zenagent-py 0.1.0__tar.gz

zenagent_py-0.1.0/PKG-INFO

@@ -0,0 +1,253 @@
+Metadata-Version: 2.3
+Name: zenagent-py
+Version: 0.1.0
+Summary: A minimal, Pythonic coding agent: streaming LLM client, async-generator agent loop, and terminal harness.
+Keywords: agent,llm,ai,cli,async,coding-agent
+Author: Justin Gao
+Author-email: Justin Gao <atomyuangao@gmail.com>
+Classifier: Intended Audience :: Developers
+Classifier: Environment :: Console
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3 :: Only
+Classifier: Topic :: Software Development :: Libraries
+Classifier: Typing :: Typed
+Requires-Dist: httpx>=0.28.1
+Requires-Dist: prompt-toolkit>=3.0.53 ; extra == 'cli'
+Requires-Dist: rich>=15.0.0 ; extra == 'cli'
+Requires-Python: >=3.15
+Project-URL: Homepage, https://github.com/encyc/zen-agent
+Project-URL: Repository, https://github.com/encyc/zen-agent
+Project-URL: Changelog, https://github.com/encyc/zen-agent/commits/main
+Provides-Extra: cli
+Description-Content-Type: text/markdown
+
+# ZenAgent
+
+A minimal coding agent written from scratch in pure Python: one interface for
+streaming LLM calls, an agent loop that is an async generator, and a terminal
+harness. No vendor SDKs, no agent framework — `httpx` is the only runtime
+dependency of the kernel.
+
+Named after the Zen of Python: explicit contracts, flat layers, and one obvious
+way to do things. (Renamed from `pyagent`; there is no compatibility shim.)
+
+Built as a learning project and as the small, hackable Python kernel that the
+ecosystem is missing (in the spirit of `pi-ai` + `pi-agent-core`).
+
+Repo: <https://github.com/encyc/zen-agent> · PyPI package: `zenagent-py` (imports as `zenagent`).
+
+## Install
+
+```bash
+uv sync --extra cli        # kernel + interactive extras (prompt_toolkit, rich)
+```
+
+Requires Python 3.15+ (the project targets the current `uv` toolchain; any
+3.11+ should work for the kernel itself).
+
+## API keys
+
+Read from the environment variable named by each provider in
+`zenagent/catalog.yaml`:
+
+| provider | env var |
+|---|---|
+| `zai-coding-cn` | `ZAI_CODING_CN_API_KEY` |
+| `deepseek` | `DEEPSEEK_API_KEY` |
+| `opencode-go` | `OPENCODE_API_KEY` |
+
+## Usage
+
+```bash
+# one-shot, exits 0 on success and 1 when the provider failed
+uv run zen -p "explain what this repo does"
+
+# pick a model / reasoning level; allow write + execute tools
+uv run zen -p "create /tmp/x/hello.txt, then read it back" -m deepseek-flash --yes --thinking high
+
+# attach images (needs a vision model; deepseek-flash is one)
+uv run zen -p "what is in this picture?" --image shot.png
+
+# guardrails and reproducibility
+uv run zen -p "..." --max-turns 20 --max-cost 0.05 --no-context-files
+uv run zen -p "analyze this" --read-only --trace run.jsonl --cwd /path/to/project
+
+# structured output (a single JSON object; no streamed text)
+uv run zen -p "say hi" --json
+
+# interactive REPL
+uv run zen
+```
+
+Read-only tools (`read_file`, `list_dir`, `glob`, `grep`) run without asking.
+Write/execute tools (`write_file`, `edit_file`, `bash`) need `--yes`, an
+interactive "yes", or an "always" answer for that tool in the session.
+
+### Diagnostics
+
+```bash
+uv run zen doctor      # keys, catalog, config, session dir, terminal
+```
+
+### Extending the agent (file-based, no restart needed)
+
+| What | Where | How it shows up |
+|---|---|---|
