docs: split license files, add citation and repository metadata

.github/CODEOWNERS

@@ -0,0 +1 @@
+* @antoinegg1

.github/dependabot.yml

@@ -0,0 +1,16 @@
+version: 2
+updates:
+  - package-ecosystem: pip
+    directory: /
+    schedule:
+      interval: monthly
+    groups:
+      pip:
+        patterns: ["*"]
+  - package-ecosystem: github-actions
+    directory: /
+    schedule:
+      interval: monthly
+    groups:
+      github-actions:
+        patterns: ["*"]

.github/workflows/offline.yml

@@ -8,8 +8,8 @@ jobs:
   tests:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v4
-      - uses: actions/setup-python@v5
+      - uses: actions/checkout@11d5960a326750d5838078e36cf38b85af677262 # v4.4.0
+      - uses: actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065 # v5.6.0
         with:
           python-version: '3.12'
       - run: python -m pip install -r requirements-repro.lock

CITATION.cff

@@ -0,0 +1,20 @@
+cff-version: 1.2.0
+message: >-
+  If you use this software, please cite it as below. The paper citation will be
+  added here with the paper release.
+type: software
+title: "Humanize MLE Agents: Multi-Agent Session Alternation Beyond Single-Agent Limits"
+abstract: >-
+  Experiment code for Humanize MLE Agents (HMA), which alternates two native
+  coding agents over a shared machine-learning workspace.
+authors:
+  - name: HMA contributors
+version: 1.0.0
+license:
+  - Apache-2.0
+  - MIT
+repository-code: "https://github.com/humanfia/hma"
+keywords:
+  - agents
+  - machine-learning-engineering
+  - mle-bench

LICENSE

@@ -1,23 +1,3 @@
-HMA - anonymous review edition
-
-Copyright 2026 HMA contributors (identity omitted for anonymous review).
-
-src/hma: HMA controller and provider adapter, Apache License 2.0.
-src/hmz: supporting agent runtime, Apache License 2.0.
-src/hma/benchmark: benchmark evaluator integration, MIT License.
-
-Identifying project metadata has been removed for anonymous review. This
-edition uses HMA module and container names, defaults to a five-experiment cap,
-removes operational comments and remote catalog links, and disables telemetry.
-These changes are packaging and anonymity adaptations. External model CLIs,
-the upstream MLE-bench grader, and task data retain their own licenses and are
-not included. Both license texts are included below in this file.
-
-========================================================================
-Apache License 2.0: controller and agent runtime
-========================================================================
-
-
                                  Apache License
                            Version 2.0, January 2004
                         http://www.apache.org/licenses/
@@ -220,28 +200,3 @@ Apache License 2.0: controller and agent runtime
    See the License for the specific language governing permissions and
    limitations under the License.
 
-========================================================================
-MIT License: benchmark integration
-========================================================================
-
-MIT License
-
-Copyright (c) HMA contributors (identity omitted for anonymous review)
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.

NOTICE

@@ -0,0 +1,17 @@
+HMA - anonymous review edition
+
+Copyright 2026 HMA contributors (identity omitted for anonymous review).
+
+src/hma: HMA controller and provider adapter, Apache License 2.0.
+src/hmz: supporting agent runtime, Apache License 2.0.
+src/hma/benchmark: benchmark evaluator integration, MIT License.
+
+SPDX-License-Identifier: Apache-2.0 AND MIT
+
+Identifying project metadata has been removed for anonymous review. This
+edition uses HMA module and container names, defaults to a five-experiment cap,
+removes operational comments and remote catalog links, and disables telemetry.
+These changes are packaging and anonymity adaptations. External model CLIs,
+the upstream MLE-bench grader, and task data retain their own licenses and are
+not included. The Apache License 2.0 text is in LICENSE; the MIT License text
+for src/hma/benchmark is in src/hma/benchmark/LICENSE.

