agent-godmode 0.1.5

MCP server and library: read/write files, run commands, list files — with strict prompts for user-provided LLMs.

mcp-agent-tools

Workspace-scoped MCP tools for building Cursor-style agents: read_file, write_file, edit_file, run_command, list_files. Includes strict, versioned system prompts (SYSTEM_PROMPT_V1) and OpenAI-style tool definitions so your app can wire any LLM with one import.

The LLM and API keys stay in your app. This package provides tool execution, sandboxing, and prompts—not a hosted model.

Install

pip install mcp-agent-tools

Editable / dev:

pip install -e ".[dev]"

Tools

All tools are scoped to a single workspace root. Paths are relative to that root (or absolute only if they resolve under it). The same operations are available over MCP (mcp-agent-tools server) and in-process via **AgentWorkspace** / **WorkspaceTools**.

Tool	Purpose
**read_file**	Read a UTF-8 text file; optional line range and byte cap.
**write_file**	Create or overwrite/append UTF-8 text; creates parent directories.
**edit_file**	Search-and-replace in an existing UTF-8 file: non-empty old_string, optional replace_all. With replace_all=false, old_string must match exactly once (use surrounding context from read_file for uniqueness). Invalid UTF-8 returns an error instead of corrupting binary data.
**list_files**	List directory entries with optional recursion, glob, depth cap, dotfile control.
**run_command**	Run a subprocess from an **argv list only** (no shell); optional cwd under the root.

For LLM integrations, tool shapes and descriptions are centralized in **OPENAI_TOOL_DEFINITIONS** and **TOOL_DESCRIPTIONS**; agent behavior is guided by **SYSTEM_PROMPT_V1**.

Tier A — Cursor (or any MCP client)

1. Pick a workspace directory (only paths under this root are allowed).

2. Add a server entry (stdio). Example for a global MCP config (paths use forward slashes on Windows):

{
  "mcpServers": {
    "agent-tools": {
      "command": "mcp-agent-tools",
      "args": [],
      "env": {
        "MCP_AGENT_TOOLS_ROOT": "D:/your/project"
      }
    }
  }
}

Or with an explicit CLI root (overrides env for that process):

{
  "mcpServers": {
    "agent-tools": {
      "command": "mcp-agent-tools",
      "args": ["--root", "D:/your/project"]
    }
  }
}

3. Paste **SYSTEM_PROMPT_V1** (from mcp_agent_tools.prompts or below) into your host’s system prompt if the client does not load server instructions automatically.

Environment variables

Variable	Meaning
MCP_AGENT_TOOLS_ROOT	Required unless --root is passed. Absolute workspace root.
MCP_AGENT_TOOLS_MAX_READ_BYTES	Max bytes per read (default 512000).
MCP_AGENT_TOOLS_COMMAND_TIMEOUT	Subprocess timeout in seconds (default 120).
MCP_AGENT_TOOLS_MAX_COMMAND_OUTPUT_BYTES	Truncate stdout/stderr combined (default 256000).
MCP_AGENT_TOOLS_LIST_MAX_ENTRIES	Cap for list_files (default 2000).
MCP_AGENT_TOOLS_ALLOWED_COMMANDS	Comma-separated basenames allowed as argv[0] (e.g. python,uv,node). If unset, all commands allowed under the sandbox.

Tier B — Python app (in-process + OpenAI)

Design notes

• Workspace root — Examples use D:\Avi-assign as a placeholder; point WORK_DIR at any directory you control.
• API key policy — OPENAI_API_KEY is required only for Chat Completions. Imports set client = OpenAI() if HAS_OPENAI_KEY else None; workspace setup and direct edit_file run without a key.
• Model-authored I/O — For write_file, persist only text returned by the model. For edit_file, the model must copy **old_string** exactly from **read_file** (see SYSTEM_PROMPT_V1).

1. Install dependencies

In a shell or a notebook cell:

pip install -q openai
pip install -q -e "D:/MCP"   # editable checkout; or: pip install mcp-agent-tools

In Jupyter, the equivalent is %pip install -q openai followed by %pip install -q -e "D:/MCP" (adjust the path to your clone). Do not place shell comments on the same line as %pip.

