agentbox-rl 0.2.0__tar.gz

agentbox_rl-0.2.0/.gitignore

@@ -0,0 +1,54 @@
+# Python
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+
+# Virtual envs
+.venv/
+venv/
+ENV/
+
+# Testing / typing
+.pytest_cache/
+.mypy_cache/
+.ruff_cache/
+.coverage
+htmlcov/
+.tox/
+
+# IDE
+.idea/
+.vscode/
+*.swp
+.DS_Store
+
+# Project artifacts
+trajectories/
+out/
+*.log
+.env
+.env.*
+
+# Docker
+.docker-cache/
+bench-results/
+
+# Pi agent artifacts
+.pi-subagents/

agentbox_rl-0.2.0/LICENSE

@@ -0,0 +1,176 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranty or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS

agentbox_rl-0.2.0/PKG-INFO

@@ -0,0 +1,427 @@
+Metadata-Version: 2.4
+Name: agentbox-rl
+Version: 0.2.0
+Summary: Fully local agentic virtual environment for trajectory building
+Project-URL: Homepage, https://github.com/konic-labs/agentbox
+Project-URL: Documentation, https://github.com/konic-labs/agentbox#readme
+Author: AgentBox Contributors
+License-Expression: Apache-2.0
+License-File: LICENSE
+Keywords: agents,docker,llm,rl,sandbox,trajectory
+Classifier: Development Status :: 3 - Alpha
+Classifier: Intended Audience :: Developers
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Python: >=3.11
+Requires-Dist: anyio>=4.0
+Requires-Dist: docker>=7.0
+Requires-Dist: httpx>=0.27
+Requires-Dist: openai>=1.40
+Requires-Dist: pydantic>=2.6
+Requires-Dist: structlog>=24.0
+Requires-Dist: typer>=0.12
+Provides-Extra: art
+Requires-Dist: openpipe-art>=0.4; extra == 'art'
+Provides-Extra: dev
+Requires-Dist: mypy>=1.11; extra == 'dev'
+Requires-Dist: pytest-asyncio>=0.24; extra == 'dev'
+Requires-Dist: pytest-timeout>=2.3; extra == 'dev'
+Requires-Dist: pytest>=8.0; extra == 'dev'
+Requires-Dist: ruff>=0.6; extra == 'dev'
+Provides-Extra: generate
+Requires-Dist: dspy>=2.5; extra == 'generate'
+Provides-Extra: rich
+Requires-Dist: rich>=13; extra == 'rich'
+Description-Content-Type: text/markdown
+
+<div align="center">
+
+# AgentBox
+
+**ART-native rollout engine for real agentic trajectories in Docker sandboxes.**
+
+[Quick Start](#quick-start) ·
+[Workflow](#workflow) ·
+[Tools](#tools) ·
+[Models](#models--providers) ·
+[Tasks](#tasks) ·
+[Trajectories](#trajectories--art) ·
+[CLI](#cli-reference) ·
+[Benchmarks](#benchmarks) ·
+[Docs](#technical-docs) ·
+[Development](#development) ·
+[License](#license)
+
+![Python](https://img.shields.io/badge/Python-3.11%2B-blue)
+![Runtime](https://img.shields.io/badge/Runtime-Docker%20%7C%20AsyncIO-black)
+![ART](https://img.shields.io/badge/ART-native-orange)
+![Package](https://img.shields.io/badge/Package-agentbox-green)
+![License](https://img.shields.io/badge/License-Apache%202.0-green)
+
+</div>
+
+AgentBox is an **ART-native** environment + rollout harness: isolated Docker
+sandboxes, structured tools (or pure shell), task seeding, verifiers, and
+multi-turn trajectories shaped for [OpenPipe ART](https://github.com/OpenPipe/ART)
+(`messages_and_choices`, rewards, groups) — for online GRPO with ART’s inference
+backend, offline SFT dumps, and the same real-rollout path for benchmarks.
+
+It is **not** a trainer. ART (or TRL / Unsloth / verl) owns training; AgentBox
+owns the virtual env and trajectory collection.
