Skip to main content

zzm-agent

A local-first personal coding agent and CLI for OpenAI-compatible chat models.

zzm-agent gives you an interactive REPL for project work, a non-interactive exec mode for scripts and CI, persistent sessions, semantic memory, prompt snapshots, and plugin-based tools for files, search, and shell commands.

Install

pip install zzm-agent

Requires Python 3.11 or newer.

First Run

Start the REPL:

zzm-agent

On the first interactive run, zzm-agent asks for:

  • Base URL, for example https://api.openai.com/v1
  • Model name, for example gpt-4o-mini
  • LLM API key

It creates user-level configuration:

~/.zzm_agent/config.yaml
~/.zzm_agent/.env

Secrets are written to .env; config.yaml keeps references such as ${LLM_API_KEY}.

You can also use existing environment variables:

LLM_BASE_URL=https://api.openai.com/v1
LLM_MODEL_NAME=gpt-4o-mini
LLM_API_KEY=your_api_key_here

Usage

Interactive REPL:

zzm-agent
zzm-agent repl --session my-session
zzm-agent repl --config path/to/config.yaml

Non-interactive one-shot tasks:

zzm-agent exec "summarize this repository"
zzm-agent exec --stdin < prompt.txt
zzm-agent exec --json "review the latest changes"
zzm-agent exec -o answer.md "write release notes"

In exec mode, medium and high risk tools are denied instead of prompting, so CI jobs and scripts do not hang on interactive approvals.

Shell completion:

zzm-agent completion bash
zzm-agent completion zsh
zzm-agent completion powershell

Configuration

Config resolution:

  1. --config <path>
  2. ZZM_AGENT_CONFIG
  3. ~/.zzm_agent/config.yaml
  4. repository config.yaml
  5. ./config.yaml
  6. ./.zzm_agent/config.local.yaml

.env files are loaded from the current directory and next to each loaded config file before ${ENV_NAME} placeholders are expanded.

Minimal config:

model:
  base_url: "${LLM_BASE_URL:-https://api.openai.com/v1}"
  api_key: "${LLM_API_KEY}"
  model_name: "${LLM_MODEL_NAME:-gpt-4o-mini}"

agent:
  stream: true
  auto_approve: false

memory:
  path: "~/.zzm_agent/memory.json"
  max_history: 50

REPL Commands

Useful commands inside the REPL:

  • /help
  • /tools
  • /models [filter]
  • /model [id]
  • /config
  • /memory
  • /sessions
  • /session <id>
  • /new
  • /remember <fact>
  • /search <keyword>
  • /evolve run
  • /exit

Source repository and full documentation are available in the project README.

Metadata

Release files for zzm-agent 0.1.3

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for zzm-agent 0.1.3
File Size Uploaded Provenance
zzm_agent-0.1.3.tar.gz 298.2 kB Without provenance Details

Built distribution (wheel)

Table of built distributions (wheels) for zzm-agent 0.1.3
File Interpreter ABI Platform
zzm_agent-0.1.3-py3-none-any.whl Python 3 none any Details

Total release size: 547.8 kB

Release files / zzm_agent-0.1.3.tar.gz

Download URL zzm_agent-0.1.3.tar.gz
Size 298.2 kB
Tags Source
SHA-256 checksum
How to use checksums
a350706b52b7fbcebaefa3cbb8f82d10e9b0f4467815cbf9dec0699f8f6d0b3c
BLAKE2b-256 checksum
How to use checksums
15d0d664e7cdaebf92beb4f51ea720dc8ada48e90ed83e42959d50d7749cce88
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.11.15

Release files / zzm_agent-0.1.3-py3-none-any.whl

Download URL zzm_agent-0.1.3-py3-none-any.whl
Size 249.6 kB
Tags Python 3
SHA-256 checksum
How to use checksums
180d79c0c913b8bdf333868343db9611452a3fdf6c3b8176e0fe6f231bd5fd66
BLAKE2b-256 checksum
How to use checksums
b4ed6d5cb43219a364d4482b9f2bc12c7c2630fa0de283d9740e93c00c075eb5
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.11.15

Security for zzm-agent 0.1.3

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page