browser wait 500

agent-browser click @e3
agent-browser wait --load networkidle
agent-browser wait 1000  # Show result

agent-browser record stop
```

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

### Memory Injection Example

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/writing-to-memory/SKILL.md

Memory entries are automatically injected into the system prompt at the start of each turn, formatted as XML. This shows the structure of injected memory.

```xml
<memory workspace="zesty_mushroom" store="preferences">
- Always archive newsletters from substack.com (2026-04-15)
- GitHub CI failures → keep unread
</memory>
```

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

### Multi-tenant Environment Setup

Source: https://github.com/friday-platform/friday-studio/blob/main/ENVIRONMENTS.md

Configure distinct home directories and HTTP/UI ports for multiple Friday instances on the same host. Each instance requires a unique FRIDAY_HOME and specific FRIDAY_PORT_* values in its .env file.

```shell
~/.friday/work/.env        FRIDAY_PORT_FRIDAY=18080  ...
~/.friday/personal/.env    FRIDAY_PORT_FRIDAY=18180  ...
```

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

### Loading Specific Agent Browser Skills

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/bundled-agents/src/web/skill/SKILL.md

Use the `agent-browser skills get` command to load skills tailored for different environments. For example, use `electron` for desktop apps and `slack` for workspace automation.

```bash
agent-browser skills get electron
```

```bash
agent-browser skills get slack
```

```bash
agent-browser skills get dogfood
```

```bash
agent-browser skills get vercel-sandbox
```

```bash
agent-browser skills get agentcore
```

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

### Basic Video Recording Workflow

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/bundled-agents/src/web/skill/references/video-recording.md

Start a video recording, perform automation actions, and then stop the recording to save the video file.

```bash
# Start recording
agent-browser record start ./demo.webm

# Perform actions
agent-browser open https://example.com
agent-browser snapshot -i
agent-browser click @e1
agent-browser fill @e2 "test input"

# Stop and save
agent-browser record stop
```

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

### Defining Structured Output Type with JSON Schema

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/agent-action-handshake/SKILL.md

When outputType is set to a document name, the runtime uses the specified JSON Schema to validate the LLM's output. Ensure the agent's prompt or examples guide the LLM to produce data matching the schema.

```yaml
documentTypes:
  EmailDraft:
    type: object
    properties:
      to: { type: string }
      subject: { type: string }
      body: { type: string }
    required: [to, subject, body]

states:
  drafting:
    entry:
      - type: agent
        agentId: drafter
        outputTo: draft
        outputType: EmailDraft
```

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

### Build and Publish Studio/Installer Script

Source: https://github.com/friday-platform/friday-studio/blob/main/scripts/README.md

A versatile script for building (uploading versioned artifacts) and publishing (updating manifest.json) studio or installer components. Use --watch to block until the run completes and prints the publish command.

```bash
./scripts/studio-release.sh build studio --watch
```

```bash
./scripts/studio-release.sh build installer --watch
```

```bash
./scripts/studio-release.sh publish studio
```

```bash
./scripts/studio-release.sh publish installer
```

```bash
./scripts/studio-release.sh publish studio 1234567890
```

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

### Guard Daemon Status and Start

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/friday-cli/references/cli.md

Checks the status of the atlas daemon and starts it in detached mode if it's not running.

```bash
# Guard daemon
deno task atlas daemon status || deno task atlas daemon start --detached
```

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

### Configure Environment Variables

Source: https://github.com/friday-platform/friday-studio/blob/main/README.md

Copies the environment variable example file and instructs to set necessary API keys, such as for an LLM provider.

```bash
cp .env.example .env
# open .env and set ANTHROPIC_API_KEY (or another provider key)
```

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

### Quick Start with @atlas/client

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/client/README.md

Demonstrates how to make a request using the client and handle potential errors with parseResult. Ensure you have the necessary imports.

```typescript
import { client, parseResult } from "@atlas/client/v2";

// Make a request and handle errors
const result = await parseResult(
  client.chatStorage[":streamId"].$get({ param: { streamId: "123" } }),
);

if (result.ok) {
  console.log(result.value.messages);
} else {
  console.error(result.error);
}
```

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

### Daemon Health Check and Start

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/friday-cli/SKILL.md

Commands to check the status of the Friday daemon and start it if it's not running, including a hard-reload option.

```APIDOC
## Daemon Health Check and Start

### Description
Commands to check the status of the Friday daemon and start it if it's not running, including a hard-reload option.

