{
"markdown": "# CiteVahti\n\n**Does the cited paper actually support the claim? Check every citation in your\nmanuscript — before a reviewer does.**\n\n
\n \n
\n\nCiteVahti walks you through your manuscript one claim at a time. You decide whether each\ncited source really backs the sentence; an optional AI gives a *blinded* second opinion\n(you rate first, it can never overrule you); and every decision is saved as a timestamped,\ntamper-evident record on your own computer.\n\n\n\n> **Free while in beta.** CiteVahti is local-first: your manuscript stays on your device\n> unless *you* choose to turn on an external AI model. No account. Nothing uploaded.\n>\n> **What beta means here:** the review workflow and your data format are stable, but the\n> tool is young and your feedback still shapes it. The app does **not update itself yet** —\n> when a new version ships you replace it yourself (steps under *Updating*, below; the\n> panel can check for a newer version, but only if you turn that on). If something is\n> confusing or judged wrong, [say so](#feedback--support) — beta reports are how it improves.\n\n## Who it's for\n\nResearchers, clinicians, and students who write papers and want to be sure every citation\nholds up — without learning new software or touching a command line. If you can open a file\non your computer, you can use CiteVahti.\n\n**What it does for you**\n\n- **Catches citations that don't support the sentence** they're attached to — the kind a\n reviewer or editor will flag.\n- **Keeps you in charge.** You make every call; the AI is only ever a second opinion, shown\n *after* you've decided.\n- **Leaves a clean paper trail** — a timestamped record of what you checked and concluded,\n ready for a supervisor, co-author, journal, or registry.\n- **Builds a reusable evidence map** so you don't re-do the same checks next project.\n- **Stays private by default** — everything lives in a folder on your Mac or PC.\n\n## Get started in 2 minutes — no terminal, ever\n\n**On a Mac, use the desktop app.** It's a normal app: download it, open it, review.\n\n1. Download **`citevahti--macos-arm64.app.zip`** from the\n [latest release](https://github.com/heidihelena/citevahti/releases/latest).\n2. Unzip it and drag **`CiteVahti.app`** into your **Applications** folder, then open it\n like any other app. *(It's signed and notarized — it opens with no security warnings.)*\n3. Choose a folder to keep your reviews in when it asks.\n4. Drag your manuscript (`.docx` or `.md`) onto the window. The panel walks you through the\n rest, one step at a time — there's nothing to memorise.\n\nWant an AI assistant to help pre-screen citations? Whenever you're ready, click the\n**CiteVahti menu-bar icon → Start Agent Server**. CiteVahti first explains exactly what an\nassistant can and cannot do — and either way, you rate every claim yourself and make every\nfinal call.\n\n### Or: use it inside Claude Desktop\n\nIf you already work in Claude, install CiteVahti as a one-click extension and drive the\nreview by chatting. *(This is also the path on Windows and Linux, where the desktop app\nisn't available yet.)*\n\n1. Download the extension **for your computer** from the\n [latest release](https://github.com/heidihelena/citevahti/releases/latest):\n - **macOS** (Apple Silicon) — `citevahti--macos-arm64.mcpb` *(signed + notarized)*\n - **Windows** — `citevahti--windows-x64.mcpb`\n - **Linux** — `citevahti--linux-x64.mcpb`\n2. **Double-click** the downloaded file to add it to Claude Desktop.\n3. Choose a folder for your reviews when asked.\n4. Tell your assistant: *\"Check the claims in my manuscript using CiteVahti.\"*\n\n> **First time on macOS** you may see a one-time security prompt — the extension is a zip,\n> which macOS can't pre-stamp even though the app inside is signed and notarized. If it\n> appears: **right-click the file → Open**, or go to **System Settings → Privacy & Security →\n> \"Open Anyway\"**. After that it installs and runs normally.