+| **Skills** | `.zenagent/skills/<name>/SKILL.md` (project) or `~/.zenagent/skills/` | listed in the prompt; the model loads one with `read_skill` |
+| **Prompt templates** | `.zenagent/commands/<name>.md` | `/name args` expands and sends it |
+| **Extensions** | `extensions = ["path.py"]` in config or `--extension` | registers tools, slash commands, prompt text, event hooks |
+| **MCP servers** | `~/.zenagent/mcp.json` | tools exposed as `mcp__<server>__<tool>` |
+| **Custom providers/models** | `[providers.custom.<id>]` in config.toml | usable with `-m <model>` |
+
+Servers that are down at start are retried before each prompt (`/mcp` shows
+state, `/mcp retry` forces it), and `/reload` re-discovers skills, templates and
+extensions mid-session.
+
+### Library / SDK
+
+```python
+from zenagent.agent import Agent
+from zenagent.tools import ALL_TOOLS
+
+agent = Agent(
+    model=model,                 # a zenagent.providers.catalog.Model
+    system_prompt="be brief",
+    tools=list(ALL_TOOLS),       # or your own AgentTool objects
+    api_key="...",
+)
+await agent.prompt("explain this repo")   # events via agent.subscribe(fn)
+```
+
+See `examples/` for a headless run, a custom tool with a fake stream (offline),
+a direct MCP call, and a `zen rpc` client.
+
+### REPL commands
+
+| command | effect |
+|---|---|
+| `/help` | list commands |
+| `/exit` | leave (Ctrl+D also works) |
+| `/clear` | forget the conversation |
+| `/model [id]` | show / switch model (next turn) |
+| `/thinking [level]` | `off` / `low` / `medium` / `high` / `max` |
+| `/compact` | summarize the session into a new session file |
+| `/undo` | drop the last turn (writes a new session file) |
+| `/sessions` | list saved sessions |
+| `/fork <n>` | branch a new session at user message n |
+| `/import <path>` | load a session file and continue it |
+| `/image <path>` | attach an image for the next prompt |
+| `/skills` | list available skills |
+| `/mcp` | MCP server status; `/mcp retry` reconnects |
+| `/reload` | re-discover skills, templates, extensions |
+| `/theme <name>` | `default` / `mono` / `solarized` / a JSON file |
+| `/hotkeys` | show active keybindings |
+
+While the agent is working, anything you type is queued as steering input
+rather than starting a second run. Ctrl+C aborts the current run cooperatively.
+
+## Configuration
+
+Optional `~/.zenagent/config.toml` (themes, keybindings and MCP live in their own files):
+
+```toml
+default_model = "deepseek-flash"
+
+[providers.deepseek]
+base_url = "https://mirror.example/v1"   # optional
+api_key_env = "MY_DEEPSEEK_KEY"          # optional
+
+# define a provider from scratch
+[providers.custom.local]
+base_url = "http://127.0.0.1:8080/v1"
+api_key_env = "LOCAL_API_KEY"
+adapter = "openai-completions"
+[[providers.custom.local.models]]
+id = "my-model"
+context_window = 32768
+max_tokens = 4096
+reasoning = false
+
+extensions = ["~/work/my-extension.py"]
+
+[tools]
+allow = ["read_*", "grep"]     # never ask for these
+deny = ["write_file"]          # always refuse
+```
+
+Themes: `~/.zenagent/keybindings.json` (`{"newline": "escape,enter"}`) and
+`--theme mono`. MCP servers: `~/.zenagent/mcp.json` with
+`{"servers": {"name": {"command": [...]} | {"url": "...", "headers": {...}}}}`.
+
+Costs are estimated from the catalog's unit prices (input, output, and cache
+read/write separately). DeepSeek bills peak/off-peak; the catalog has no time