### Commands

*   **Check Status:**
    ```bash
    deno task atlas daemon status
    ```
*   **Check Status (using resolved URL):**
    ```bash
    curl -k -sf "<resolved-url>/health" && echo OK
    ```
*   **Start Daemon (detached):**
    ```bash
    deno task atlas daemon start --detached
    ```
*   **Restart Daemon (force):**
    ```bash
    deno task atlas daemon restart --force
    ```
```

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

### Start Friday Daemon (Detached)

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/friday-cli/SKILL.md

Starts the Friday daemon in detached mode, allowing it to run in the background. Useful for development environments.

```bash
deno task atlas daemon start --detached
```

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

### Run Chat Replay Development Server

Source: https://github.com/friday-platform/friday-studio/blob/main/tools/chat-replay/README.md

Navigate to the chat-replay directory and start the development server using npm. This command is used to run the application locally for development and testing.

```bash
cd chat-replay
npm run dev
```

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

### Bad Description Examples

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/authoring-skills/references/frontmatter.md

These examples highlight common mistakes in skill descriptions, such as being too vague, using the wrong person, or lacking specific triggers.

```yaml
description: Helps with documents         # vague, no trigger
```

```yaml
description: I can help with PDFs.        # first person
```

```yaml
description: You can use this to process  # second person
```

```yaml
description: Processes data               # no specifics, no trigger
```

```yaml
description: Does stuff with files        # meaningless
```

```yaml
description: A skill for PDF work.        # self-reference, redundant
```

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

### Create, Register, and Execute Agent with Cancellation

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/core/src/agent-server/README.md

Demonstrates creating an agent using the SDK, registering it with the server, and executing it with support for cancellation. Ensure the server instance is available and session data is correctly provided.

```typescript
import { createAgent } from "@atlas/agent-sdk";
import { AtlasAgentsMCPServer } from "@atlas/agent-server";

// Create agent using SDK
const agent = createAgent({
  name: "test-agent",
  handler: async (prompt, context) => {
    // Check for cancellation
    if (context.abortSignal?.aborted) {
      throw new Error("Cancelled");
    }
    return { response: `Processed: ${prompt}` };
  },
});

// Register with server
await server.registerAgent(agent);

// Execute via MCP with cancellation support
const requestId = crypto.randomUUID();
const executePromise = server.executeAgent("test-agent", "Hello", sessionData, {
  requestId,
});

// Cancel the execution
await server.handleCancellation(requestId);

// Execution will throw cancellation error
try {
  await executePromise;
} catch (error) {
  console.log("Execution cancelled");
}
```

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

### daemon restart

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/friday-cli/references/cli.md

Restarts the HTTP server by stopping it and then starting it in detached mode. Supports similar flags to the `start` command and a force option.

```APIDOC
## daemon restart

### Description
Stop → 3s wait → start detached. Same flags as `start`. `--force` bypasses the active-workspace check.

### Method
`daemon restart [--port 8080] [--force] [--max-workspaces <n>] [--idle-timeout <s>]`
```

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

### Full Frontmatter Example with Optional Fields

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/authoring-skills/references/frontmatter.md

Demonstrates a complete SKILL.md frontmatter, including required 'name' and 'description', along with optional fields like 'allowed-tools', 'context', and 'user-invocable'.

```yaml
---
name: creating-workspaces
description: Scaffolds new Friday workspaces. Use when the user asks to create a workspace, add a new workflow, or convert a recurring task into automation.
allowed-tools:
  - create_artifact
  - artifacts_update
  - webfetch
context: workspace-authoring
user-invocable: true
argument-hint: <workspace name>
---
```

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

### Python Agent Example

Source: https://github.com/friday-platform/friday-studio/blob/main/README.md

Example of defining a user-type agent in Python for mechanical tasks, leveraging the 'friday-agent-sdk' for capabilities like LLM, HTTP, and MCP.

```python
# Example usage of friday-agent-sdk within a user-type agent
# from friday_agent_sdk import Agent, LLM, HTTP, MCP

# class MyAgent(Agent):
#     def __init__(self):
#         super().__init__()
#         self.llm = LLM()
#         self.http = HTTP()
#         self.mcp = MCP()

#     def run(self):
#         # Mechanical decision logic here
#         pass

```

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

### Example Workspace Configuration

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/mcp-server/README.md

This YAML configuration enables MCP, defines discoverable capabilities and jobs, and sets rate limits for requests and concurrent sessions.

