slife 0.3.18

Terminal-based AI agent — a function-calling loop with minimum harness

Slife

Terminal-based AI agent — a function-calling loop with minimum harness. Chat with an LLM that calls tools, remembers everything, and orchestrates other agents over MQTT.

┌─────────────────────────────────────────────────────────────┐
│  Terminal UI (Textual)                                      │
│  ────────────────────────────────────────────────────────── │
│  Agent Loop — LLM + Tools + Stream + Memory + A2A + Inbox  │
│  ┌───────────┬──────────┬───────────┬─────────────────────┐ │
│  │ MCP Proxy │ A2A Mesh │ Subagents │ Built-in Plugins    │ │
│  │ (gateway) │ (MQTT)   │ (workers) │ Memory · MCP · WX   │ │
│  └───────────┴──────────┴───────────┴─────────────────────┘ │
│  Permanent Memory — hybrid search (grep + FTS5 + semantic)  │
└─────────────────────────────────────────────────────────────┘

Install

Zero prerequisites. The install script auto-installs uv and Node.js if needed.

Install script (recommended)

macOS / Linux / WSL:

curl -fsSL https://raw.githubusercontent.com/juzcn/slife/main/install.sh | bash

Windows PowerShell:

powershell -ExecutionPolicy Bypass -Command "irm https://raw.githubusercontent.com/juzcn/slife/main/install.ps1 | iex"

Try without installing

uvx --from git+https://github.com/juzcn/slife.git slife

Update

Re-run the install script — it auto-preserves optional packages (llama-cpp-python, sentence-transformers) from the previous install.

Uninstall

# macOS / Linux / WSL
curl -fsSL https://raw.githubusercontent.com/juzcn/slife/main/uninstall.sh | bash

# Windows PowerShell
powershell -ExecutionPolicy Bypass -Command "irm https://raw.githubusercontent.com/juzcn/slife/main/uninstall.ps1 | iex"

Uninstall removes the binaries. User data (~/.slife/, ~/.credstore/) is listed but not removed — delete manually for a full reset.

Quick Start

credstore set-password              # first time only — encrypted backup
credstore set DEEPSEEK_API_KEY       # store API key (masked input)
slife

The default config ships with pre-configured MCP servers: filesystem + shell, code search, web fetch, search APIs.

How It Works

Slife is a function-calling loop: you type → the LLM decides what tools to call → Slife executes them → the LLM responds → repeat.

You: "Find all TODO comments and create GitHub issues for them"
  → LLM calls search_content("TODO")
  → LLM calls github__create_issue(...) for each one
  → LLM: "Created 7 issues. All linked above."

Every turn is permanently recorded. On restart, recent conversations are restored.

Configuration

Two-layer model — secrets in OS keyring, config in JSON5:

Layer	Storage	Contents
Secrets	OS keyring (credstore)	API keys, tokens — encrypted at OS level
Config	~/.slife/slife.json5 → env:	${VAR} references + non-secret values

env: {
  DEEPSEEK_API_KEY: "${DEEPSEEK_API_KEY}",   // → resolved from keyring at runtime
}

models: {
  providers: {
    deepseek: {
      base_url: "https://api.deepseek.com",
      api_key: "${DEEPSEEK_API_KEY}",
      models: [{ model: "deepseek-v4-pro", name: "DeepSeek V4 Pro", reasoning: true }],
    },
  },
},
active_model: "deepseek/deepseek-v4-pro",

Secrets never reach the LLM context. All tool output is sanitized before reaching the model — API key patterns are auto-masked.

Features

Tools

All unified as OpenAI function definitions. The LLM sees no difference between native, MCP, or A2A tools.

Category	Description
Native	System info, Python execution, env/config management, skill loading, CLI discovery
MCP / REST	Filesystem, shell, code search, web fetch, any MCP server (stdio or HTTP)
Skills	On-demand plugins loaded via list_skills / use_skill
CLI	Auto-discovered external commands, persisted across restarts
A2A	Agent discovery, task routing, subagents, broadcast (13 tools)

Memory — Always On

Every conversation turn is permanently recorded. Hybrid search across four modes:

Mode	Best for
grep	Exact strings — error messages, file paths, code
fts5	Topic / keyword search with ranked snippets
hybrid	Semantic recall (FTS5 + vec0 vector search, RRF merge)
time	Browse by date

Embedding backends: local GGUF (BGE-M3, ~300 MB, offline), HuggingFace transformers, or OpenAI-compatible API. Keyword search works without any embedding backend.

