PaperSpine5 skill
The paper-spine skill takes a topic, materials and data through a scientific paper: literature, text, figures, review and Word, LaTeX and PDF files
- 6,035 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The installer downloads and runs a full suite with its own Python and a local web server, and the skill updates it itself on every call
- The skill reads the folder with research materials, which can hold private data and unpublished results
- The agent writes the paper files and the check scripts run on your machine
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/paperspine5Detects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The installer downloads and runs a full suite with its own Python and a local web server, and the skill updates it itself on every call
- The skill reads the folder with research materials, which can hold private data and unpublished results
- The agent writes the paper files and the check scripts run on your machine
This is third-party code. Review the repository files before installing.
What it does
PaperSpine5 carries a scientific paper in a single task: from a research direction and materials to finished files. The `paper-spine` skill starts it, or the `/paperspine` command. It first opens a web workspace on the local machine where you set options and choose materials, then the host agent searches and reads literature, builds the contribution and outline, writes the text, prepares figures (data plots, mechanism diagrams, method overview diagrams), checks references and makes revisions. The same workspace serves for previews, choosing variants, feedback and downloading the result. Scholarly judgment stays with the agent, while the workspace and scripts handle storage, rendering and delivery. The skill's rules forbid inventing data, metrics, citations, figures, methods, ethics and funding, and treat the user's materials as the authority for claims about the study. Three research subagents study strong papers and the scene on the topic, three reviewers check evidence, method and writing. The checking tools include the scripts `artifact_check.py`, `reference_inventory.py`, `citation_bank_check.py`, `latex_guard.py` and `word_guard.py`. The output is editable Word or LaTeX sources and a PDF for the chosen venue. References are kept with claims tied to sources, and corrections go into the same task and version history, with no second copy. Every skill call first checks the update channel and, if a new version is out, replaces the suite, the web part and the updater itself while keeping task data.
Who it is for. For researchers, graduate students and report authors who write a paper from their own data in Codex or Claude Code and want venue-ready files together with figures and verified references.
Good fit when
- You have a direction, experiments or notes and need a paper with text, plots and diagrams in one task
- A paper must be prepared for a specific venue with LaTeX or Word sources and a final PDF
- You need a survey of strong papers on the topic and every reference checked against its source, not a list from the model's memory
- A draft needs independent review of evidence, method and writing before submission
- The research materials are private and should stay on your machine by default
Not a fit when
- You want to install from one small archive: a single-skill version is no longer released, only the full suite of 26 to 56 MB for your platform
- You work on Linux with an ARM processor or on Alpine: the author does not claim support for Linux arm64 and musl
- You want the service to submit the paper to a journal, pay for publication or write to an editor: the product does none of this and sends nothing without your explicit permission
- You have no verifiable materials or data: the rules forbid inventing results, so an empty request will not produce a paper
Example request
Run paper-spine on my experiments folder and prepare a machine learning conference paper with plots and LaTeX textLimitations
Version 0.4.0-alpha.3 is a prerelease without a separate cryptographic signature. The installer verifies archive size and SHA-256, but the macOS packages are not signed or notarized, so on first launch the system may ask for explicit permission. Installation needs the full suite for your system: Windows x64, Linux glibc x86_64, macOS on Apple Silicon or Intel.
On every call the skill checks for updates and installs a new version without your involvement unless you explicitly turned auto-update off; offline it continues with the existing installation. The profile and task data live in `~/.paperspine5/profiles/default` and are not erased when the skill is removed. The author stresses that a successful tool run, hashes and task status do not prove scientific, editorial or visual readiness of the paper: judge the result by the text itself. Voluntary support of the author unlocks no features. MIT license.
How to disable. Close the host and delete the `paper-spine` folder from `~/.codex/skills` or `~/.claude/skills`. Task data in `~/.paperspine5` stays, delete it separately if you no longer need it. Auto-update can be turned off with an explicit setting, see `UPDATE.md`, and rollback is described in `ROLLBACK.md`.
FAQ
What is in the full suite and why is there no small archive?
The suite for your platform includes its own Python, the web workspace, the `paper-spine` skill itself and updating with rollback. The old 0.70 MB archive without the web part and runtime is not suitable for a first install and does not repair a lost suite pointer: without the web workspace a new task cannot be started.
Where do my materials go?
Task files are stored locally by default. Reading public literature is part of the work. Uploading private materials, submitting the paper, paying and external contact each need separate explicit permission that neither task status nor a product release replaces. The texts themselves are read by your host agent's model, so its data handling rules apply here too.
