claimset 0.40.0__tar.gz

claimset-0.40.0/LICENSE

@@ -0,0 +1,201 @@
+                                 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 warranties 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
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

claimset-0.40.0/PKG-INFO

@@ -0,0 +1,198 @@
+Metadata-Version: 2.4
+Name: claimset
+Version: 0.40.0
+Summary: Tools for Claimset, an open format for research papers written by and for LLMs: validation, lab records and export.
+Author: Claimset contributors
+License-Expression: Apache-2.0
+Project-URL: Homepage, https://github.com/entrylib/claimset
+Project-URL: Issues, https://github.com/entrylib/claimset/issues
+Keywords: research,scholarly communication,LLM,agents,claims,reproducibility
+Classifier: Programming Language :: Python :: 3
+Classifier: Operating System :: OS Independent
+Classifier: Topic :: Scientific/Engineering
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Dynamic: license-file
+
+# Claimset
+
+Claimset is an open format for research papers written by AI agents for AI readers, and the tools that go with it.
+
+A claimset is a JSON file of atomic, self-contained **claims**, each with its scope, assumptions, limitations, kind and strength of support, and links to the evidence, results and artefacts behind it. Agents write the papers; a human supervisor and guarantor answer for them. Humans read them by asking an LLM.
+
+Research is iterative, so the format has a companion: the **lab record**, an append-only, hash-chained log of questions, hypotheses, plans, runs, results and decisions kept while the research happens. The claimset is distilled from it, negative results included.
+
+> Status: early draft. The format and tools are changing quickly; expect breaking changes before 1.0.
+
+## What is in this repository
+
+| Path | Contents |
+|---|---|
+| `spec/` | The claimset specification and the lab record specification (normative) |
+| `src/claimset/` | The `claimset` Python package and command-line tool |
+| `skills/` | Instructions for AI agents (Agent Skills), bundling everything they need |
+| `agents/` | What `claimset init` writes for coding agents: the instructions block and the guard's plugin |
+| `examples/` | Valid example claimsets |
+| `tests/` | Test suite |
+| `tools/` | Maintenance scripts |
+
+## Install
+
+```
+pip install claimset                                   # a release, from PyPI
+pip install git+https://github.com/entrylib/claimset    # the development version
+```
+
+The package has no dependencies beyond the Python standard library (Python 3.10 or later). It carries the specifications and the agent skills of its version:
+
+```
+claimset spec                  # the claimset specification this version validates against
+claimset spec --lab            # the lab record specification
+claimset spec --revision       # the revision specification: reviews, responses, revised versions, scores
+claimset init                  # set the project up for the coding agent running this shell (see below)
+claimset skill install         # install or update the skills in .claude/skills (--dir for another place)
+claimset skill check           # are the installed skills this version's?
+```
+
+The tools and skills work with any coding agent. `claimset init` sets a project up for the ones its people use, from one template:
+
+```
+claimset init --agent codex --agent gemini --agent claude    # or --agent all; --list shows every agent
+claimset init --agent opencode --guard                        # also the lab record's guard, where the agent has hooks
+```
+
+It installs the skills in the fewest folders that reach every agent chosen (`.agents/skills/` for Codex, Gemini CLI, Qwen Code, opencode, Kilo Code, GitHub Copilot, Cursor, Windsurf, Amp, Goose, Factory, Crush, Zed and MiniMax Code; `.claude/skills/` for Claude Code; their own for Junie, Kiro and Deep Code), writes one instructions block to `AGENTS.md` that `CLAUDE.md` and `GEMINI.md` import, and with `--guard` registers the guard in each agent's hook configuration (Claude Code, Codex, Gemini CLI, Qwen Code, opencode, GitHub Copilot, Cursor, Cline). Running it again updates what it wrote and leaves the rest of those files alone. Any model works: DeepSeek, MiniMax, Qwen or Kimi through opencode, Kilo Code, Crush or another harness are set up the same way, and the harness an agent runs in is recorded with it (lab record §5.2). An agent that cannot tell which model it is takes it from `CLAIMSET_MODEL`, which the human sets.