+
+## Highlights
+
+- **ART-native trajectories** — `to_art_dict()` / `to_art()`; GRPO groups via `ParallelRunner`
+- **One container per rollout** — clean state, parallel isolation, labeled GC
+- **OpenAI tools protocol** — Chat Completions + tool calling (ART-compatible wire format)
+- **Provider-agnostic models** — Featherless, OpenRouter, vLLM, Ollama, ART backend client, …
+- **Builtins + custom tools** — `BaseTool` / `@tool`, override-by-name
+- **Task seeding** — `starter_files` + `setup_commands` before the agent starts
+- **Objective rewards** — pytest / command verifiers (hybrid-ready with ART RULER)
+- **ParallelRunner** — concurrent rollouts and GRPO-style groups
+- **Trajectory export** — JSON, JSONL, ART-native dicts
+- **Typer CLI** — `doctor`, `generate`, `bench`, `traj`, `run`, `export`, `prune`
+- **Real-rollout benchmarks** — freeze tasks + env; multi-student scoring
+- **Task generation** — batch + two-stage stubs + static/Docker/LLM QC (no private scripts)
+- **Trajectory HTML** — official verifier vs agent self-check dashboard
+
+## Quick Start
+
+**Prerequisites:** Python 3.11+, [Docker](https://docs.docker.com/get-docker/)
+daemon running, optional [`uv`](https://github.com/astral-sh/uv).
+
+```bash
+git clone https://github.com/konic-labs/agentbox.git
+cd agentbox
+uv venv .venv --seed --python 3.12
+source .venv/bin/activate
+uv pip install -e ".[dev]"
+
+# Optional
+uv pip install -e ".[rich]"      # progress bars
+uv pip install -e ".[art]"       # live art.Trajectory
+uv pip install -e ".[generate]"  # DSPy task generation
+```
+
+Sanity checks (no external LLM required):
+
+```bash
+agentbox version
+agentbox doctor
+uv run pytest tests/unit -q
+python examples/hello_rollout.py
+```
+
+### Minimal Python rollout (mock)
+
+```python
+import asyncio
+from agentbox import Task, Rollout
+from agentbox.config import SandboxConfig, ResourceLimits
+from agentbox.model import MockModelClient, ModelResponse
+from agentbox.trajectory.schema import ToolCall, FunctionCall
+import json
+
+async def main():
+    task = Task.from_json("examples/tasks/fix_fizzbuzz/task.json")
+    fixed = 'def fizzbuzz(n):\n    return "Fizz" if n % 3 == 0 else str(n)\n'
+    mock = MockModelClient([
+        ModelResponse(
+            content=None,
+            tool_calls=[ToolCall(
+                id="c1",
+                function=FunctionCall(
+                    name="write_file",
+                    arguments=json.dumps({"path": "fizzbuzz.py", "content": fixed}),
+                ),
+            )],
+        ),
+        ModelResponse(content="done", tool_calls=[]),
+    ])
+    traj = await Rollout.run(
+        task,
+        model=mock,
+        sandbox=SandboxConfig(
+            limits=ResourceLimits(network_disabled=False),
+        ),
+    )
+    print(traj.final_status, traj.reward)
+    traj.save("out/traj.json")
+
+asyncio.run(main())
+```
+
+### Real model (Ollama)
+
+```bash
+agentbox run examples/tasks/fix_fizzbuzz/task.json \
+  --model qwen2.5-coder:7b \
+  --base-url http://localhost:11434/v1 \
+  --api-key ollama \
+  --network \
+  --out trajectories/
+```
+
+```python
+from agentbox import Agent, Task, Rollout
+
+agent = Agent(
+    model="qwen2.5-coder:7b",
+    base_url="http://localhost:11434/v1",
+    api_key="ollama",
+    max_steps=20,
+)
+traj = await Rollout.run(task=Task.from_json("examples/tasks/fix_fizzbuzz/task.json"), agent=agent)
+```
+
+### Parallel / GRPO-style groups
+
+```python
+from agentbox import ParallelRunner, Agent
+
+runner = ParallelRunner(concurrency=16, agent=agent)
+trajs = await runner.run_tasks(tasks, n_per_task=1)
+groups = await runner.run_groups(tasks, group_size=4)
+art = [[t.to_art_dict() for t in g] for g in groups]
+```
+
+## Workflow
+
+| Step | What happens | Artifact |
+| --- | --- | --- |
+| 0. Task | Load or generate Task definition | `task.json` |
+| 1. Sandbox | Create labeled Docker container | running container |
+| 2. Seed | Write `starter_files`, run `setup_commands` | `/workspace` ready |
+| 3. Agent | Multi-turn tools via OpenAI protocol | messages + tool records |
+| 4. Verify | pytest / command exit code | reward |
+| 5. Record | Finalize trajectory, destroy container | JSON / ART dict |
+
+```txt
+task → create sandbox → seed → agent loop → verifier → trajectory → destroy
+```
+
+## Tools
+
+| Builtin | Role |
+| --- | --- |
+| `list_files` | List workspace paths |
+| `read_file` | Read file content |
+| `write_file` | Create / overwrite |
+| `edit_file` | Exact one-occurrence replace (`old_string` / `new_string`) |
+| `run_command` | Shell in container |
+| `run_tests` | Default `python -m pytest -q` |
+
+Modes: `structured` (all builtins), `shell` (`run_command` only), `custom`.
