Base Skill

Every Expert automatically has access to @perstack/base — a built-in skill that provides file operations, runtime control, and other essential tools.

No configuration needed. It’s always available.

Design Philosophy

Runtime-Coupled Skill

Base Skill is the only skill tightly coupled with @perstack/runtime. The name “base” reflects this fundamental role — Perstack cannot operate without it, and the runtime assumes Base Skill is always present.

This coupling enables MCP-native runtime control. Rather than implementing special control mechanisms outside MCP, runtime operations (task completion, thinking, todos) are exposed as standard MCP tools.

Thinking Tools and Observability

The think and todo tools serve two purposes:

1. Test-time scaling for any model — These tools enable extended reasoning for models that don’t natively support it, allowing step-by-step problem solving regardless of the underlying model.

2. Observable reasoning — For reasoning models that already have test-time scaling, these tools don’t interfere with native capabilities. Instead, they ensure thought processes are recorded in checkpoints rather than hidden internally.

This design guarantees:

• All reasoning steps are visible in execution history
• Debugging and auditing are always possible
• No hidden Chain-of-Thought that cannot be inspected

Binary Data Handling

For binary files (images, PDFs), Base Skill returns only path and mimeType. The runtime then:

1. Reads the file from the returned path
2. Base64 encodes the content
3. Passes the encoded data to the LLM

This separation keeps Base Skill simple while letting the runtime handle LLM-specific formatting.

No External Network Access

Base Skill never accesses external networks. This is intentional:

• Keeps the attack surface minimal
• Forces explicit network dependencies via separate skills
• Makes network access auditable

If your Expert needs network access, define a dedicated skill for it.

Workspace Isolation

All file operations are restricted to the workspace directory (where perstack run was executed).

• Experts cannot read, write, or access files outside the workspace
• Path traversal attempts (e.g., ../) are blocked
• The .perstack directory is hidden from directory listings

Tool Reference

System Execution

healthCheck

Returns Perstack runtime health status and diagnostics.

Parameters: None

Returns:

{
  "status": "ok",
  "workspace": "/path/to/workspace",
  "uptime": "42s",
  "memory": { "heapUsed": "25MB", "heapTotal": "50MB" },
  "pid": 12345
}

Use cases:

• Verify Perstack runtime is running and responsive
• Check workspace configuration
• Monitor runtime uptime and memory usage
• Debug connection issues

exec

Executes system commands within the workspace.

Parameters:

Returns:

{ "output": "command output here" }

Behavior:

• Executes command using Node.js execFile
• Validates cwd is within the workspace
• Merges provided env with process.env
• Returns “Command executed successfully, but produced no output.” if no output captured
• Returns timeout error if command exceeds timeout

Constraints:

• Working directory must be within workspace
• Do not execute long-running foreground commands (e.g., tail -f)
• Be cautious with resource-intensive commands

Security Note: While cwd is validated, the executed command itself can still access files outside the workspace (e.g., cat /etc/passwd). For production deployments, use infrastructure-level isolation (containers, sandboxes) to enforce strict boundaries.

Runtime Control

attemptCompletion

Signals task completion with automatic todo validation.

Parameters: None

Returns (remaining todos):

{
  "remainingTodos": [
    { "id": 0, "title": "Incomplete task 1", "completed": false },
    { "id": 2, "title": "Incomplete task 2", "completed": false }
  ]
}

Returns (all todos complete or no todos):

{}

Behavior:

• Checks the current todo list for incomplete items
• If incomplete todos exist: returns them in remainingTodos and continues the agent loop
• If no incomplete todos (or no todos at all): triggers run result generation and ends the agent loop
• The Expert should complete remaining todos before calling again

Run result:

When todo validation passes, the LLM generates a run result — a summary of the work done. This ensures every completed run has a clear outcome, regardless of whether the Expert was delegated or run directly.

For delegated Experts, the run result is returned to the delegating Expert as the tool call result, maintaining context isolation.

Best Practice:

• Mark all todos as complete before calling attemptCompletion
• Use clearTodo if you want to reset and start fresh
• The tool prevents premature completion by surfacing forgotten tasks

think

Sequential thinking tool for step-by-step problem analysis.