\n\n> **First time on Windows** you may see a SmartScreen warning (\"Windows protected your PC\")\n> — the Windows build is unsigned, so Windows can't check who published it. If it appears:\n> click **More info**, then **Run anyway**. After that it installs and runs normally.\n\n\nUpdating — your review data is never touched; only the program is replaced\n\n- **Desktop app:** quit `CiteVahti.app`, download the newest\n `citevahti--macos-arm64.app.zip` from the\n [latest release](https://github.com/heidihelena/citevahti/releases/latest), and replace the\n old `CiteVahti.app` in Applications with it. Your reviews live in the project folder you\n chose, not inside the app, so nothing is lost.\n- **Claude Desktop extension:** remove the existing CiteVahti extension, fully quit and reopen\n Claude Desktop, then add the newest `.mcpb` from the\n [latest release](https://github.com/heidihelena/citevahti/releases/latest). Ask your\n assistant to run the **`status`** tool to confirm the running version. *(macOS Intel: install\n via `pip install \"citevahti[mcp]\"`, or run the Apple-Silicon `.mcpb` under Rosetta.)*\n- **pip:** `pip install --no-cache-dir --upgrade \"citevahti[mcp]\"`, then `pip show citevahti`\n to confirm the version.\n\n\n\n\nFor developers & technical readers — pip, CLI, architecture, security\n\nEverything below assumes a terminal; none of it is needed to use the app.\n\n**Try it on synthetic data** (nothing of yours touched — builds an invented ledger and\nopens the panel showing every claim state):\n\n```bash\npip install \"citevahti[mcp]\"\ncitevahti demo\n```\n\n**Run the review panel from the terminal:**\n\n```bash\npip install \"citevahti[mcp]\"\ncitevahti run\n```\n\nCiteVahti opens the review panel on your machine at `127.0.0.1` and tells you the one next\nthing to do.\n\n**Run the desktop app from source** (the same supervised shell as the packaged `.app`):\n\n```bash\npip install \"citevahti[app]\"\ncitevahti-app\n```\n\n**What the app actually runs:** `CiteVahti.app` is a thin shell supervising two local\nprocesses — the review panel engine (for you) and an optional MCP agent server (for a chat\nclient; it can suggest and stage, but it can never rate for you, see your rating first, or\nmake the final call). Both bind only to `127.0.0.1`, write rotating logs (menu-bar icon →\n*Open Logs Folder*), exit on their own if the app dies, and shut down cleanly when you quit.\nThe agent server is off until you enable it and can be stopped anytime from the menu bar.\n\n**Deeper reading:**\n\n- [Architecture](https://github.com/heidihelena/citevahti/blob/main/docs/ARCHITECTURE.md) · [CLI reference](https://github.com/heidihelena/citevahti/blob/main/docs/CLI.md)\n- [Safety invariants](https://github.com/heidihelena/citevahti/blob/main/docs/SAFETY_INVARIANTS.md) · [Security policy](https://github.com/heidihelena/citevahti/blob/main/SECURITY.md)\n- [Building the desktop app / extension](https://github.com/heidihelena/citevahti/blob/main/desktop-extension/BUILD.md)\n- [Status & capabilities](https://github.com/heidihelena/citevahti/blob/main/docs/STATUS.md) · [Dependencies / SBOM](https://github.com/heidihelena/citevahti/blob/main/docs/SBOM.md)\n\n\n\n## How it works\n\nCiteVahti always tells you the **one next thing** to do — there's no command to remember and\nnothing to set up.\n\n\n\nYou start by adding your manuscript — choose a Word or Markdown file, drag it onto the window,\nor paste the text. Nothing is uploaded.\n\n\n\nThen you work through each claim on a simple card that walks a visible\n**Rate → Reveal → Decide → Write** path, so you always know where you are:\n\n1. **Rate** — you judge whether the cited paper supports the claim.\n2. **Reveal** — *now* the optional AI second opinion appears (never before).\n3. **Decide** — accept, caution, review, or reject. The decision is yours.