2. Imports and API key handling

import os
from pathlib import Path

# OPENAI_API_KEY is required only for steps that call Chat Completions (LLM + agent loops).
# Workspace + direct edit_file work without a key.
# Set via OS env or Jupyter: %env OPENAI_API_KEY sk-...
# Local-only optional override — never commit a real key:
# os.environ["OPENAI_API_KEY"] = "sk-..."

from openai import OpenAI

from mcp_agent_tools import (
    AgentWorkspace,
    OPENAI_TOOL_DEFINITIONS,
    SYSTEM_PROMPT_V1,
    run_agent_loop,
)

HAS_OPENAI_KEY = bool(os.environ.get("OPENAI_API_KEY"))
client = OpenAI() if HAS_OPENAI_KEY else None
MODEL = "gpt-4o-mini"

if not HAS_OPENAI_KEY:
    print(
        "Note: OPENAI_API_KEY not set — Chat Completions cells will raise until you set it. "
        "Workspace + direct edit_file still work."
    )

3. Workspace bootstrap and seed file

# Fixed workspace — all reads/writes/commands stay under this folder
WORK_DIR = Path(r"D:\Avi-assign")
WORK_DIR.mkdir(parents=True, exist_ok=True)
print("Workspace:", WORK_DIR.resolve())

hello = WORK_DIR / "hello.txt"
if not hello.exists():
    hello.write_text("Hello from Avi-assign workspace.\n", encoding="utf-8")

ws = AgentWorkspace(WORK_DIR)
print(ws.read_file("hello.txt"))
print("--- list_files ---")
print(ws.list_files(".", recursive=False))

4. LLM-authored file body (no tool calls)

Requires OPENAI_API_KEY. Skip if you are only exercising tools without the API.

if client is None:
    raise ValueError(
        "Set OPENAI_API_KEY to run this cell (Jupyter: %env OPENAI_API_KEY sk-...). "
        "Skip this cell if you only want workspace / edit_file demos."
    )

# 1) Context from disk (read-only)
context = ws.read_file("hello.txt")

# 2) Ask the model to author the entire new file; no static template for the body
user_prompt = (
    "Here is the current contents of hello.txt in my workspace:\n\n"
    f"---\n{context}\n---\n\n"
    "Write ONLY the body of a new Markdown file (no preamble, no code fences) "
    "with a title line and two bullet points explaining what this greeting is for."
)

resp = client.chat.completions.create(
    model=MODEL,
    messages=[
        {
            "role": "system",
            "content": "You output only the file body the user asked for. No extra commentary.",
        },
        {"role": "user", "content": user_prompt},
    ],
)

generated = (resp.choices[0].message.content or "").strip()
if not generated:
    raise RuntimeError("LLM returned empty content; nothing to write.")

# 3) Persist exactly what the LLM produced
out_rel = "llm_generated_notes.md"
ws.write_file(out_rel, generated, mode="overwrite")
print(f"Wrote {out_rel!r} ({len(generated)} chars from model)\n")
print(ws.read_file(out_rel))

5. Direct edit_file (no Chat Completions)

No API key required. The next lines create ws if you have not run the workspace section yet (same root).

# Direct edit_file (no Chat Completions call).
# If you run this before the main workspace cell, the next few lines create `ws` (same root as that cell).
from pathlib import Path

from mcp_agent_tools import AgentWorkspace

if "ws" not in globals():
    WORK_DIR = Path(r"D:\Avi-assign")
    WORK_DIR.mkdir(parents=True, exist_ok=True)
    ws = AgentWorkspace(WORK_DIR)

demo_edit = "notebook_edit_demo.txt"
ws.write_file(
    demo_edit,
    "version: 1\nstatus: draft\nfooter: end\n",
    mode="overwrite",
)
print("--- before ---")
print(ws.read_file(demo_edit), end="")
print(ws.edit_file(demo_edit, old_string="status: draft", new_string="status: ready"))
print("--- after ---")
print(ws.read_file(demo_edit), end="")

6. Agent loop: model calls write_file

Requires OPENAI_API_KEY.

if client is None:
    raise ValueError(
        "Set OPENAI_API_KEY to run this cell. "
        "Skip if you only need workspace or direct edit_file."
    )


def complete(messages, tools):
    """One Chat Completions turn; return OpenAI-shaped dict for run_agent_loop."""
    resp = client.chat.completions.create(
        model=MODEL,
        messages=messages,
        tools=tools,
        tool_choice="auto",
    )
    return resp.model_dump()


answer = run_agent_loop(
    complete,
    "Use tools only. List the workspace root, read hello.txt, then call write_file on "
    "agent_notes.txt. The `content` argument must be your own freshly written summary "
    "(several sentences) based only on what you read—do not paste boilerplate.",
    ws,
    system_prompt=SYSTEM_PROMPT_V1,
    max_turns=12,
)
print("--- final answer ---")
print(answer)
print("--- agent_notes.txt (if created by tool write_file) ---")
p = WORK_DIR / "agent_notes.txt"
print(p.read_text(encoding="utf-8") if p.exists() else "(missing)")