+
+Studies and papers live in the research repository, each study on its own branch `claimset/<study>` that is never merged with the code. The tool checks it out at `claimset/<study>/` (ignored by the code branches), and commits and pushes it after every change to the record (see the lab record specification §3):
+
+```
+claimset setup --study <study> --title …   # new study: branch claimset/<study>, checked out at claimset/<study>/
+claimset setup                             # in a fresh clone: check out every study branch
+claimset setup --ci                        # GitHub check: pull requests never carry study records
+claimset studies                           # the studies and their state
+claimset/<study>/lab/                      # the lab record, on the study branch
+claimset/<study>/<paper>.claimset.json     # a paper drawn from it   (claimset lab export [--paper <paper>])
+claimset-code/<study>/                     # code only this study uses, on the code branches
+claimset tag claimset/<study>/<paper>.claimset.json    # mark a submitted version: git tag claimset/<study>/<paper>/v<n>
+```
+
+Skills copied into a project do not follow later versions of the package; `claimset skill check` finds an out-of-date copy and `claimset skill install` replaces it.
+
+## Use
+
+Validate a claimset before submission:
+
+```
+claimset validate claimset.json
+claimset validate claimset.json --json        # machine-readable report
+claimset validate claimset.json --published   # check a published file
+claimset render claimset.json -o claimset.txt  # the reading view: plain text for reading models
+claimset renumber claimset.json                # before the first submission: IDs in reading order
+claimset migrate claimset.json                 # bring a file from an earlier version; lists what needs judgement
+claimset package claimset.json -o paper.zip    # the submission package: the claimset and its files (BagIt)
+claimset validate paper.zip                    # validate a package: files, checksums and the claimset
+```
+
+The report lists errors (the file is rejected), warnings (accepted, recorded) and the verifiability the platform will compute for each claim.
+
+Submit to EntryLib and follow the submission:
+
+```
+claimset submit paper.zip                      # a package (zip or folder) or a claimset JSON file; prints the consent link
+claimset submit https://zenodo.org/…/paper.zip # a package deposited elsewhere: the platform fetches it
+claimset status sub_…                          # pending, confirmed, publishing, published, refused or expired
+```
+
+A local file is validated first and not sent if it has errors. The answer gives a consent link for the guarantor, who signs in with ORCID within 48 hours and confirms; it is shown only once. A new version of a published claimset sets `paper.id` to its identifier. `--portal` (or `$CLAIMSET_PORTAL`) names another platform.
+
+Keep a lab record while doing research:
+
+```
+claimset lab init --title "Working title"
+claimset lab actor --kind agent --name "Research agent" --model "Anthropic|Claude Opus|5.5"   # the harness is recorded too
+export CLAIMSET_ACTOR=AC1
+claimset lab question "Does X improve Y?"
+claimset lab hypothesis "X improves Y on monthly series." --question Q1
+claimset lab plan --kind confirmatory --hypotheses H1 --design "…" --analysis "…" \
+    --framework frequentist --parameter alpha=0.05 --parameter test="paired t-test" --parameter sides=greater \
+    --parameter correction=none --parameter sample="400 series, fixed" \
+    --quantity p="p-value" --quantity d="mean difference" --supported "p < alpha and d > 0" --not-supported "…"
+claimset lab run --plan PL1 --seed 7 -- python3 experiment.py
+claimset lab verdict PL1     # the verdict the plan's rule gives on the run's results
+claimset lab status H1 supported --plan PL1 --why "Rule of PL1 met." --evidence RS1 RS2
+claimset lab summary     # current state, for whoever picks up the work
+claimset lab verify      # hash chain, rules and run files
+```
+
+A confirmatory plan fixes everything that decides the outcome before any run: the test, significance level, sidedness, correction for multiple tests, sample size and stopping rule, the results the decision is made on, and the rules for *supported* and *not supported* (anything else is *inconclusive*). The tool computes the verdict from those rules and refuses a status that disagrees or a second attempt at a decided test, and every planned test that ran is exported as a claim, whatever its result. Exploration stays light: record the runs you learn from; nothing else is enforced. Weaknesses that do not break a rule are flagged rather than refused: code changed after the plan, held-out data used elsewhere, a test that reproduced the result that suggested its hypothesis; `plan`, `run` and `verdict` also warn about a hypothesis from exploration without `--new-data`, exclusions without counts and the same computation run under two plans.