\n4. **Write** — optionally update your manuscript or Zotero, always after a preview.\n\n\n\n## A record you can hand to anyone\n\nEvery decision is saved to a **timestamped, tamper-evident review record** — a plain-language\nlog of what you reviewed and in what order. Export it as a Word document, a print-ready report,\nor a review-packet `.zip` (which even includes an auto-filled methods paragraph with your\nreview's real numbers) for a methods section, a co-author, a supervisor, or a journal.\n\n\n\n## Build an evidence map you can reuse\n\n**Stop losing evidence work after each project.** Most citation checks vanish after one\nmanuscript or one deadline. Every claim you assess in CiteVahti is saved to your own private\nevidence map, so next time you work on the same topic you don't start from zero — pull back\nearlier judgments, see what you already checked, and spot where the literature is thin,\ncontradictory, or overclaimed.\n\nIt stays **on your machine**. Sharing into the wider, **de-identified** community evidence map\nis a separate, opt-in step, and you preview the exact payload before anything is sent — never\nyour manuscript, search history, patient data, registry data, or private project identifiers.\n\n[How your data is handled →](https://github.com/heidihelena/citevahti/blob/main/docs/CONTRIBUTOR_PRIVACY.md)\n\n## The AI is a second opinion — never the judge\n\nYou always rate first. CiteVahti can then show a second rating from an AI, and compare the two\n— but the AI value stays hidden until your rating is in, and the final decision is always\nyours. There's no hidden AI subscription: pick the mode you want, or leave AI off entirely.\n\n\n\nYou can get the second opinion from:\n\n- your MCP assistant (e.g. Claude Desktop),\n- a **local model on your own computer** (free, fully private — see below), or\n- your own OpenAI/Anthropic-compatible API key.\n\n\nRun the AI on your own machine (free, nothing leaves your device)\n\n1. Install [Ollama](https://ollama.com/download) (macOS: `brew install ollama`).\n2. Pull a model — `qwen2.5` is recommended for claim checking:\n ```bash\n ollama pull qwen2.5\n ```\n (Pick a smaller tag if your machine is tight on memory, e.g. `ollama pull qwen2.5:3b`.)\n3. In the panel, open **Settings → AI second opinion → Local AI**. CiteVahti detects the\n installed model and pins its exact version so the rating stays auditable.\n\nTo update later, re-pull the same name (`ollama pull qwen2.5`).\n\n\n\n## Privacy Policy\n\nCiteVahti is local-first. Full policy: ****\n(source: [`docs/PRIVACY.md`](https://github.com/heidihelena/citevahti/blob/main/docs/PRIVACY.md)). In short:\n\n- **No telemetry, no analytics, no account** — CiteVahti collects nothing about you and\n nothing for us, and never phones home.\n- Your manuscript, ratings, and review record stay in a folder on your computer.\n- The panel runs only on your own machine (loopback, `127.0.0.1`).\n- AI is optional and blinded; local and bring-your-own-key modes are supported.\n- Changes to Zotero or your manuscript always require a preview and your confirmation, and can\n be undone.\n- **Outbound calls are enumerated, not hidden.** The main ones: the literature services you\n choose to search (PubMed, OpenAlex, Semantic Scholar, Crossref), your Zotero if you connect\n it, the optional AI second opinion (off by default; API mode sends the claim plus the paper's\n title and abstract to the provider *you* configure), a user-initiated check to PyPI for a\n newer version, a one-time Pandoc download if you ask for a Word export without Pandoc\n installed, and opt-in RFC 3161 timestamping (audit-head hash only). The complete table —\n destination, payload, trigger, default — is in [`docs/PRIVACY.md`](docs/PRIVACY.md).