```yaml
server:
  mcp:
    enabled: true
    discoverable:
      capabilities:
        - "workspace_jobs_*" # Allow job listing and description
      jobs:
        - "build-*" # Allow all build jobs
        - "deploy-prod" # Allow specific deploy job
    rate_limits:
      requests_per_hour: 1000
      concurrent_sessions: 5
```

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

### Skill Registry Configuration

Source: https://github.com/friday-platform/friday-studio/blob/main/ENVIRONMENTS.md

Control remote skill installation and expose the internal kernel workspace. Set FRIDAY_ALLOW_REMOTE_SKILLS to "false" for hardened installs.

```shell
FRIDAY_ALLOW_REMOTE_SKILLS
FRIDAY_EXPOSE_KERNEL
```

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

### Generate Commit Message - Fix Example

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/authoring-skills/references/patterns.md

Example of generating a commit message for a bug fix, following the conventional commit format. Use this for correcting issues.

```markdown
fix(reports): correct date formatting in timezone conversion

Use UTC timestamps consistently across report generation
```

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

### Basic Login Flow with agent-browser

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/bundled-agents/src/web/skill/references/authentication.md

Automate a basic login process by navigating to the login page, filling in credentials, and submitting the form. Use `snapshot -i` to identify elements and `wait --load networkidle` for synchronization.

```bash
# Navigate to login page
agent-browser open https://app.example.com/login
agent-browser wait --load networkidle

# Get form elements
agent-browser snapshot -i
# Output: @e1 [input type="email"], @e2 [input type="password"], @e3 [button] "Sign In"

# Fill credentials
agent-browser fill @e1 "user@example.com"
agent-browser fill @e2 "password123"

# Submit
agent-browser click @e3
agent-browser wait --load networkidle

# Verify login succeeded
agent-browser get url  # Should be dashboard, not login
```

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

### Database Migration Command (Low Freedom)

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/authoring-skills/references/control-calibration.md

Use this low-freedom example for critical operations like database migrations. The command must be run exactly as specified, with no modifications.

```bash
python scripts/migrate.py --verify --backup
```

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

### Enable TLS for Development

Source: https://github.com/friday-platform/friday-studio/blob/main/README.md

Sets up TLS certificates for development and configures environment variables for the daemon URL and CA.

```bash
bash scripts/setup-tls.sh
```

```bash
set -a
. "${FRIDAY_HOME:-$HOME/.friday/local}/.env" 2>/dev/null \
  || . "$HOME/.atlas/.env" 2>/dev/null || true
set +a
```

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

### NDJSON Log Format Example

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/friday-cli/references/session-and-logs.md

This is an example of the NDJSON format used for log entries. Each line is a self-contained JSON object with timestamp, level, message, and context details.

```json
{
  "timestamp": "string"          // ISO 8601
  "level": "trace" | "debug" | "info" | "warn" | "error" | "fatal"
  "message": "string"
  "context": {
    "workspaceId": "?string"
    "sessionId": "?string"
    "agentId": "?string"
    "chatId": "?string"          // synonym for streamId
    "streamId": "?string"
    "error": { "name", "message", "stack"?, "cause"? }
    "[key: string]": "unknown"
  }
  "stack_trace": "?string"
}
```

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

### Generate Commit Message - Feat Example

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/authoring-skills/references/patterns.md

Example of generating a commit message for a new feature, following the conventional commit format. Use this for adding new functionality.

```markdown
feat(auth): implement JWT-based authentication

Add login endpoint and token validation middleware
```

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

### Example Skill Directory Layout

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/authoring-skills/references/scripts.md

Illustrates the recommended directory structure for a skill, separating main instructions, references, and executable scripts.

```directory
my-skill/
├── SKILL.md                   # main instructions (loaded when triggered)
├── references/
│   ├── schema.md              # field-by-field reference (loaded on demand)
│   └── examples.md            # worked examples (loaded on demand)
└── scripts/
    ├── analyze.py             # executed, not loaded into context
    ├── validate.py            # executed
    └── emit.py                # executed
```

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

### Check or Start Daemon

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/friday-cli/references/recipes.md

This snippet checks the status of the Friday daemon and starts it in detached mode if it is not running. It also shows a pure HTTP method to check the daemon's health.

```bash
deno task atlas daemon status || deno task atlas daemon start --detached
# or pure HTTP:
curl -k -sf "$FRIDAYD_URL"/health > /dev/null || echo "daemon down"
```

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

### Bad Handoff Example

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/delegate-handoff/SKILL.md

Illustrates a suboptimal handoff that leads to inefficient tool usage. Avoid vague goals and missing context.