+dimension, so peak rates are used — treat `usage.cost` as a conservative upper
+bound. Providers with subscription pricing (zai, opencode-go) carry 0.
+
+Sessions are append-only JSONL under `~/.zenagent/sessions/`; `--resume` continues
+the most recent one. Destructive-looking operations never rewrite a file: `/undo`,
+`/fork` and `/compact` all write a new session and leave the old one intact. The file format is exactly the public message model, so it
+can be inspected with `jq` or replayed elsewhere.
+
+## Architecture
+
+```
+cli/        REPL, headless entry point, rendering, permission gate, /compact   ← harness
+agent/      loop (pure function) + Agent (state, events, steering, abort)      ← runtime
+providers/  openai-completions (+ zai/deepseek dialects), anthropic-messages   ← client
+            shared SSE transport, compat flags, packaged model catalog
+types.py    messages, content blocks, usage                     ← vocabulary
+events.py   stream events, agent events, abort token
+```
+
+Dependencies point strictly downward: `providers` knows nothing about `agent`,
+`agent` knows nothing about `cli`. The kernel (`types`, `events`, `providers`,
+`agent`, `tools`) imports only `httpx`; `prompt_toolkit`/`rich` live in the CLI
+extra.
+
+### Contracts worth knowing
+
+- **`StreamFn` never raises.** Every failure (network, 4xx/5xx, parse error,
+  cancellation) becomes a terminal `done`/`error` event. `safe_stream` enforces
+  this for any stream function, including injected fakes.
+- **Cancellation is a protocol state, not an exception.** The transport raises
+  `Aborted` for cooperative cancellation; `asyncio.CancelledError` is reserved
+  for real task cancellation.
+- **Tools cannot crash the loop.** Unknown tools, invalid arguments, blocked
+  calls, exceptions, and truncated (`stop_reason == "length"`) calls all become
+  `is_error` tool results. Truncated calls are never executed and always get a
+  result, so the transcript never carries a dangling `tool_call`.
+- **Every event carries a `partial` snapshot**, so the UI never assembles deltas
+  itself; assembly belongs to the adapter and happens once.
+
+## Tools
+
+`read_file`, `write_file`, `edit_file` (exactly-one-match replacement),
+`list_dir`, `bash` (merged output, 30 KB head+tail truncation, process-group
+kill on timeout/abort), `glob`, `grep` (ripgrep when available, Python fallback).
+
+## Tests
+
+```bash
+bash scripts/check.sh     # the gate: sync + ruff + pytest (CI runs exactly this)
+bash scripts/check.sh     # the gate: sync + ruff + pytest (CI runs exactly this)
+uv run pytest -q          # fully offline: fake stream fns + replayed SSE fixtures
+./scripts/eval.sh         # behavioral evals against a real model (opt-in, costs money)
+```
+
+Behavioral evals (real model, small cost) check the agent rather than the
+code; they are separate from the offline gate:
+
+```bash
+./scripts/eval.sh --list                      # case ids
+./scripts/eval.sh --model deepseek-flash --repeats 3
+```
+
+`scripts/capture_sse.py` records real provider traffic into
+`zenagent/providers/fixtures/` so the parser can be checked against the wire; the
+replay tests skip when a capture is absent.
+
+## Status
+
+Milestones 1–2 are complete and verified against a live provider; the
+interactive harness and the second protocol adapter are implemented. Not in
+scope (v1): MCP, subagents, sandboxing, a full-screen TUI, custom providers.

