All authors
Claude Skills by LARi-UQAC
github.com/LARi-UQAC45 skills0 installs27 views
- Aider SetupSet up and maintain the aider nightly local-coding pipeline: two local Ollama models (a writer that codes and tests, a reviewer that never edits) run one aider process per plan overnight, with no Claude Code involved, driven by aider-plan.ps1/aider-night.ps1 and gated by measured token budgets.Votes: 0GitHub stars: 9
- DeliberationTwo-round multi-model deliberation panel: Gemini and GitHub Copilot debate a near-final draft (via deliberate.py), enriched with Consensus and optional Scopus.AI evidence, after which Claude arbitrates per the canonical protocol, validates new references against Scopus, and writes a Deliberation Log.Votes: 0GitHub stars: 9
- Drawio2tikzConvert one sheet of a .drawio file into a coordinate-exact TikZ fragment for embedding in LaTeX. Use when the user asks to turn a draw.io diagram/figure into TikZ, when a LaTeX figure must reproduce a draw.io sheet exactly (mechanical drawings, flowcharts, module layouts), or for /drawio2tikz.Votes: 0GitHub stars: 9
- Extract ContributionsExtract the stated scientific contribution of a cited paper from its full text, to check that a citation says what the paper actually claims. Companion to extract-futureworks: that one mines what a corpus says remains to be done, this one mines what each paper says it did. Two modes.Votes: 0GitHub stars: 9
- Extract FutureworksFuture-works analysis skill with two modes. Mode audit: review one manuscript''s own future-works / conclusion / limitations section (.tex/.md) — presence, testability, link to a stated limitation, and novelty against the literature — and emit flagged findings for the host agent''s plan.Votes: 0GitHub stars: 9
- Extract Paper IdeaExtract a paper''s own content (contribution, novelty, method, results, limitations, future work) into one JSON artifact from its full text, the basis for drafting or refreshing its abstract - or, for a UQAC thesis, its Resume (French) + Abstract (English) pair.Votes: 0GitHub stars: 9
- Extract StatisticStatistical analysis skill with two modes. Mode audit: review the statistics of one manuscript (.tex/.md) — test selection, assumptions, effect size, presentation, and cross-validation between text, tables, and figures — and emit flagged findings for the host agent''s plan.Votes: 0GitHub stars: 9
- Form ServiceStateless mechanics for official PDF forms from any institution (proven on UQAC''s Decanat des etudes thesis forms and Service des ressources financieres travel and expense forms): retrieve a form over https with the validated ingest contract, and later fill and cryptographically sign one.Votes: 0GitHub stars: 9
- GeolocalisationMap, geolocate, or produce a study-location map for a literature-review corpus from a BibTeX file: turn a .bib into study_locations.csv, a KML/GeoJSON/world-map figure, or answer a question about a corpus''s geographic coverage, spatial distribution, or country breakdown.Votes: 0GitHub stars: 9
- Latex HygieneMeasure LaTeX manuscript hygiene mechanically: forbidden characters, an AI-usage risk score, prose, track-changed and per-section word counts against a grant-form cap, abstract length, brace balance, par-inside-changes-argument corruption, and citation-key coverage between a .tex and its .bib.Votes: 0GitHub stars: 9
- Loop EngineerBudget-bounded develop-and-improve loop: design, plan, code, comment, test, review, score, and correct in a loop until a composite quality gate is met or a hard budget/iteration/no-progress stop is hit.Votes: 0GitHub stars: 9
- Narrative CvUse when drafting, updating, or tailoring a narrative ''CV descriptif'' for a Quebec/Canadian grant competition (FRQ''s CV-FRQ, structurally identical to the tri-agency CIHR/NSERC/SSHRC CV commun des trois organismes): three sections (career/skills, up to ten contributions and experiences, supervision and mentoring), matched to one program''s own objectives and evaluation criteria.Votes: 0GitHub stars: 9
- Obsidian CliUse when a task needs Claude to act on the Obsidian vault - read or search a note, list or count notes, inspect tags, tasks, links or properties, move or rename a file, or deposit a captured learning for the vault.Votes: 0GitHub stars: 9
- Opt Local Stt VramBenchmark speech-to-text models (Whisper, faster-whisper) for this GPU beside the resident local LLM, all on VRAM, and print a scored table to pick the best one: transcript error against your own reference words, live-caption speed, and how much VRAM each takes and pushes the LLM out of.Votes: 0GitHub stars: 9