Plugins

Three built-in plugins run as independent processes on Streamable HTTP transport:

Plugin	Role
slife-mcp	Gateway for external MCP servers (stdio + HTTP)
slife-memory	Diary database with hybrid search
slife-wechat	Bidirectional WeChat via iLink ClawBot

External MCP servers (filesystem, fetch, search, any OpenAPI spec) are configured in slife.json5 and auto-connected at startup. Third-party MCP servers need no Slife SDK — any stdio or HTTP MCP server works.

A2A — Agent-to-Agent

Two transports, one interface:

Transport	Use case
MQTT	Remote peers over Mosquitto broker
HTTP Streamable	Direct agent-to-agent
Subagent	Local child processes (always available)

Unified inbox serializes human, WeChat, MQTT, and subagent messages through a single queue.

Progressive Disclosure

Not all tools are in every request. Three categories use lightweight summaries:

Category	Browse	Load
Memory	memory_search	memory_open
Skills	list_skills	use_skill
MCP	mcp_list_tools	mcp_set_disclosure("eager")

Keyboard Shortcuts

Key	Action
Ctrl+C (in input)	Quit
Ctrl+C (elsewhere)	Copy
Esc	Cancel agent loop
Ctrl+L	Focus input
Home / End	Scroll to top / bottom

CLI

Flag	Description
--agent <id>	Agent identity — memory isolation key + A2A mesh name (default: slife)

Optional Extras

Extra	Enables
slife[gguf]	Local GGUF embeddings (offline)
slife[transformer]	HuggingFace transformer embeddings (~2 GB)
slife[embeddings]	Both of the above

Linux / macOS — builds from source (C compiler is standard on these platforms):

uv tool install "slife[gguf]" --reinstall

Windows — no C++ compiler by default, so use a pre-built wheel. Pick the one that matches your setup:

Your setup	Wheel	Notes
No compiler, any GPU or none	v0.3.34-vulkan	Safest — uses Vulkan if GPU present, falls back to CPU
NVIDIA GPU + CUDA 12	v0.3.34-cu132	CUDA 12.x
NVIDIA GPU + CUDA 11	v0.3.34-cu125	CUDA 11.x
AMD GPU	v0.3.34-hip-radeon	ROCm

uv tool install "slife[gguf]" --reinstall
# Then install your chosen wheel into slife's venv:
$py = (uv tool list --show-paths 2>$null | Select-String 'slife v' | Out-String) -replace '.*\((.*?)\).*', '$1\Scripts\python.exe'
uv pip install --python $py "llama-cpp-python @ https://github.com/abetlen/llama-cpp-python/releases/download/v0.3.34-vulkan/llama_cpp_python-0.3.34-py3-none-win_amd64.whl"

First use — download a GGUF model and enable it:

curl -LO https://huggingface.co/ChristianAzinn/bge-m3-gguf/resolve/main/bge-m3-Q4_K_M.gguf

Then launch slife and tell it: enable local embeddings with bge-m3-Q4_K_M.gguf

Development

git clone https://github.com/juzcn/slife.git
cd slife
uv sync --all-extras

uv run credstore set-password        # first time
uv run credstore set DEEPSEEK_API_KEY
uv run slife

Dev mode auto-detects: data files stay in the project directory. Production installs use ~/.slife/.

# Run tests
uv run pytest

# With coverage
uv run pytest --cov=slife --cov=credstore --cov-report=term-missing

Architecture

Slife is a minimum-harness agent. The harness only does what the LLM physically cannot: execute tools, maintain conversation state, stream responses, persist memory. Everything else — reasoning, planning, tool selection, error recovery — is the LLM's job.

See DESIGN.md for the full architecture: agent loop, tool system, plugin contract, MCP gateway, memory database, A2A mesh, credential security model, and project structure.

License

MIT