+
+Custom tools:
+
+```python
+from agentbox.tools import tool
+
+@tool(description="Count lines")
+async def count_lines(sandbox, path: str) -> str:
+    r = await sandbox.exec(f"wc -l -- {path}")
+    return r.stdout.strip()
+
+agent = Agent(model="...", base_url="...", custom_tools=[count_lines])
+```
+
+Same `name` overrides a builtin. Details: [docs/tools.md](docs/tools.md).
+
+## Models & Providers
+
+OpenAI Chat Completions + tools only. Configure any compatible endpoint:
+
+| Type | Examples | Config |
+| --- | --- | --- |
+| External | Featherless, OpenRouter, Together, Fireworks | `base_url` + `api_key` |
+| Local | vLLM, SGLang, Ollama `/v1` | `base_url`; `api_key` optional |
+
+```python
+Agent(model="glm-5.2", base_url="https://api.featherless.ai/v1", api_key="...")
+Agent(model="Qwen/Qwen2.5-7B-Instruct", base_url="http://localhost:8000/v1")
+```
+
+## Tasks
+
+```json
+{
+  "task_id": "fix_fizzbuzz_001",
+  "description": "Fix fizzbuzz so tests pass.",
+  "starter_files": { "fizzbuzz.py": "...", "test_fizzbuzz.py": "..." },
+  "setup_commands": ["pip install -q pytest"],
+  "verifier": {
+    "type": "pytest",
+    "command": "python -m pytest -q",
+    "success_exit_code": 0
+  },
+  "metadata": { "difficulty": "easy", "tags": ["python"] }
+}
+```
+
+Automated generation (optional):
+
+```python
+from agentbox.tasks.generate import TaskGenerator, GenerateConfig
+
+gen = TaskGenerator(GenerateConfig(
+    model="glm-5.2",
+    base_url="https://api.featherless.ai/v1",
+    api_key="...",
+    validate_in_docker=True,
+))
+task = await gen.generate(difficulty="easy", domain="python")
+```
+
+## Trajectories & ART
+
+AgentBox is built to drop into ART workflows:
+
+| Flow | AgentBox role | ART role |
+| --- | --- | --- |
+| **Online GRPO** | Docker env, tools, tasks, optional verifier | Inference (vLLM/LoRA) + `backend.train` |
+| **Offline SFT** | Bulk rollouts from any API | SFT on exported trajectories |
+| **Rewards** | Verifier scores | Optional RULER on groups (hybrid OK) |
+
+```python
+traj.save("trajectories/run.json")
+art = traj.to_art_dict()   # ART-native: messages_and_choices, reward, metrics, metadata
+# uv pip install -e ".[art]"
+# live = traj.to_art()     # openpipe-art Trajectory object
+```
+
+```bash
+agentbox export traj.json --format art -o traj.art.json
+python examples/art_integration.py
+```
+
+## Production workflows
+
+End-to-end lab path **without private scripts**:
+
+```bash
+# 0) Optional project config (teacher + students)
+cp examples/agentbox.yaml ./agentbox.yaml   # edit endpoints
+
+# 1) Probe tool-calling before a long suite
+agentbox doctor --model MODEL --base-url http://localhost:8000/v1
+
+# 2) Generate a QC-passed task set
+agentbox generate batch -m MODEL --base-url URL -n 20 -c 8 --two-stage \
+  -o generated/tasks
+
+# 3) Freeze a suite and run multi-student bench
+agentbox bench create suites/coding-v1 --from-tasks generated/tasks \
+  --suite-id coding-v1 --name "coding v1"
+agentbox bench run suites/coding-v1 \
+  --student small=qwen3.5:0.8b@http://localhost:11434/v1 \
+  --student big=MODEL@http://localhost:8000/v1 \
+  --probe --out bench-results/run1
+
+# 4) Inspect
+agentbox bench show bench-results/run1/report.json --out REPORT.md
+agentbox traj render bench-results/run1 -o traj-dash.html
+
+# 5) ART offline groups
+python examples/art_grpo_offline.py
+```
+
+Details: [docs/generation.md](docs/generation.md) · [docs/art_grpo.md](docs/art_grpo.md) · [docs/trajectories.md](docs/trajectories.md).