\n\n## What CiteVahti writes\n\nCiteVahti never silently changes Zotero or your manuscript. Every write follows the same gate:\n\n```\nPreview → Confirm → Audit → Undo available\n```\n\n## What it does and does not check\n\nCiteVahti checks **citation support, not the truth of the underlying claims.** It tests\nwhether the cited paper backs the sentence — it does not certify that every claim in the\nmanuscript was entered, and it is not a clinical or scientific oracle. Blinding is enforced by\nthe workflow (the AI value is withheld until you rate) and recorded in the review record with\ntimestamps, so the order is auditable, not assumed.\n\n## Responsibility & disclosure\n\nCiteVahti is a **local-first citation-support audit tool for manuscript claims** — it helps\nauthors test whether the cited evidence supports each claim before submission.\n\nUsing CiteVahti **does not certify** scientific truth, manuscript quality, publication\nsuitability, or the absence of citation problems. It provides structured decision support and\nan audit trail; **final responsibility remains with the human author, reviewer, editor, or\ninstitution.** CiteVahti is developed by the author — any publication, review, teaching, or\nevaluation that uses it should **disclose the tool use and the developer relationship where\nrelevant.** Ready-to-adapt disclosure text and the full statement are in\n[docs/DISCLOSURE.md](https://github.com/heidihelena/citevahti/blob/main/docs/DISCLOSURE.md).\n\n## Learn more\n\n- [Quickstart — your first citation check, step by step](https://github.com/heidihelena/citevahti/blob/main/docs/QUICKSTART.md)\n- [Writing good (atomic) claims](https://github.com/heidihelena/citevahti/blob/main/docs/WRITING_GOOD_CLAIMS.md)\n- [Reporting in your methods section](https://github.com/heidihelena/citevahti/blob/main/docs/REPORTING.md)\n- [Known limitations](https://github.com/heidihelena/citevahti/blob/main/docs/KNOWN_LIMITATIONS.md)\n- [Responsibility, disclosure & conflict of interest](https://github.com/heidihelena/citevahti/blob/main/docs/DISCLOSURE.md)\n- [Contributor privacy](https://github.com/heidihelena/citevahti/blob/main/docs/CONTRIBUTOR_PRIVACY.md)\n\n*Technical documentation (architecture, CLI, safety invariants, SBOM) lives under the\n\"For developers\" toggle above and in [docs/](https://github.com/heidihelena/citevahti/tree/main/docs).*\n\n## Feedback & support\n\nCiteVahti is a free beta and your feedback shapes it.\n\n- **Found a bug, or a citation judged wrong?** Open an issue — there are quick forms for\n [a bug](https://github.com/heidihelena/citevahti/issues/new?template=bug_report.yml) and for\n [a wrong support judgment](https://github.com/heidihelena/citevahti/issues/new?template=wrong_support_judgment.yml).\n- **Have an idea?** [Request a feature](https://github.com/heidihelena/citevahti/issues/new?template=feature_request.yml).\n- **Anything private or security-related?** Email **privacy@vahtian.com** — please don't open a\n public issue.\n\nNew here? Open the app and drag a manuscript in — or follow the\n[Quickstart](https://github.com/heidihelena/citevahti/blob/main/docs/QUICKSTART.md) step by step.\n\n## Companion: FullVahti\n\n[**FullVahti**](https://github.com/heidihelena/fullvahti) is a sibling Zotero plugin that finds\nfree, legal open-access PDFs for your references and writes CiteVahti's check results back as\ntags — so the citation check and the full text live in one place. See the\n[FullVahti README](https://github.com/heidihelena/fullvahti) to install (a two-click Zotero\nplugin, no terminal).\n\n## License\n\nApache License 2.0 — see [LICENSE](https://github.com/heidihelena/citevahti/blob/main/LICENSE) and [NOTICE](https://github.com/heidihelena/citevahti/blob/main/NOTICE).\n\n\n",
"bytes": 16881,
"sha": "121190a7afce207b61a01d9ac4dd7a2d48a5ebc2e379b9c35beb5f6bca37fe0f",
"repo_slug": "heidihelena/citevahti",
"fonte": "repo",
"truncated": false,
"api": "https://agentalog.com/api/listings/mcp_io_github_heidihelena_citevahti_711ff314/readme"
}