+
+`run` records the command, code version, environment and every output file, and turns the numbers the script writes to `$CLAIMSET_RESULTS` into result events. The log is append-only and hash-chained; the tool refuses writes that break the rules (e.g. confirming a hypothesis with the result that suggested it).
+
+When the research is done, draft the claimset from the record:
+
+```
+claimset lab export -o claimset.json
+```
+
+The draft carries everything the record states: a claim for every tested hypothesis (negative results included) with its outcome and whether the test was planned, evidence with the runs and results behind it, contributors, models and the exploration counts. Export also prints a completion list of what only people or the writing agent can decide: licence, guarantor, principal claims, the highlights a reader sees first (up to five, chosen by the guarantor), wording and scope. The `claimset-submit` skill works through it.
+
+Several agents working at once each keep their own log, merged when they finish:
+
+```
+claimset lab start --by AC2     # AC2's events now go to lab/agents/AC2.jsonl, with IDs like AC2.H1
+claimset lab start --by AC3
+claimset lab merge              # appends both logs to the main log, interleaved by time
+```
+
+An agent sees the main log as it was when it started, plus its own events. The merge is refused as a whole if any event no longer fits the main log, and `verify` checks that every merged event matches its agent log.
+
+### Answering a review
+
+A published claimset is reviewed, and every comment states in advance what evidence would settle it. The author may answer with a revised version and a response; reviews, responses and versions are kept side by side, never rewritten, and the reviewer's score of each claim rests on comments the author can answer. The commands read the claimset's record folder (`v1/claimset.json`, `v1/reviews/RV1.json`, …):
+
+```
+claimset revision status <record>                    # comments, their states and tests; what to answer
+claimset lab plan --kind confirmatory --hypotheses H9 --comment doi:…@v1/RV1.c2 --review RV1.json --analysis "…"
+                                                     # run a comment's resolution test as a planned test
+claimset revision respond <record> --revision revised.json --lab lab.jsonl -o response.json
+claimset revision check <record> --revision revised.json --response response.json --lab lab.jsonl
+```
+
+### Protecting the lab record
+
+The hash chain makes any edit to the log detectable. Three optional layers stop edits from happening or make them impossible to hide (spec §4.5):
+
+```
+claimset lab anchor --git     # push the log to a `claimset-lab` branch after every plan
+claimset lab verify --remote  # check the local log against the anchored copy
+claimset lab protect          # append-only attribute on the log files (Linux, needs sudo)
+```
+
+- **Anchoring** commits the log to its own branch without touching your working tree and never force-pushes. A confirmatory run is refused until its plan has been anchored, so the plan provably existed before the results. Protect the branch on your git host against force-pushes and deletion to make the remote copy append-only.
+- **The guard** is a hook for agent harnesses that refuses tool calls that would edit or delete files in the lab directory other than through `claimset lab`. `claimset init --guard --agent NAME` installs it for Claude Code, Codex, Gemini CLI, Qwen Code, opencode, GitHub Copilot, Cursor or Cline (`claimset lab guard --print-config --agent NAME` prints the configuration instead). Codex, Gemini CLI, Qwen Code and Copilot run a project's hooks only once you trust them.
+- **The append-only attribute** makes the kernel refuse anything but appending, as long as the agent cannot use `sudo`.
+
+## For AI agents
+
+`skills/claimset-submit/` teaches an agent to turn finished research, or a lab record, into a valid claimset submission. It is self-contained: `claimset init` installs it where your agent reads skills, or copy the folder there yourself. Agents without skills support are pointed at its `SKILL.md` by the block `claimset init` writes to `AGENTS.md`.
+
+`skills/claimset-lab/` teaches an agent to keep a lab record while it does research: registering itself honestly, planning before testing, running code through the tool, recording failures and negative results, working alongside other agents, and handing over to `claimset-submit` at the end. It bundles a copy of the `claimset` package, so it works without installing anything (`python3 scripts/claimset lab ...`).
+
+`skills/claimset-revise/` teaches an agent to answer the reviews of a published claimset: settling each comment with evidence (running its resolution test as a planned test where it can be computed), building the revised version and writing the response. It bundles the package too.