How do I move from versions V3 and V4?
Run the installer with `--clean-legacy` (`-CleanLegacy` on Windows). It first previews what it found and archives the known skill folders of earlier versions only on a conflict. Task data, host settings and unknown files are not deleted.
Can it be installed offline?
Yes, with two files: the manifest and the suite archive for your platform. On Windows pass `-ManifestPath` and `-BundlePath`, on macOS and Linux the `--manifest` and `--bundle` flags. A single archive without a manifest still goes online for the current manifest.
Which hosts are supported?
The installer accepts the target `codex`, `claude-code` or `both`, and the repository holds skill projections for Codex, Claude and OpenClaw. A `/paperspine` command is added for Claude Code.
Discussion
No questions or reviews yet.
Related
SkillsA skill that stress-tests a research paper before submission: mocks a review, gives a verdict, helps revise and re-verify
Eight skills for finished papers and theses in LaTeX and Typst: compile, references, language polish, cover letters and a reviewer-style audit
A skill set for academic papers: deep method analysis, visual slides and a coherent write-up from a paper PDF
A skill for writing and revising ML, CV and NLP papers: section structure, paragraph flow and reviewer-style self-review
Takes an agent from a topic to an arXiv review paper in LaTeX IEEEtran: approved plan, every citation checked against a live source, PDF build
A skills pack for writing a research paper: a multi-agent pipeline from outline and plots to literature review and refinement, with no LLM keys
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Repository stars | Main difference |
|---|---|---|---|
| PaperSpine5 (this entry) | no | 6,035 | |
| PaperOrchestra | no | 684 | a five-step multi-agent pipeline for writing a paper, without its own web part or an installer with updates |
| Academic Research Skills | no | 53,488 | four skills for literature review, writing, peer review and a pipeline, delivered as plain skills with no suite |
README in short
The description exists in nine languages, Russian included. It opens with the promise to go from nothing to a paper with figures and explains the working cycle with the web workspace, then lists download sizes for four platforms, the commands to install and check for updates, the support boundaries and the repository layout. The repository holds detailed notes for three prereleases, instructions for installation, updating, rollback and removal, the skill source, host projections, tests and the product site.
README badge
Original SKILL.md
--- name: paper-spine description: Research, write, review and deliver evidence-bound papers in one task, with user choices, real files, editable outputs and same-task revision. # Suite Update Authority; Blocks unsupported readiness claims. --- # PaperSpine Orchestrator Before paper production or resume, read [paper-spine-production-protocol.md](references/paper-spine-production-protocol.md). It is the operational 17-step handoff for the methods below. The current host Agent performs research, citation checking, interpretation, writing, figures and revision, and arranges independent review. Web, local files and small scripts support interaction, persistence, rendering and delivery. When configuration is missing, launch the intake UI automatically before inspecting materials; do not hand-write configuration or silently choose a paper root. For a visible launcher action, use `require_escalated` as required by the host. After the update preflight below, launch is the first paper-production action when configuration is missing. **Product outcome:** an evidence-bound, coherent paper with informative figures, applicable venue format and editable sources. Learn from strong same-field and venue papers. Let users configure, choose contribution/figures, preview, download and revise the same task. Tool success, hashes and task status do not establish scientific, editorial, visual or submission readiness; assess the actual paper. ## Authority and boundaries 1. The host Agent owns scholarly judgment and the current paper files. 2. The user's materials/results are authoritative for claims about this study. Never invent data, metrics, p-values, citations, figures, methods, authors, ethics, funding, permissions or review conclusions. 3. Use one task ID, one material/workspace binding and one continuous revision history. Never create a second task because a launcher, profile or transport failed. Never silently switch to another directory, profile, task or paper. 4. Web is the supported path for configuration, user choices, previews, downloads and feedback. It supports the configuration and choice boundaries below; existing confirmed scope permits independent local corrections during a Web outage. 5. Task status never authorizes uploads, publication, submission, payment, licensing, telemetry or external contact; obtain explicit authorization. 6. Private materials stay local unless explicitly authorized for sharing. Apply the saved research mode as described below; public literature reading is normal. ## Update once on invocation, then continue the paper Before starting or resuming paper production, run the installed Skill's `scripts/paperspine_update.py --preflight --yes` with its verified Python. This checks the official channel on every invocation: if current, continue; if newer, transactionally update the Skill, Web/runtime and updater together.