+
+## CLI Reference
+
+```bash
+agentbox doctor [--prune] [--model M --base-url URL]
+agentbox build-image [--tag agentbox/sandbox:latest]
+agentbox generate one|batch|validate-llm|validate-docker …
+agentbox run TASK.json -m MODEL --base-url URL [--network] [--out trajectories/]
+agentbox run-dir tasks/ -m MODEL --base-url URL -c 16 --n 4
+agentbox bench create DIR --from-tasks tasks/ --suite-id ID --name NAME
+agentbox bench freeze DIR && agentbox bench validate DIR --strict
+agentbox bench run DIR -m MODEL --base-url URL --model-id LABEL -o bench-results/run1
+agentbox bench run DIR --student id=model@url --probe --limit 5
+agentbox bench show bench-results/run1/report.json --out extended.md
+agentbox traj show traj.json
+agentbox traj render bench-results/run1 -o traj-dash.html
+agentbox export traj.json --format art -o out.json
+agentbox prune
+```
+
+Full flag list: [docs/cli.md](docs/cli.md) · [docs/benchmarks.md](docs/benchmarks.md).
+
+## Benchmarks
+
+Same Docker rollouts as training collection, frozen as a suite, scored against
+any OpenAI-compatible endpoint (local or external).
+
+```bash
+python examples/bench_run.py   # hermetic mock solver (Docker required)
+
+agentbox bench run examples/benchmarks/coding-mini \
+  --model-id ollama-qwen \
+  --model qwen2.5-coder:7b \
+  --base-url http://localhost:11434/v1 \
+  --api-key ollama \
+  --out bench-results/run1
+```
+
+Setup checks run after seed (env health); task verifiers own success/reward.
+Details: [docs/benchmarks.md](docs/benchmarks.md).
+
+## Sandbox image
+
+Default: **`python:3.12-slim-bookworm`** (glibc; not Alpine).
+
+```bash
+agentbox build-image   # optional baked image with pytest + unix tools
+```
+
+Override: `SandboxConfig(image="my-org/env:1.2")` or presets in
+`agentbox.sandbox.images`.
+
+## Technical Docs
+
+| Doc | Concern |
+| --- | --- |
+| [docs/index.md](docs/index.md) | Documentation map |
+| [docs/setup.md](docs/setup.md) | Install & first runs |
+| [docs/architecture.md](docs/architecture.md) | Module boundaries & data flow |
+| [docs/configuration.md](docs/configuration.md) | All config fields |
+| [docs/sandbox.md](docs/sandbox.md) | Docker lifecycle, jail, prune |
+| [docs/tools.md](docs/tools.md) | Builtins & custom tools |
+| [docs/models.md](docs/models.md) | Providers & clients |
+| [docs/agent.md](docs/agent.md) | Loop, timeouts, prompts |
+| [docs/tasks.md](docs/tasks.md) | Schema, seeder, verifier |
+| [docs/generation.md](docs/generation.md) | Automated task generation |
+| [docs/trajectories.md](docs/trajectories.md) | Formats & ART export |
+| [docs/runner.md](docs/runner.md) | Rollout & parallel |
+| [docs/cli.md](docs/cli.md) | CLI reference |
+| [docs/benchmarks.md](docs/benchmarks.md) | Real-rollout multi-model suites |
+| [docs/security.md](docs/security.md) | Isolation model |
+| [docs/development.md](docs/development.md) | Tests & extension |
+
+## Development
+
+```bash
+uv pip install -e ".[dev]"
+uv run pytest tests/unit -q
+uv run pytest tests/integration -q -m docker
+uv run agentbox doctor
+```
+
+Conventional Commits (`feat:`, `fix:`, `docs:`, `test:`, …). See
+[docs/development.md](docs/development.md).
+
+## Security (short)
+
+Tools run **only** inside Docker; paths jailed under `/workspace`; network off
+by default; never mount the Docker socket into sandboxes. Details:
+[docs/security.md](docs/security.md).
+
+## License
+
+Apache License 2.0. See [LICENSE](LICENSE).