README.md

@@ -25,8 +25,11 @@ model of this process and evaluates the workflow on 75 MLE-bench tasks.
 - [Installation](#installation)
 - [Reproducing the Experiments](#reproducing-the-experiments)
 - [Dataset and Models](#dataset-and-models)
+- [Use HMA with humanize](#use-hma-with-humanize)
 - [Citation](#citation)
 - [Acknowledgments](#acknowledgments)
+- [Maintainers](#maintainers)
+- [License](#license)
 
 ## HMA
 
@@ -93,8 +96,8 @@ to the paper's tables and figures.
 Clone the repository and install the Python dependencies:
 
 ```bash
-git clone https://github.com/humanfia/humanize-mle-agents.git
-cd humanize-mle-agents
+git clone https://github.com/humanfia/hma.git
+cd hma
 python3.12 -m venv .venv
 source .venv/bin/activate
 pip install -r requirements-repro.lock
@@ -227,9 +230,18 @@ multi-node execution remain unvalidated here. See the [validation record](docs/v
 for the completed checks and remaining limitations. [AGENTS.md](AGENTS.md) provides
 the maintenance and execution runbook.
 
+## Use HMA with humanize
+
+This repository reproduces the paper. To use HMA's fixed-*k* alternation in your
+own work, use [fixed-interrupt-flame-chase-flow](https://github.com/humanfia/fixed-interrupt-flame-chase-flow),
+which brings it to [humanize](https://github.com/humanfia/humanize) as a flow.
+See the [humanize documentation](https://docs.humanfia.ai/humanize/) for running flows.
+
 ## Citation
 
-The arXiv link and citation will be added with the paper release.
+The arXiv link and citation will be added with the paper release. Until then,
+cite this repository with [`CITATION.cff`](CITATION.cff) (GitHub's "Cite this
+repository").
 
 ## Acknowledgments
 
@@ -238,3 +250,18 @@ This repository builds on [MLE-bench](https://github.com/openai/mle-bench),
 [MLEvolve](https://github.com/InternScience/MLEvolve), and
 [ScienceFlow](https://github.com/science-learner/ScienceFlow).
 We thank their authors for releasing the code and benchmark resources.
+
+## Maintainers
+
+[@antoinegg1](https://github.com/antoinegg1).
+
+Issues and pull requests are welcome; see the humanfia
+[contributing guide](https://github.com/humanfia/.github/blob/main/CONTRIBUTING.md).
+
+## License
+
+`src/hma` and `src/hmz` are licensed under [Apache-2.0](LICENSE);
+`src/hma/benchmark` is licensed under [MIT](src/hma/benchmark/LICENSE)
+(SPDX: `Apache-2.0 AND MIT`). See [NOTICE](NOTICE). External model CLIs, the
+upstream MLE-bench grader, and task data retain their own licenses and are not
+included.

docs/provenance.md

@@ -38,8 +38,9 @@ upstream trees or pretrained weights; builds fetch the pinned revisions and
 verify their allowlisted files. Consult each source's original license for use.
 The adapter provenance source does not carry a separate license grant in the
 imported folder; confirm the authors' redistribution terms when publishing it.
-The existing repository [LICENSE](../LICENSE) is preserved rather than assigning
-new third-party ownership or licensing terms.
+The repository's own terms ([LICENSE](../LICENSE), [NOTICE](../NOTICE) and
+[src/hma/benchmark/LICENSE](../src/hma/benchmark/LICENSE)) are preserved rather
+than assigning new third-party ownership or licensing terms.
 
 ## Hardware and protocol provenance
 

pyproject.toml

@@ -9,7 +9,7 @@ description = "HMA: capped alternation of native agents for long-horizon machine
 readme = "README.md"
 requires-python = ">=3.12"
 license = "Apache-2.0 AND MIT"
-license-files = ["LICENSE"]
+license-files = ["LICENSE", "NOTICE", "src/hma/benchmark/LICENSE"]
 dependencies = ["docker>=7.1,<8", "pydantic>=2.13,<3", "requests>=2.32,<3", "pyyaml>=6", "python-dotenv>=1.2", "matplotlib>=3.9", "uv==0.12.9"]
 
 [project.optional-dependencies]

src/hma/benchmark/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) HMA contributors (identity omitted for anonymous review)
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.