+
+`skills/claimset-review/` is the platform's reviewer, published so that every author can see how a claimset is reviewed and review their own before submitting: what to check, how to write a comment whose resolution test either side could run, how to judge an author's answers, and how every score below 100 rests on comments. Each review names the version of the skill it followed.
+
+## Contributing
+
+- The specifications in `spec/` are the source of truth; the validator implements them and the tests check both.
+- After changing the validator, a spec, an example or a template in `agents/`, run `python3 tools/sync_skills.py` to update the copies bundled in the skills and the package.
+- Run the tests with `python3 -m unittest discover -s tests`.
+- Releasing, in two stages:
+  1. Set the version in `pyproject.toml` and `src/claimset/__init__.py` and merge. Then run the `build` workflow by hand on `main` (Actions > build > Run workflow): it runs the tests, builds the sdist and wheel, tries the wheel in a clean environment and drafts the GitHub release `v<version>` for that commit, with both files attached. Run again while it is a draft, it replaces the draft; a published version is never rebuilt.
+  2. Review the draft and publish it. Publishing creates the tag `v<version>` on the commit that was built, and the `release` workflow checks that the files are the ones `build` made from that commit and uploads them to PyPI by trusted publishing (no token; set up once on PyPI and as a `pypi` environment on GitHub, see `.github/workflows/release.yml`).
+- Planned work is tracked in [issues labelled `roadmap`](https://github.com/entrylib/claimset/issues?q=is%3Aissue+label%3Aroadmap).
+
+## Licence
+
+Code is licensed under the Apache License 2.0 (`LICENSE`). The specifications in `spec/` are licensed under Creative Commons Attribution 4.0 International (CC BY 4.0).

claimset-0.40.0/README.md

@@ -0,0 +1,181 @@
+# Claimset
+
+Claimset is an open format for research papers written by AI agents for AI readers, and the tools that go with it.
+
+A claimset is a JSON file of atomic, self-contained **claims**, each with its scope, assumptions, limitations, kind and strength of support, and links to the evidence, results and artefacts behind it. Agents write the papers; a human supervisor and guarantor answer for them. Humans read them by asking an LLM.
+
+Research is iterative, so the format has a companion: the **lab record**, an append-only, hash-chained log of questions, hypotheses, plans, runs, results and decisions kept while the research happens. The claimset is distilled from it, negative results included.
+
+> Status: early draft. The format and tools are changing quickly; expect breaking changes before 1.0.
+
+## What is in this repository
+
+| Path | Contents |
+|---|---|
+| `spec/` | The claimset specification and the lab record specification (normative) |
+| `src/claimset/` | The `claimset` Python package and command-line tool |
+| `skills/` | Instructions for AI agents (Agent Skills), bundling everything they need |
+| `agents/` | What `claimset init` writes for coding agents: the instructions block and the guard's plugin |
+| `examples/` | Valid example claimsets |
+| `tests/` | Test suite |
+| `tools/` | Maintenance scripts |
+
+## Install
+
+```
+pip install claimset                                   # a release, from PyPI
+pip install git+https://github.com/entrylib/claimset    # the development version
+```
+
+The package has no dependencies beyond the Python standard library (Python 3.10 or later). It carries the specifications and the agent skills of its version:
+
+```
+claimset spec                  # the claimset specification this version validates against
+claimset spec --lab            # the lab record specification
+claimset spec --revision       # the revision specification: reviews, responses, revised versions, scores
+claimset init                  # set the project up for the coding agent running this shell (see below)
+claimset skill install         # install or update the skills in .claude/skills (--dir for another place)
+claimset skill check           # are the installed skills this version's?
+```
+
+The tools and skills work with any coding agent. `claimset init` sets a project up for the ones its people use, from one template:
+
+```
+claimset init --agent codex --agent gemini --agent claude    # or --agent all; --list shows every agent
+claimset init --agent opencode --guard                        # also the lab record's guard, where the agent has hooks
+```
+
+It installs the skills in the fewest folders that reach every agent chosen (`.agents/skills/` for Codex, Gemini CLI, Qwen Code, opencode, Kilo Code, GitHub Copilot, Cursor, Windsurf, Amp, Goose, Factory, Crush, Zed and MiniMax Code; `.claude/skills/` for Claude Code; their own for Junie, Kiro and Deep Code), writes one instructions block to `AGENTS.md` that `CLAUDE.md` and `GEMINI.md` import, and with `--guard` registers the guard in each agent's hook configuration (Claude Code, Codex, Gemini CLI, Qwen Code, opencode, GitHub Copilot, Cursor, Cline). Running it again updates what it wrote and leaves the rest of those files alone. Any model works: DeepSeek, MiniMax, Qwen or Kimi through opencode, Kilo Code, Crush or another harness are set up the same way, and the harness an agent runs in is recorded with it (lab record §5.2). An agent that cannot tell which model it is takes it from `CLAIMSET_MODEL`, which the human sets.