zenagent_py-0.1.0/README.md

@@ -0,0 +1,230 @@
+# ZenAgent
+
+A minimal coding agent written from scratch in pure Python: one interface for
+streaming LLM calls, an agent loop that is an async generator, and a terminal
+harness. No vendor SDKs, no agent framework — `httpx` is the only runtime
+dependency of the kernel.
+
+Named after the Zen of Python: explicit contracts, flat layers, and one obvious
+way to do things. (Renamed from `pyagent`; there is no compatibility shim.)
+
+Built as a learning project and as the small, hackable Python kernel that the
+ecosystem is missing (in the spirit of `pi-ai` + `pi-agent-core`).
+
+Repo: <https://github.com/encyc/zen-agent> · PyPI package: `zenagent-py` (imports as `zenagent`).
+
+## Install
+
+```bash
+uv sync --extra cli        # kernel + interactive extras (prompt_toolkit, rich)
+```
+
+Requires Python 3.15+ (the project targets the current `uv` toolchain; any
+3.11+ should work for the kernel itself).
+
+## API keys
+
+Read from the environment variable named by each provider in
+`zenagent/catalog.yaml`:
+
+| provider | env var |
+|---|---|
+| `zai-coding-cn` | `ZAI_CODING_CN_API_KEY` |
+| `deepseek` | `DEEPSEEK_API_KEY` |
+| `opencode-go` | `OPENCODE_API_KEY` |
+
+## Usage
+
+```bash
+# one-shot, exits 0 on success and 1 when the provider failed
+uv run zen -p "explain what this repo does"
+
+# pick a model / reasoning level; allow write + execute tools
+uv run zen -p "create /tmp/x/hello.txt, then read it back" -m deepseek-flash --yes --thinking high
+
+# attach images (needs a vision model; deepseek-flash is one)
+uv run zen -p "what is in this picture?" --image shot.png
+
+# guardrails and reproducibility
+uv run zen -p "..." --max-turns 20 --max-cost 0.05 --no-context-files
+uv run zen -p "analyze this" --read-only --trace run.jsonl --cwd /path/to/project
+
+# structured output (a single JSON object; no streamed text)
+uv run zen -p "say hi" --json
+
+# interactive REPL
+uv run zen
+```
+
+Read-only tools (`read_file`, `list_dir`, `glob`, `grep`) run without asking.
+Write/execute tools (`write_file`, `edit_file`, `bash`) need `--yes`, an
+interactive "yes", or an "always" answer for that tool in the session.
+
+### Diagnostics
+
+```bash
+uv run zen doctor      # keys, catalog, config, session dir, terminal
+```
+
+### Extending the agent (file-based, no restart needed)
+
+| What | Where | How it shows up |
+|---|---|---|
+| **Skills** | `.zenagent/skills/<name>/SKILL.md` (project) or `~/.zenagent/skills/` | listed in the prompt; the model loads one with `read_skill` |
+| **Prompt templates** | `.zenagent/commands/<name>.md` | `/name args` expands and sends it |
+| **Extensions** | `extensions = ["path.py"]` in config or `--extension` | registers tools, slash commands, prompt text, event hooks |
+| **MCP servers** | `~/.zenagent/mcp.json` | tools exposed as `mcp__<server>__<tool>` |
+| **Custom providers/models** | `[providers.custom.<id>]` in config.toml | usable with `-m <model>` |
+
+Servers that are down at start are retried before each prompt (`/mcp` shows
+state, `/mcp retry` forces it), and `/reload` re-discovers skills, templates and
+extensions mid-session.
+
+### Library / SDK
+
+```python
+from zenagent.agent import Agent
+from zenagent.tools import ALL_TOOLS
+
+agent = Agent(
+    model=model,                 # a zenagent.providers.catalog.Model
+    system_prompt="be brief",
+    tools=list(ALL_TOOLS),       # or your own AgentTool objects
+    api_key="...",
+)
+await agent.prompt("explain this repo")   # events via agent.subscribe(fn)
+```
+
+See `examples/` for a headless run, a custom tool with a fake stream (offline),
+a direct MCP call, and a `zen rpc` client.
+
+### REPL commands
+
+| command | effect |
+|---|---|
+| `/help` | list commands |
+| `/exit` | leave (Ctrl+D also works) |
+| `/clear` | forget the conversation |
+| `/model [id]` | show / switch model (next turn) |
+| `/thinking [level]` | `off` / `low` / `medium` / `high` / `max` |
+| `/compact` | summarize the session into a new session file |
+| `/undo` | drop the last turn (writes a new session file) |
+| `/sessions` | list saved sessions |
+| `/fork <n>` | branch a new session at user message n |
+| `/import <path>` | load a session file and continue it |
+| `/image <path>` | attach an image for the next prompt |
+| `/skills` | list available skills |
+| `/mcp` | MCP server status; `/mcp retry` reconnects |
+| `/reload` | re-discover skills, templates, extensions |
+| `/theme <name>` | `default` / `mono` / `solarized` / a JSON file |
+| `/hotkeys` | show active keybindings |
+
+While the agent is working, anything you type is queued as steering input