- Opt Local Vram LlmTune a local Ollama model for this machine''s GPU: retain the largest context window that keeps the model 100 percent resident in VRAM, among configurations whose decode throughput clears a floor.Votes: 0GitHub stars: 9
- Paper2talkTurn an accepted paper into a conference talk: slides on the lab gabarit plus timed speaker notes, in PowerPoint, LaTeX Beamer, or a self-contained web deck.Votes: 0GitHub stars: 9
- Recommendation LetterGenerate support, recommendation, appreciation, acceptance, and dispense (short-stay invitation) letters for a supervisor''s students and candidates, in LaTeX compiled to PDF, signed by the active profile''s author.Votes: 0GitHub stars: 9
- Rt ObserveReport the state of this toolkit as one local page - the mirror matrix of every agent, skill, command and rule against every harness dialect, plus live sessions, services and plan progress. Trigger on rt-dashboard, mirror matrix, mirror drift, is my toolkit deployed, which harness is missing an agent, toolkit status, harness state.Votes: 0GitHub stars: 9
- Scholar EvaluationSystematic framework for evaluating scholarly and research work based on the ScholarEval methodology. This skill should be used when assessing research papers, evaluating literature reviews, scoring research methodologies, analyzing scientific writing quality, or applying structured evaluation criteria to academic work.Votes: 0GitHub stars: 9
- Scientific WritingCore skill for the deep research and writing tool. Write scientific manuscripts in full paragraphs (never bullet points). Use two-stage process: (1) create section outlines with key points using research-lookup, (2) convert to flowing prose.Votes: 0GitHub stars: 9
- ScopusUse this skill to search Scopus for academic references, validate existing references, or produce literature reviews. Trigger on: /scopus, mentions of Scopus search, reference validation requests, finding papers on a topic.Votes: 0GitHub stars: 9
- Word2latexConvert a Word .docx template (Mitacs, CRSNG, FRQNT, UQAC, partner forms) into a faithful LaTeX source. Use when the user asks to translate a Word gabarit to LaTeX, when a generated .tex must match a .docx visually, or for /word2latex.Votes: 0GitHub stars: 9
- Aider SetupSet up and maintain the aider nightly local-coding pipeline: two local Ollama models (a writer that codes and tests, a reviewer that never edits) run one aider process per plan overnight, with no Claude Code involved, driven by aider-plan.ps1/aider-night.ps1 and gated by measured token budgets. Use this skill whenever the user wants to install, tune, or fix the aider kit on this machine; asks to run, resume, or debug an overnight aider plan; wants the distribution zip for students rebuilt or ...Votes: 0GitHub stars: 9
- Test Driven DevelopmentUse when implementing any feature or bugfix, before writing implementation codeVotes: 0GitHub stars: 9
- DeliberationTwo-round multi-model deliberation panel: Gemini and GitHub Copilot debate a near-final draft (via deliberate.py), enriched with Consensus and optional Scopus.AI evidence, after which Claude arbitrates per the canonical protocol, validates new references against Scopus, and writes a Deliberation Log. Used as the standardized Deliberation step inside the academic auditor and researcher agents (scopus-auditor, paper-auditor, thesis-auditor, thesis-proposal-auditor, scopus-researcher, reviewer-r...Votes: 0GitHub stars: 9
- Drawio2tikzConvert one sheet of a .drawio file into a coordinate-exact TikZ fragment for embedding in LaTeX. Use when the user asks to turn a draw.io diagram/figure into TikZ, when a LaTeX figure must reproduce a draw.io sheet exactly (mechanical drawings, flowcharts, module layouts), or for /drawio2tikz. Drives scripts/drawio2tikz.py (absolute coordinates, group-offset resolution, edge anchoring to shape perimeters, waypoints, braces, FR->EN translation).Votes: 0GitHub stars: 9
- Extract ContributionsExtract the stated scientific contribution of a cited paper from its full text, to check that a citation says what the paper actually claims. Companion to extract-futureworks: that one mines what a corpus says remains to be done, this one mines what each paper says it did. Two modes. Mode validate: for one manuscript, pair every cited key with its paper's own contribution sentences and flag the citations whose claim the paper does not support. Mode mine: sweep a refs/ corpus and tabulate what...Votes: 0GitHub stars: 9