+
+Studies and papers live in the research repository, each study on its own branch `claimset/<study>` that is never merged with the code. The tool checks it out at `claimset/<study>/` (ignored by the code branches), and commits and pushes it after every change to the record (see the lab record specification §3):
+
+```
+claimset setup --study <study> --title …   # new study: branch claimset/<study>, checked out at claimset/<study>/
+claimset setup                             # in a fresh clone: check out every study branch
+claimset setup --ci                        # GitHub check: pull requests never carry study records
+claimset studies                           # the studies and their state
+claimset/<study>/lab/                      # the lab record, on the study branch
+claimset/<study>/<paper>.claimset.json     # a paper drawn from it   (claimset lab export [--paper <paper>])
+claimset-code/<study>/                     # code only this study uses, on the code branches
+claimset tag claimset/<study>/<paper>.claimset.json    # mark a submitted version: git tag claimset/<study>/<paper>/v<n>
+```
+
+Skills copied into a project do not follow later versions of the package; `claimset skill check` finds an out-of-date copy and `claimset skill install` replaces it.
+
+## Use
+
+Validate a claimset before submission:
+
+```
+claimset validate claimset.json
+claimset validate claimset.json --json        # machine-readable report
+claimset validate claimset.json --published   # check a published file
+claimset render claimset.json -o claimset.txt  # the reading view: plain text for reading models
+claimset renumber claimset.json                # before the first submission: IDs in reading order
+claimset migrate claimset.json                 # bring a file from an earlier version; lists what needs judgement
+claimset package claimset.json -o paper.zip    # the submission package: the claimset and its files (BagIt)
+claimset validate paper.zip                    # validate a package: files, checksums and the claimset
+```
+
+The report lists errors (the file is rejected), warnings (accepted, recorded) and the verifiability the platform will compute for each claim.
+
+Submit to EntryLib and follow the submission:
+
+```
+claimset submit paper.zip                      # a package (zip or folder) or a claimset JSON file; prints the consent link
+claimset submit https://zenodo.org/…/paper.zip # a package deposited elsewhere: the platform fetches it
+claimset status sub_…                          # pending, confirmed, publishing, published, refused or expired
+```
+
+A local file is validated first and not sent if it has errors. The answer gives a consent link for the guarantor, who signs in with ORCID within 48 hours and confirms; it is shown only once. A new version of a published claimset sets `paper.id` to its identifier. `--portal` (or `$CLAIMSET_PORTAL`) names another platform.
+
+Keep a lab record while doing research:
+
+```
+claimset lab init --title "Working title"
+claimset lab actor --kind agent --name "Research agent" --model "Anthropic|Claude Opus|5.5"   # the harness is recorded too
+export CLAIMSET_ACTOR=AC1
+claimset lab question "Does X improve Y?"
+claimset lab hypothesis "X improves Y on monthly series." --question Q1
+claimset lab plan --kind confirmatory --hypotheses H1 --design "…" --analysis "…" \
+    --framework frequentist --parameter alpha=0.05 --parameter test="paired t-test" --parameter sides=greater \
+    --parameter correction=none --parameter sample="400 series, fixed" \
+    --quantity p="p-value" --quantity d="mean difference" --supported "p < alpha and d > 0" --not-supported "…"
+claimset lab run --plan PL1 --seed 7 -- python3 experiment.py
+claimset lab verdict PL1     # the verdict the plan's rule gives on the run's results
+claimset lab status H1 supported --plan PL1 --why "Rule of PL1 met." --evidence RS1 RS2
+claimset lab summary     # current state, for whoever picks up the work
+claimset lab verify      # hash chain, rules and run files
+```
+
+A confirmatory plan fixes everything that decides the outcome before any run: the test, significance level, sidedness, correction for multiple tests, sample size and stopping rule, the results the decision is made on, and the rules for *supported* and *not supported* (anything else is *inconclusive*). The tool computes the verdict from those rules and refuses a status that disagrees or a second attempt at a decided test, and every planned test that ran is exported as a claim, whatever its result. Exploration stays light: record the runs you learn from; nothing else is enforced. Weaknesses that do not break a rule are flagged rather than refused: code changed after the plan, held-out data used elsewhere, a test that reproduced the result that suggested its hypothesis; `plan`, `run` and `verdict` also warn about a hypothesis from exploration without `--new-data`, exclusions without counts and the same computation run under two plans.