+rather than starting a second run. Ctrl+C aborts the current run cooperatively.
+
+## Configuration
+
+Optional `~/.zenagent/config.toml` (themes, keybindings and MCP live in their own files):
+
+```toml
+default_model = "deepseek-flash"
+
+[providers.deepseek]
+base_url = "https://mirror.example/v1"   # optional
+api_key_env = "MY_DEEPSEEK_KEY"          # optional
+
+# define a provider from scratch
+[providers.custom.local]
+base_url = "http://127.0.0.1:8080/v1"
+api_key_env = "LOCAL_API_KEY"
+adapter = "openai-completions"
+[[providers.custom.local.models]]
+id = "my-model"
+context_window = 32768
+max_tokens = 4096
+reasoning = false
+
+extensions = ["~/work/my-extension.py"]
+
+[tools]
+allow = ["read_*", "grep"]     # never ask for these
+deny = ["write_file"]          # always refuse
+```
+
+Themes: `~/.zenagent/keybindings.json` (`{"newline": "escape,enter"}`) and
+`--theme mono`. MCP servers: `~/.zenagent/mcp.json` with
+`{"servers": {"name": {"command": [...]} | {"url": "...", "headers": {...}}}}`.
+
+Costs are estimated from the catalog's unit prices (input, output, and cache
+read/write separately). DeepSeek bills peak/off-peak; the catalog has no time
+dimension, so peak rates are used — treat `usage.cost` as a conservative upper
+bound. Providers with subscription pricing (zai, opencode-go) carry 0.
+
+Sessions are append-only JSONL under `~/.zenagent/sessions/`; `--resume` continues
+the most recent one. Destructive-looking operations never rewrite a file: `/undo`,
+`/fork` and `/compact` all write a new session and leave the old one intact. The file format is exactly the public message model, so it
+can be inspected with `jq` or replayed elsewhere.
+
+## Architecture
+
+```
+cli/        REPL, headless entry point, rendering, permission gate, /compact   ← harness
+agent/      loop (pure function) + Agent (state, events, steering, abort)      ← runtime
+providers/  openai-completions (+ zai/deepseek dialects), anthropic-messages   ← client
+            shared SSE transport, compat flags, packaged model catalog
+types.py    messages, content blocks, usage                     ← vocabulary
+events.py   stream events, agent events, abort token
+```
+
+Dependencies point strictly downward: `providers` knows nothing about `agent`,
+`agent` knows nothing about `cli`. The kernel (`types`, `events`, `providers`,
+`agent`, `tools`) imports only `httpx`; `prompt_toolkit`/`rich` live in the CLI
+extra.
+
+### Contracts worth knowing
+
+- **`StreamFn` never raises.** Every failure (network, 4xx/5xx, parse error,
+  cancellation) becomes a terminal `done`/`error` event. `safe_stream` enforces
+  this for any stream function, including injected fakes.
+- **Cancellation is a protocol state, not an exception.** The transport raises
+  `Aborted` for cooperative cancellation; `asyncio.CancelledError` is reserved
+  for real task cancellation.
+- **Tools cannot crash the loop.** Unknown tools, invalid arguments, blocked
+  calls, exceptions, and truncated (`stop_reason == "length"`) calls all become
+  `is_error` tool results. Truncated calls are never executed and always get a
+  result, so the transcript never carries a dangling `tool_call`.
+- **Every event carries a `partial` snapshot**, so the UI never assembles deltas
+  itself; assembly belongs to the adapter and happens once.
+
+## Tools
+
+`read_file`, `write_file`, `edit_file` (exactly-one-match replacement),
+`list_dir`, `bash` (merged output, 30 KB head+tail truncation, process-group
+kill on timeout/abort), `glob`, `grep` (ripgrep when available, Python fallback).
+
+## Tests
+
+```bash
+bash scripts/check.sh     # the gate: sync + ruff + pytest (CI runs exactly this)
+bash scripts/check.sh     # the gate: sync + ruff + pytest (CI runs exactly this)
+uv run pytest -q          # fully offline: fake stream fns + replayed SSE fixtures
+./scripts/eval.sh         # behavioral evals against a real model (opt-in, costs money)
+```
+
+Behavioral evals (real model, small cost) check the agent rather than the
+code; they are separate from the offline gate:
+
+```bash
+./scripts/eval.sh --list                      # case ids
+./scripts/eval.sh --model deepseek-flash --repeats 3
+```
+
+`scripts/capture_sse.py` records real provider traffic into
+`zenagent/providers/fixtures/` so the parser can be checked against the wire; the
+replay tests skip when a capture is absent.
+
+## Status
+
+Milestones 1–2 are complete and verified against a live provider; the
+interactive harness and the second protocol adapter are implemented. Not in
+scope (v1): MCP, subagents, sandboxing, a full-screen TUI, custom providers.

