Agent Claude Code
Vahtian's company voice + SEO agent. Rewrites and reviews page titles, meta descriptions, and headings so they read in the Vahtian voice AND meet the SEO-compliant company standard (docs/seo-copy-standard.md). Use when auditing or improving titles/descriptions/H1s across vahtian.com, standardising SERP copy before…
Skill Claude CodeCodex
Run, serve, and screenshot the Vahtian static site (vahtian.com) locally. Use when asked to launch, preview, build, or take a screenshot of the Vahtian homepage or any product page (CiteVahti, StudyVahti, MethodVahti, DictVahti, MatchVahti-Lite, ReviewVahti, ExtractVahti, GuidelineVahti, FullVahti, SynthVahti), or to…
Skill Claude CodeCodex
Draft channel-ready announcement posts for a shipped Vahtian change — GitHub release note, Zotero community post, research-methods post, and a short blurb — then route every draft through vahtian-copy-editor and vahtian-brand-safety and queue them for the founder's approval. Use when a product ships or updates and…
Skill Claude CodeCodex
Review Vahtian copy for research-integrity safety. Use when reviewing or writing Vahtian UI labels, product descriptions, agent-workflow text, documentation, prompts, GitHub READMEs, LinkedIn posts, ad copy, or landing pages. Preserves Vahtian's invariant — it assesses claim-source support and records auditable…
Skill Claude CodeCodex
Edit Vahtian copy for craft — clarity, rhythm, voice, and stripping the tells that make writing read as AI-generated. Use when writing or revising Vahtian headlines, ledes, product descriptions, microcopy (buttons, tags, empty states), docs, READMEs, LinkedIn/launch posts, or any outbound copy, and the question is…
Skill Claude CodeCodex
Make minimal, coherent edits to the Vahtian static site. Use when editing Vahtian HTML, CSS, inline JS, page structure, navigation, product cards, CTA sections, accessibility markup, or responsive layout. Preserves static-site simplicity, zero trackers, local-first positioning, and research-integrity copy. Implements…
Skill Claude CodeCodex
Set up and run publishing for the Vahtian packages from the monorepo — the Python package to PyPI (via Trusted Publishing / OIDC, no API tokens) and the R package to R-universe and CRAN. Use when configuring releases, adding a publish workflow, cutting a version, reserving the vahtian name on PyPI/CRAN, or wiring the…
Skill Claude CodeCodex
Build and maintain the Vahtian static site (vahtian.com) to a professional standard. Use when adding a new page or tool, doing a site-wide quality pass, or whenever you touch HTML/meta/sitemap/footer — it enforces the per-page completeness, consistency, accessibility, SEO/GEO, security, and editorial standards that…
Skill Claude CodeCodex
Audit Vahtian web pages, product pages, landing pages, carousels, onboarding flows, or docs for UI/UX clarity. Use when reviewing the Vahtian site for first-screen comprehension, navigation, decision paths, CTA clarity, information hierarchy, mobile readability, accessibility, and whether a researcher understands what…
Instructions file
Claude Code instructions for heidihelena/vahtian, covering vahtian — repo guide for claude, the invariant (applies to every edit, every session), which skill for what and hard floor (even when no skill fires).
Skill Claude CodeCodex
Use when a researcher is deciding or defending a qualitative sample size, or appraising how defensible a study's methodological decisions are before results exist. Triggers on sample size justification, how many interviews, how many focus groups, saturation, information power, thematic saturation, defend my sample…
Skill Claude CodeCodex
Use when a researcher needs literature-search, screening, evidence-synthesis, reference-checking, or citation-checking support — a systematic or scoping review, a reproducible multi-source search, blinded dual screening, inter-rater reliability (Cohen's κ / Krippendorff's α), full-text retrieval, checking a reference…
Skill Claude CodeCodex
Use when a finished manuscript is about to be submitted and the question is what to fix first: a pre-submission check, a submission-readiness pass, "what will get me desk rejected", reporting-guideline gaps (STROBE, CONSORT, PRISMA, COREQ), a missing ethics or funding or competing-interests or data-availability…
MCP server Claude CodeCodexCursor
Reproducible, provenance-first evidence tooling — freeze a corpus, verify it, keep a hash-chained audit trail. Human-first, auditable, local-first. Runs locally from the vahtian Python package.
Plugin Claude Code
CiteVahti — citation integrity for manuscripts. Human-first blinded claim checking with PubMed, decision-gated Zotero write-back, and a hash-chained audit trail; plus an AI-assisted drafting skill that writes only from checked claims.
Skill Claude CodeCodex
Benchmark local LLMs (Ollama models such as qwen3:14b, hermes3:8b) as CiteVahti citation prescreening agents on a chosen topic, and build the publication-style prescreen report + a dedicated CiteVahti store from the run. Use this whenever the user wants to test how well small/local models rate claim-vs-source support…
Skill Claude CodeCodex
The secure front-door for building, developing, releasing, and updating CiteVahti. Use this whenever you are about to cut a release or bump the version; build, sign, or notarize the desktop app, the .mcpb extension, or the PyInstaller binary; set up or change automatic updates (tufup / code signing); add or change an…
Skill Claude CodeCodex
Use before publishing ANY public CiteVahti artifact — site copy, README, LinkedIn post, release notes, docs, conference slide, screenshot caption — to audit it against the must-not-claim list and the evidence-tier language rules. Also use when drafting such copy, or when asked "can we say this about CiteVahti?".…
Skill Claude CodeCodex
Use when checking manuscript claims against evidence, assessing whether a citation actually supports what it is cited for, adding audited references to Zotero, generating a citation-integrity report, or when a researcher mentions claims, citations, PubMed, Zotero, PRISMA, blinded rating, or peer review of evidence.
Skill Claude CodeCodex
Use when measuring or publishing CiteVahti's own accuracy — running the automatic claim-lexicon evaluation, checking it against its regression baseline, reasoning about how AI second-rater models are rated, or answering "how accurate is CiteVahti?". Follows ADR-0009's three evaluation objects (automatic lexicon eval…
Skill Claude CodeCodex
Use when choosing or comparing the AI second-rater model in CiteVahti — picking a model, running the same topic through several models to see where they diverge, reading a model's rating/scoreboard standing, acting on a low-rated model, or setting up a multi-model pre-check before a guideline group starts. Follows…
Skill Claude CodeCodex
Use when writing or regenerating CiteVahti install and quickstart documentation for a distribution channel — the desktop app, the .mcpb in Claude Desktop, MCP config for Claude Code or other clients, the VS Code extension, or pip — or when checking that channel docs still match the shipped release. Maintainer-facing…
Skill Claude CodeCodex
Use when shipping a CiteVahti release — cutting a version, checking feature parity across the MCP server, desktop extension (.mcpb), loopback panel, and VS Code extension, bumping versions in lockstep, writing the changelog entry, minting the release DOI, or preparing rollback notes. Maintainer-facing. Runs the…
Skill Claude CodeCodex
Use when packaging a completed CiteVahti audit for a journal, supervisor, or co-author — writing the methods-section paragraph that describes the claim-checking workflow, exporting the hash-chained audit trail, assembling an evidence appendix, or preparing a pre-submission integrity summary. Runs downstream of…