+
+`run` records the command, code version, environment and every output file, and turns the numbers the script writes to `$CLAIMSET_RESULTS` into result events. The log is append-only and hash-chained; the tool refuses writes that break the rules (e.g. confirming a hypothesis with the result that suggested it).
+
+When the research is done, draft the claimset from the record:
+
+```
+claimset lab export -o claimset.json
+```
+
+The draft carries everything the record states: a claim for every tested hypothesis (negative results included) with its outcome and whether the test was planned, evidence with the runs and results behind it, contributors, models and the exploration counts. Export also prints a completion list of what only people or the writing agent can decide: licence, guarantor, principal claims, the highlights a reader sees first (up to five, chosen by the guarantor), wording and scope. The `claimset-submit` skill works through it.
+
+Several agents working at once each keep their own log, merged when they finish:
+
+```
+claimset lab start --by AC2     # AC2's events now go to lab/agents/AC2.jsonl, with IDs like AC2.H1
+claimset lab start --by AC3
+claimset lab merge              # appends both logs to the main log, interleaved by time
+```
+
+An agent sees the main log as it was when it started, plus its own events. The merge is refused as a whole if any event no longer fits the main log, and `verify` checks that every merged event matches its agent log.
+
+### Answering a review
+
+A published claimset is reviewed, and every comment states in advance what evidence would settle it. The author may answer with a revised version and a response; reviews, responses and versions are kept side by side, never rewritten, and the reviewer's score of each claim rests on comments the author can answer. The commands read the claimset's record folder (`v1/claimset.json`, `v1/reviews/RV1.json`, …):
+
+```
+claimset revision status <record>                    # comments, their states and tests; what to answer
+claimset lab plan --kind confirmatory --hypotheses H9 --comment doi:…@v1/RV1.c2 --review RV1.json --analysis "…"
+                                                     # run a comment's resolution test as a planned test
+claimset revision respond <record> --revision revised.json --lab lab.jsonl -o response.json
+claimset revision check <record> --revision revised.json --response response.json --lab lab.jsonl
+```
+
+### Protecting the lab record
+
+The hash chain makes any edit to the log detectable. Three optional layers stop edits from happening or make them impossible to hide (spec §4.5):
+
+```
+claimset lab anchor --git     # push the log to a `claimset-lab` branch after every plan
+claimset lab verify --remote  # check the local log against the anchored copy
+claimset lab protect          # append-only attribute on the log files (Linux, needs sudo)
+```
+
+- **Anchoring** commits the log to its own branch without touching your working tree and never force-pushes. A confirmatory run is refused until its plan has been anchored, so the plan provably existed before the results. Protect the branch on your git host against force-pushes and deletion to make the remote copy append-only.
+- **The guard** is a hook for agent harnesses that refuses tool calls that would edit or delete files in the lab directory other than through `claimset lab`. `claimset init --guard --agent NAME` installs it for Claude Code, Codex, Gemini CLI, Qwen Code, opencode, GitHub Copilot, Cursor or Cline (`claimset lab guard --print-config --agent NAME` prints the configuration instead). Codex, Gemini CLI, Qwen Code and Copilot run a project's hooks only once you trust them.
+- **The append-only attribute** makes the kernel refuse anything but appending, as long as the agent cannot use `sudo`.
+
+## For AI agents
+
+`skills/claimset-submit/` teaches an agent to turn finished research, or a lab record, into a valid claimset submission. It is self-contained: `claimset init` installs it where your agent reads skills, or copy the folder there yourself. Agents without skills support are pointed at its `SKILL.md` by the block `claimset init` writes to `AGENTS.md`.