- Extract FutureworksFuture-works analysis skill with two modes. Mode audit: review one manuscript's own future-works / conclusion / limitations section (.tex/.md) — presence, testability, link to a stated limitation, and novelty against the literature — and emit flagged findings for the host agent's plan. Mode mine: extract the stated future works / open problems of a corpus from full text (download via download_pdf.py any-format, parse via extract_text.py --section-scan) and synthesize a corpus future-works tab...Votes: 0GitHub stars: 9
- Extract Paper IdeaExtract a paper's own content (contribution, novelty, method, results, limitations, future work) into one JSON artifact from its full text, the basis for drafting or refreshing its abstract - or, for a UQAC thesis, its Resume (French) + Abstract (English) pair. Ships no reader of its own: reuses paper2talk's LaTeX flattening, extract-statistic's future-works section scan, and extract-contributions' marker scan, pointed at the paper's own text rather than a cited one. Trigger on: extract-paper...Votes: 0GitHub stars: 9
- Extract StatisticStatistical analysis skill with two modes. Mode audit: review the statistics of one manuscript (.tex/.md) — test selection, assumptions, effect size, presentation, and cross-validation between text, tables, and figures — and emit flagged findings for the host agent's plan. Mode mine: extract the reported statistics from the full-text PDFs of a corpus (download via download_pdf.py, parse via extract_text.py) and synthesize a corpus statistics table plus a statistical-improvement opportunity li...Votes: 0GitHub stars: 9
- Form ServiceStateless mechanics for official PDF forms from any institution (proven on UQAC's Decanat des etudes thesis forms and Service des ressources financieres travel and expense forms): retrieve a form over https with the validated ingest contract, and later fill and cryptographically sign one. Holds no catalogue and no personal data: the form catalogue, field maps and profile live in the consuming application (ThesisTracker, for the UQAC deployment). Trigger on: /fetchform, fetch an official form ...Votes: 0GitHub stars: 9
- GeolocalisationMap, geolocate, or produce a study-location map for a literature-review corpus from a BibTeX file: turn a .bib into study_locations.csv, a KML/GeoJSON/world-map figure, or answer a question about a corpus's geographic coverage, spatial distribution, or country breakdown. Trigger on: /geolocalisation, 'map the corpus', 'geolocate the studies', 'bib to map'. Also triggers on 'carte des études' and 'où ont eu lieu les études'. Extracts each paper's study / case-study location automatically (per-...Votes: 0GitHub stars: 9
- Latex HygieneMeasure LaTeX manuscript hygiene mechanically: forbidden characters, an AI-usage risk score, prose, track-changed and per-section word counts against a grant-form cap, abstract length, brace balance, par-inside-changes-argument corruption, and citation-key coverage between a .tex and its .bib. Backs the /texcheck command and the aiscan / wc checks that paper-auditor and submit-checker already describe in prose. Trigger on: /texcheck, LaTeX hygiene, AI-usage score, word count, brace balance, c...Votes: 0GitHub stars: 9
- Loop EngineerBudget-bounded develop-and-improve loop: design, plan, code, comment, test, review, score, and correct in a loop until a composite quality gate is met or a hard budget/iteration/no-progress stop is hit. Keeps the best cloud model (Fable 5) as orchestrator and judge, uses cheaper cloud tiers (Opus to plan, Sonnet to execute and review) for the actions, and delegates code and comments to the local-coder / local-writer agents over the Bash bridge so the heavy generation is free. Trigger on: /loo...Votes: 0GitHub stars: 9
- Narrative CvUse when drafting, updating, or tailoring a narrative 'CV descriptif' for a Quebec/Canadian grant competition (FRQ's CV-FRQ, structurally identical to the tri-agency CIHR/NSERC/SSHRC CV commun des trois organismes): three sections (career/skills, up to ten contributions and experiences, supervision and mentoring), matched to one program's own objectives and evaluation criteria. Trigger on: CV-FRQ, CV descriptif, CV trois organismes, narrative CV, tri-agency CV, tailor my CV to this grant, /cv.Votes: 0GitHub stars: 9
- Obsidian CliUse when a task needs Claude to act on the Obsidian vault - read or search a note, list or count notes, inspect tags, tasks, links or properties, move or rename a file, or deposit a captured learning for the vault. Also use when a vault write appeared to succeed but the note is absent, when a numbered duplicate such as "Decisions 1.md" appears, or when Obsidian shows "A JavaScript error occurred in the main process". Skip for questions about the Obsidian GUI, themes, plugins or sync conflicts...Votes: 0GitHub stars: 9