```yaml
goal: "Find the right org for these repos"
handoff: "The user is ken@tempest.team. Try each org."
```

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

### Signal Configuration Schema Example

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/debugging-runtime-errors/SKILL.md

Illustrates the correct structure for signal configuration, particularly the nested 'config' object containing the 'path'. This is crucial for avoiding 'Invalid signal config' errors.

```json
{
  "id": "<signalId>",
  "config": {
    "provider": "http",
    "description": "...",
    "config": { "path": "/your-path" }
  }
}
```

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

### Restart Friday Studio Daemon

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/friday-cli/references/cli.md

Restarts the Friday Studio daemon by stopping it, waiting for 3 seconds, and then starting it in detached mode. Supports the same flags as the 'start' command, with --force bypassing the active-workspace check.

```bash
atlas daemon restart [--port 8080] [--force] [--max-workspaces <n>] [--idle-timeout <s>]
```

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

### Creating a Binary Artifact (User Agent SDK)

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/writing-to-memory/SKILL.md

For binary content, use `create_artifact` with `contentEncoding: "base64"` and specify the `mimeType`.

```javascript
# MCP / user-agent path:
create_artifact(
  data={type:"file", content:"<base64>", contentEncoding:"base64", mimeType:"image/png", originalName:"chart.png"},
  ...
)
```

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

### Incorrect Debugging Approach Example

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/debugging-broken-jobs/SKILL.md

This example illustrates an incorrect debugging approach where a user prematurely assumes a platform bug without following the prescribed diagnostic steps. It highlights the importance of adhering to the 'diagnostic discipline' outlined in the documentation.

```text
USER: run stale
ASSISTANT [calls job tool]
JOB TOOL: { summary: "" }
ASSISTANT: This looks like a platform bug. Let me try migrating from
Pattern A to Pattern B FSM…
```

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

### Register and Use Provider Registry

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/signals/README.md

Demonstrates how to register built-in providers, get a registry instance, and load a provider from configuration. Ensure the registry is initialized before use.

```typescript
import { ProviderRegistry } from "@atlas/signals";

// Register built-in providers
ProviderRegistry.registerBuiltinProviders();

// Get registry instance
const registry = ProviderRegistry.getInstance();

// Create provider from configuration
const config = {
  id: "my-signal",
  provider: "http",
  config: {
    path: "/webhook",
    method: "POST",
  },
};

const provider = await registry.loadFromConfig(config);
await provider.setup();
```

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

### Tab Label Workflow Example

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/bundled-agents/src/web/skill/references/commands.md

Demonstrates using user-assigned labels for managing tabs in multi-tab workflows, including switching, snapshotting, and closing by label.

```bash
agent-browser tab new --label docs https://docs.example.com
agent-browser tab new --label app  https://app.example.com
agent-browser tab docs                   # switch to docs
agent-browser snapshot                   # populate refs for docs
agent-browser click @e1                  # ref click on docs
agent-browser tab app                    # switch to app
agent-browser tab close docs             # close by label
```

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

### Agent Browser Full Skill Reference

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/bundled-agents/src/web/skill/SKILL.md

To access a comprehensive list of all commands, flags, and environment variables, use the `--full` flag with the `core` skill.

```bash
agent-browser skills get core --full
```

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

### Skills API

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/friday-cli/references/http.md

Endpoints for managing skills, including listing, installing, and publishing.

```APIDOC
## Skills API

### Description
Manage skills within the Friday ecosystem, including installation and publishing.

### Endpoints
- `GET /skills` - List available skills.
- `POST /skills/install` - Install a new skill.
- `POST /skills/publish` - Publish a new skill.
```

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

### Get Workspace Configuration

Source: https://github.com/friday-platform/friday-studio/blob/main/packages/system/skills/workspace-api/SKILL.md

Retrieves the full parsed configuration of a specific workspace.

```APIDOC
## Get Workspace Configuration

### Description
Retrieves the full parsed configuration of a specific workspace.

### Method
GET

### Endpoint
/api/workspaces/$WS/config

### Parameters
#### Path Parameters
- **WS** (string) - Required - The ID of the workspace whose configuration to retrieve.

### Request Example
```bash
curl -k -s "$FRIDAYD_URL/api/workspaces/$WS/config" | jq
```

### Response Example
```json
{
  "config": {
    "version": "1.0",
    "workspace": {
      "name": "my-workspace"
    }
  }
}
```
```