+
+`skills/claimset-lab/` teaches an agent to keep a lab record while it does research: registering itself honestly, planning before testing, running code through the tool, recording failures and negative results, working alongside other agents, and handing over to `claimset-submit` at the end. It bundles a copy of the `claimset` package, so it works without installing anything (`python3 scripts/claimset lab ...`).
+
+`skills/claimset-revise/` teaches an agent to answer the reviews of a published claimset: settling each comment with evidence (running its resolution test as a planned test where it can be computed), building the revised version and writing the response. It bundles the package too.
+
+`skills/claimset-review/` is the platform's reviewer, published so that every author can see how a claimset is reviewed and review their own before submitting: what to check, how to write a comment whose resolution test either side could run, how to judge an author's answers, and how every score below 100 rests on comments. Each review names the version of the skill it followed.
+
+## Contributing
+
+- The specifications in `spec/` are the source of truth; the validator implements them and the tests check both.
+- After changing the validator, a spec, an example or a template in `agents/`, run `python3 tools/sync_skills.py` to update the copies bundled in the skills and the package.
+- Run the tests with `python3 -m unittest discover -s tests`.
+- Releasing, in two stages:
+  1. Set the version in `pyproject.toml` and `src/claimset/__init__.py` and merge. Then run the `build` workflow by hand on `main` (Actions > build > Run workflow): it runs the tests, builds the sdist and wheel, tries the wheel in a clean environment and drafts the GitHub release `v<version>` for that commit, with both files attached. Run again while it is a draft, it replaces the draft; a published version is never rebuilt.
+  2. Review the draft and publish it. Publishing creates the tag `v<version>` on the commit that was built, and the `release` workflow checks that the files are the ones `build` made from that commit and uploads them to PyPI by trusted publishing (no token; set up once on PyPI and as a `pypi` environment on GitHub, see `.github/workflows/release.yml`).
+- Planned work is tracked in [issues labelled `roadmap`](https://github.com/entrylib/claimset/issues?q=is%3Aissue+label%3Aroadmap).
+
+## Licence
+
+Code is licensed under the Apache License 2.0 (`LICENSE`). The specifications in `spec/` are licensed under Creative Commons Attribution 4.0 International (CC BY 4.0).

claimset-0.40.0/pyproject.toml

@@ -0,0 +1,33 @@
+[build-system]
+requires = ["setuptools>=77"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "claimset"
+version = "0.40.0"
+description = "Tools for Claimset, an open format for research papers written by and for LLMs: validation, lab records and export."
+readme = "README.md"
+license = "Apache-2.0"
+license-files = ["LICENSE"]
+requires-python = ">=3.10"
+authors = [{name = "Claimset contributors"}]
+keywords = ["research", "scholarly communication", "LLM", "agents", "claims", "reproducibility"]
+classifiers = [
+  "Programming Language :: Python :: 3",
+  "Operating System :: OS Independent",
+  "Topic :: Scientific/Engineering",
+]
+dependencies = []
+
+[project.urls]
+Homepage = "https://github.com/entrylib/claimset"
+Issues = "https://github.com/entrylib/claimset/issues"
+
+[project.scripts]
+claimset = "claimset.cli:main"
+
+[tool.setuptools.packages.find]
+where = ["src"]
+
+[tool.setuptools.package-data]
+claimset = ["data/spec/*.md", "data/spec/extensions/*.md", "data/examples/*.json", "data/skills/*/SKILL.md", "data/skills/claimset-lab/claimset", "data/agents/*"]

claimset-0.40.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

claimset-0.40.0/src/claimset/__init__.py

@@ -0,0 +1,12 @@
+"""Claimset: an open format for research papers written by and for LLMs."""
+
+__version__ = "0.40.0"
+
+#: Version of the claimset specification this package implements.
+SPEC_VERSION = "0.40"
+
+#: Version of the lab record specification this package implements.
+LAB_SPEC_VERSION = "0.21"
+
+#: Version of the revision specification (reviews, responses, revised versions and scores) this package implements.
+REVISION_SPEC_VERSION = "0.6"