- Opt Local Stt VramBenchmark speech-to-text models (Whisper, faster-whisper) for this GPU beside the resident local LLM, all on VRAM, and print a scored table to pick the best one: transcript error against your own reference words, live-caption speed, and how much VRAM each takes and pushes the LLM out of. Downloads the named models, measures each in a fresh process, gates out any model that demotes too much of the LLM, and stops before adoption. Trigger on: /opt-local-stt-vram, 'which speech-to-text model fits...Votes: 0GitHub stars: 9
- Opt Local Vram LlmTune a local Ollama model for this machine's GPU: retain the largest context window that keeps the model 100 percent resident in VRAM, among configurations whose decode throughput clears a floor. Builds the tuned -gpu tag from a base tag, sweeps num_ctx against OLLAMA_KV_CACHE_TYPE (restarting the daemon between axis values and proving it took them), writes the measurement, and declares the tag as a candidate for its role. Trigger on: /opt-local-vram-llm, 'optimize the local model for my VRAM...Votes: 0GitHub stars: 9
- Paper2talkTurn an accepted paper into a conference talk: slides on the lab gabarit plus timed speaker notes, in PowerPoint, LaTeX Beamer, or a self-contained web deck. Use when the user asks to build a presentation, a deck, slides, or a talk from a paper or an article; for /talk; for 'prepare the presentation for <conference>', 'my paper was accepted, I need the slides', 'speaker notes', 'presentation from this accepted paper', 'passer de l article a la presentation'. Asks the six questions that set th...Votes: 0GitHub stars: 9
- Recommendation LetterGenerate support, recommendation, appreciation, acceptance, and dispense (short-stay invitation) letters for a supervisor's students and candidates, in LaTeX compiled to PDF, signed by the active profile's author. Ingests the candidate's own files (CV, transcript, project description, motivation letter) and highlights both the candidate's dossier and the professor's own experience with the candidate. Trigger on: recommendation letter, support letter, appreciation letter, acceptance letter, in...Votes: 0GitHub stars: 9
- Rt ObserveReport the state of this toolkit as one local page - the mirror matrix of every agent, skill, command and rule against every harness dialect, plus live sessions, services and plan progress. Trigger on rt-dashboard, mirror matrix, mirror drift, is my toolkit deployed, which harness is missing an agent, toolkit status, harness state. Runs on any OS from a bare clone with no harness installed, because the matrix compares files rather than asking a tool.Votes: 0GitHub stars: 9
- Scholar EvaluationSystematic framework for evaluating scholarly and research work based on the ScholarEval methodology. This skill should be used when assessing research papers, evaluating literature reviews, scoring research methodologies, analyzing scientific writing quality, or applying structured evaluation criteria to academic work. Provides comprehensive assessment across multiple dimensions including problem formulation, literature review, methodology, data collection, analysis, results interpretation, ...Votes: 0GitHub stars: 9
- Scientific WritingCore skill for the deep research and writing tool. Write scientific manuscripts in full paragraphs (never bullet points). Use two-stage process: (1) create section outlines with key points using research-lookup, (2) convert to flowing prose. IMRAD structure, citations (IEEE/APA/AMA/Vancouver), figures/tables, reporting guidelines (CONSORT/STROBE/PRISMA), for research papers and journal submissions.Votes: 0GitHub stars: 9
- ScopusUse this skill to search Scopus for academic references, validate existing references, or produce literature reviews. Trigger on: /scopus, mentions of Scopus search, reference validation requests, finding papers on a topic.Votes: 0GitHub stars: 9
- Word2latexConvert a Word .docx template (Mitacs, CRSNG, FRQNT, UQAC, partner forms) into a faithful LaTeX source. Use when the user asks to translate a Word gabarit to LaTeX, when a generated .tex must match a .docx visually, or for /word2latex. Drives pandoc + applies the standard patch sequence (Arial, full-grid tables, centered sections, landscape geometry, first-page banner). Delegates the actual patch work to the word-to-latex agent.Votes: 0GitHub stars: 9