Parameters:

Returns:

{ "nextThoughtNeeded": true, "thoughtHistoryLength": 3 }

Behavior:

• Records each thinking step sequentially
• Maintains thought history across calls within the session
• Returns current thought count and continuation status

Observability: The runtime records all tool calls (including think) into checkpoints, making thought processes visible in execution history.

todo

Task list manager for tracking work items.

Parameters:

Returns:

{
  "todos": [
    { "id": 0, "title": "Task 1", "completed": false },
    { "id": 1, "title": "Task 2", "completed": true }
  ]
}

Behavior:

• Each todo gets a unique incremental ID when created
• Returns the full todo list after every operation
• State persists across calls within the session

Observability: The runtime records all tool calls (including todo) into checkpoints, making task progress visible in execution history.

clearTodo

Resets the todo list to empty state.

Parameters: None

Returns:

{ "todos": [] }

File Operations

readTextFile

Reads text files with optional line range support.

Parameters:

Returns:

{ "path": "file.txt", "content": "file content", "from": 0, "to": 10 }

Behavior:

• Reads as UTF-8 encoded text
• Supports partial reading via line range
• Defaults to reading entire file if range not specified

Constraints:

• File must exist
• Binary files will cause errors or corrupted output

readImageFile

Validates and returns image file metadata for runtime processing.

Parameters:

Returns:

{ "path": "/workspace/image.png", "mimeType": "image/png", "size": 12345 }

Runtime Integration: The runtime reads the file at path, base64 encodes it, and passes it to the LLM as an inline image.

Supported formats: PNG, JPEG, GIF, WebP

Constraints:

• Maximum file size: 15MB
• File must exist and be a supported image format

readPdfFile

Validates and returns PDF file metadata for runtime processing.

Parameters:

Returns:

{ "path": "/workspace/doc.pdf", "mimeType": "application/pdf", "size": 54321 }

Runtime Integration: The runtime reads the file at path, base64 encodes it, and passes it to the LLM as an inline file.

Constraints:

• Maximum file size: 30MB
• File must exist and be a valid PDF

writeTextFile

Creates or overwrites text files.

Parameters:

Returns:

{ "path": "/workspace/file.txt", "text": "written content" }

Behavior:

• Creates parent directories automatically
• Overwrites existing files
• Writes as UTF-8 encoded text
• Pass empty string to clear file contents

Constraints:

• Maximum 10,000 characters per call
• Use appendTextFile for larger files

appendTextFile

Appends content to existing files.

Parameters:

Returns:

{ "path": "/workspace/file.txt", "text": "appended content" }

Behavior:

• Appends to end of file without modifying existing content
• Does not add newline automatically

Constraints:

• File must exist
• Maximum 2,000 characters per call
• Call multiple times for larger content

editTextFile

Performs search-and-replace in text files.

Parameters:

Returns:

{ "path": "/workspace/file.txt", "oldText": "old", "newText": "new" }

Behavior:

• Performs exact string replacement (first occurrence only)
• Normalizes line endings (CRLF → LF) before matching
• File must contain exact match of oldText

Constraints:

• File must exist
• oldText must exist in file
• Maximum 2,000 characters for both oldText and newText

moveFile

Moves or renames files.

Parameters:

Returns:

{ "source": "/workspace/old.txt", "destination": "/workspace/new.txt" }

Behavior:

• Creates destination directory if needed
• Performs atomic move operation

Constraints:

• Source must exist
• Destination must not exist
• Source must be writable

deleteFile

Removes files from the workspace.

Parameters:

Returns:

{ "path": "/workspace/deleted.txt" }

Constraints:

• File must exist
• File must be writable
• Cannot delete directories (use directory-specific tools)

getFileInfo

Retrieves detailed file or directory metadata.

Parameters:

Returns:

{
  "exists": true,
  "path": "file.txt",
  "absolutePath": "/workspace/file.txt",
  "name": "file.txt",
  "directory": "/workspace",
  "extension": ".txt",
  "type": "file",
  "mimeType": "text/plain",
  "size": 1234,
  "sizeFormatted": "1.21 KB",
  "created": "2024-01-01T00:00:00.000Z",
  "modified": "2024-01-02T00:00:00.000Z",
  "accessed": "2024-01-03T00:00:00.000Z",
  "permissions": {
    "readable": true,
    "writable": true,
    "executable": false
  }
}

Behavior:

• Works for both files and directories
• Returns null for extension and mimeType on directories
• Formats size in human-readable format (B, KB, MB, GB, TB)

Directory Operations

listDirectory

Lists directory contents with metadata.