zenagent_py-0.1.0/pyproject.toml

@@ -0,0 +1,61 @@
+[project]
+name = "zenagent-py"
+version = "0.1.0"
+description = "A minimal, Pythonic coding agent: streaming LLM client, async-generator agent loop, and terminal harness."
+readme = "README.md"
+keywords = [
+    "agent",
+    "llm",
+    "ai",
+    "cli",
+    "async",
+    "coding-agent",
+]
+classifiers = [
+    "Intended Audience :: Developers",
+    "Environment :: Console",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3 :: Only",
+    "Topic :: Software Development :: Libraries",
+    "Typing :: Typed",
+]
+requires-python = ">=3.15"
+dependencies = ["httpx>=0.28.1"]
+
+[[project.authors]]
+name = "Justin Gao"
+email = "atomyuangao@gmail.com"
+
+[project.scripts]
+zen = "zenagent.cli.main:main"
+zenagent = "zenagent.cli.main:main"
+
+[project.optional-dependencies]
+cli = [
+    "prompt-toolkit>=3.0.53",
+    "rich>=15.0.0",
+]
+
+[project.urls]
+Homepage = "https://github.com/encyc/zen-agent"
+Repository = "https://github.com/encyc/zen-agent"
+Changelog = "https://github.com/encyc/zen-agent/commits/main"
+
+[build-system]
+requires = ["uv_build>=0.11.24,<0.12.0"]
+build-backend = "uv_build"
+
+[tool.uv.build-backend]
+module-root = ""
+module-name = "zenagent"
+
+[tool.pytest.ini_options]
+asyncio_mode = "auto"
+asyncio_default_fixture_loop_scope = "function"
+
+[dependency-groups]
+dev = [
+    "pytest>=9.2.0",
+    "pytest-asyncio>=1.4.1",
+    "ruff>=0.17.3",
+]

zenagent_py-0.1.0/pyproject.toml.orig

@@ -0,0 +1,55 @@
+[project]
+name = "zenagent-py"
+version = "0.1.0"
+description = "A minimal, Pythonic coding agent: streaming LLM client, async-generator agent loop, and terminal harness."
+readme = "README.md"
+keywords = ["agent", "llm", "ai", "cli", "async", "coding-agent"]
+classifiers = [
+    "Intended Audience :: Developers",
+    "Environment :: Console",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3 :: Only",
+    "Topic :: Software Development :: Libraries",
+    "Typing :: Typed",
+]
+authors = [
+    { name = "Justin Gao", email = "atomyuangao@gmail.com" }
+]
+requires-python = ">=3.15"
+dependencies = [
+    "httpx>=0.28.1",
+]
+
+[project.scripts]
+zen = "zenagent.cli.main:main"
+zenagent = "zenagent.cli.main:main"
+
+[project.optional-dependencies]
+cli = [
+    "prompt-toolkit>=3.0.53",
+    "rich>=15.0.0",
+]
+
+[project.urls]
+Homepage = "https://github.com/encyc/zen-agent"
+Repository = "https://github.com/encyc/zen-agent"
+Changelog = "https://github.com/encyc/zen-agent/commits/main"
+
+[build-system]
+requires = ["uv_build>=0.11.24,<0.12.0"]
+build-backend = "uv_build"
+
+[tool.uv.build-backend]
+module-root = ""
+module-name = "zenagent"
+
+[dependency-groups]
+dev = [
+    "pytest>=9.2.0",
+    "pytest-asyncio>=1.4.1",
+    "ruff>=0.17.3",
+]
+
+[tool.pytest.ini_options]
+asyncio_mode = "auto"
+asyncio_default_fixture_loop_scope = "function"

zenagent_py-0.1.0/zenagent/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.1.0"

zenagent_py-0.1.0/zenagent/agent/__init__.py

@@ -0,0 +1,24 @@
+"""Agent package: the loop, its config/state/tool surface, and the Agent shell.
+
+``zenagent.agent.loop`` is the pure core; ``zenagent.agent.agent.Agent`` is the
+stateful wrapper harnesses drive. Tool packages (later tasks) import
+``AgentTool``/``ToolResultAgent`` from here to build instances.
+"""
+
+from zenagent.agent.agent import Agent
+from zenagent.agent.loop import (
+    AgentConfig,
+    AgentState,
+    AgentTool,
+    ToolResultAgent,
+    agent_loop,
+)
+
+__all__ = [
+    "Agent",
+    "AgentConfig",
+    "AgentState",
+    "AgentTool",
+    "ToolResultAgent",
+    "agent_loop",
+]