7. Agent loop: model calls edit_file

Requires OPENAI_API_KEY and the complete function from the previous section.

from pathlib import Path

from mcp_agent_tools import AgentWorkspace

if "ws" not in globals():
    WORK_DIR = Path(r"D:\Avi-assign")
    WORK_DIR.mkdir(parents=True, exist_ok=True)
    ws = AgentWorkspace(WORK_DIR)
if "complete" not in globals():
    raise NameError("Run the cell above that defines `complete` (and imports) before this one.")
if client is None:
    raise ValueError(
        "Set OPENAI_API_KEY to run this cell. "
        "The direct edit_file cell above works without a key."
    )

target = "edit_agent_target.txt"
ws.write_file(
    target,
    "# Demo\nThere are three erorrs in this sentance.\n",
    mode="overwrite",
)
edit_answer = run_agent_loop(
    complete,
    (
        f"Use tools only. Read `{target}`. Then use edit_file (not write_file) to fix typos: "
        "change erorrs to errors and sentance to sentence. "
        "Copy old_string exactly from read_file; use two edit_file calls or replace_all where appropriate."
    ),
    ws,
    system_prompt=SYSTEM_PROMPT_V1,
    max_turns=14,
)
print("--- agent (edit_file) answer ---")
print(edit_answer)
print("--- file after agent ---")
print(ws.read_file(target), end="")

8. Optional: custom tool loop without run_agent_loop

Use **OPENAI_TOOL_DEFINITIONS**, call the Chat Completions API with tools=..., parse **tool_calls**, and route each call through **ws.dispatch(name, json.loads(arguments))** (requires import json). For **write_file**, the **content** field should be whatever the model authored; for **edit_file**, pass **old_string**, **new_string**, and **replace_all** exactly as the model returned.

Optional limits on AgentWorkspace

ws = AgentWorkspace(
    r"D:\Avi-assign",
    allowed_commands=frozenset({"python", "uv"}),
    command_timeout_sec=60.0,
)

Lower-level (WorkspaceTools + OPENAI_TOOL_DEFINITIONS)

Same sandbox without AgentWorkspace: use config_from_root(...) and WorkspaceTools. Pass **OPENAI_TOOL_DEFINITIONS** to your provider as tools= when you implement your own loop instead of run_agent_loop.

from mcp_agent_tools import WorkspaceTools, OPENAI_TOOL_DEFINITIONS, SYSTEM_PROMPT_V1
from mcp_agent_tools.config import config_from_root

tools = WorkspaceTools(config_from_root(r"D:\Avi-assign"))
print(tools.read_file("hello.txt"))

Compose the system message:

final_system = SYSTEM_PROMPT_V1 + "\n\n" + "Your org rules here."

Imports reference

• AgentWorkspace — pass a directory path; use read_file / write_file / edit_file / list_files / run_command on that tree only
• SYSTEM_PROMPT_V1, SYSTEM_PROMPT_CHANGELOG, TOOL_DESCRIPTIONS
• OPENAI_TOOL_DEFINITIONS — same shapes as MCP tools (for tools= in chat completions)
• build_server(config) — build a FastMCP app (stdio via build_server(cfg).run())
• run_agent_loop — minimal multi-turn executor with your complete callable (accepts WorkspaceConfig or AgentWorkspace)

Safety model

• Python: all paths are resolved under the directory you passed to AgentWorkspace(...) or config_from_root(...).
• MCP / CLI: same rule via MCP_AGENT_TOOLS_ROOT or --root (no .. escape).
• read_file, write_file, and edit_file only touch UTF-8 text paths under that root; edit_file requires valid UTF-8 (strict decode).
• run_command uses **argv only** (no shell). Optional allowlist via MCP_AGENT_TOOLS_ALLOWED_COMMANDS.
• Subprocess inherits the current environment; avoid passing secrets you do not want child processes to see.

CLI

mcp-agent-tools --root D:/your/project

Runs the MCP server on stdio (default for Cursor).

Agent loop (conceptual)

1. System = SYSTEM_PROMPT_V1 (+ optional suffix).
2. User message + OPENAI_TOOL_DEFINITIONS → your LLM.
3. For each tool_call, run WorkspaceTools.dispatch (or MCP call_tool) — including read_file, write_file, **edit_file**, list_files, and run_command as defined by the server.
4. Append tool results; repeat until the model returns text without tools.

run_agent_loop implements steps 2–4 given your complete() function.

License

MIT