Parameters:

Returns:

{
  "path": "/workspace/src",
  "items": [
    { "name": "index.ts", "path": "index.ts", "type": "file", "size": 256, "modified": "2024-01-01T00:00:00.000Z" },
    { "name": "lib", "path": "lib", "type": "directory", "size": 4096, "modified": "2024-01-01T00:00:00.000Z" }
  ]
}

Behavior:

• Lists only immediate children (non-recursive)
• Sorts entries alphabetically
• Excludes .perstack directory from results

Constraints:

• Path must exist
• Path must be a directory

createDirectory

Creates directories with recursive parent creation.

Parameters:

Returns:

{ "path": "/workspace/new/nested/dir" }

Behavior:

• Creates all parent directories as needed
• Uses mkdir with recursive: true

Constraints:

• Directory must not already exist
• Parent directory must be writable

deleteDirectory

Removes directories from the workspace.

Parameters:

Returns:

{ "path": "/workspace/old-dir" }

Behavior:

• Validates directory existence and permissions
• Removes directory (and contents if recursive: true)

Constraints:

• Directory must exist
• Directory must be writable
• Non-empty directories require recursive: true
• Cannot delete files (use deleteFile instead)

Example: Agent Loop Flow

Here’s how an Expert uses Base Skill tools in a typical agent loop:

┌─────────────────────────────────────────────────────────────────┐
│  User: "Organize files in this directory by type"              │
└─────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│  1. THINK                                                       │
│  ─────────────────────────────────────────────────────────────  │
│  tool: think                                                    │
│  input: { thought: "I need to scan the directory first..." }   │
│  output: { nextThoughtNeeded: false, thoughtHistoryLength: 1 } │
└─────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│  2. LIST DIRECTORY                                              │
│  ─────────────────────────────────────────────────────────────  │
│  tool: listDirectory                                            │
│  input: { path: "." }                                           │
│  output: { items: [                                             │
│    { name: "photo.jpg", type: "file" },                         │
│    { name: "report.pdf", type: "file" },                        │
│    { name: "notes.txt", type: "file" }                          │
│  ]}                                                             │
└─────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│  3. TODO                                                        │
│  ─────────────────────────────────────────────────────────────  │
│  tool: todo                                                     │
│  input: { newTodos: [                                           │
│    "Create images/ directory",                                  │
│    "Create documents/ directory",                               │
│    "Move files to appropriate directories"                      │
│  ]}                                                             │
│  output: { todos: [{ id: 0, title: "...", completed: false }] } │
└─────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│  4. CREATE DIRECTORIES                                          │
│  ─────────────────────────────────────────────────────────────  │
│  tool: createDirectory                                          │
│  input: { path: "images" }                                      │
│  output: { path: "/workspace/images" }                          │
│                                                                 │
│  tool: createDirectory                                          │
│  input: { path: "documents" }                                   │
│  output: { path: "/workspace/documents" }                       │
└─────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│  5. MOVE FILES                                                  │
│  ─────────────────────────────────────────────────────────────  │
│  tool: moveFile                                                 │
│  input: { source: "photo.jpg", destination: "images/photo.jpg" }│
│  output: { source: "...", destination: "..." }                  │
│                                                                 │
│  tool: moveFile                                                 │
│  input: { source: "report.pdf", destination: "documents/..." }  │
│  ...                                                            │
└─────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│  6. MARK TODOS COMPLETE                                         │
│  ─────────────────────────────────────────────────────────────  │
│  tool: todo                                                     │
│  input: { completedTodos: [0, 1, 2] }                           │
│  output: { todos: [{ id: 0, completed: true }, ...] }           │
└─────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│  7. COMPLETE                                                    │
│  ─────────────────────────────────────────────────────────────  │
│  tool: attemptCompletion                                        │
│  input: {}                                                      │
│  output: {} (no remaining todos)                                │
│  → Agent loop ends                                              │
└─────────────────────────────────────────────────────────────────┘

The Expert definition for this workflow:

[experts."file-organizer"]
description = "Organizes files in the workspace by type"
instruction = """
You organize files in the current directory.
1. List all files using listDirectory
2. Create subdirectories by file type (images/, documents/, etc.)
3. Move files to appropriate directories using moveFile
"""

All tools come from Base Skill — no skill definitions needed.