All authors
Claude Skills by ruc-datalab
github.com/ruc-datalab16 skills0 installs14 views
- Build OntologyBuild an initial ontology from data and analytical goals, prepare relevant questions from supplied needs and project history, and automatically show the evidence-grounded result.Votes: 0GitHub stars: 523
- Evolve OntologyImprove an existing ontology using real project tasks and grounded exploration, run missing baselines, validate candidates and automatically show the outcome. This skill is triggered when the user requests to "self-evolve the current ontology layer system". Autonomously evolve an ontology-layer system through a four-step loop of diagnosis, attribution, patching, and Parent/Candidate evaluation until a reproducibly better version is obtained, or external conditions prevent trustworthy continua...Votes: 0GitHub stars: 523
- Explore OntologyExplore ontology questions, grounded concepts, evidence, actual task results and evolution outcomes in a read-only interactive version explorer.Votes: 0GitHub stars: 523
- Build OntologyBuild an initial ontology from data and analytical goals, prepare relevant questions from supplied needs and project history, and automatically show the evidence-grounded result.Votes: 0GitHub stars: 523
- Evolve OntologyImprove an existing ontology using real project tasks and grounded exploration, run missing baselines, validate candidates and automatically show the outcome. This skill is triggered when the user requests to "self-evolve the current ontology layer system". Autonomously evolve an ontology-layer system through a four-step loop of diagnosis, attribution, patching, and Parent/Candidate evaluation until a reproducibly better version is obtained, or external conditions prevent trustworthy continua...Votes: 0GitHub stars: 523
- Explore OntologyExplore ontology questions, grounded concepts, evidence, actual task results and evolution outcomes in a read-only interactive version explorer.Votes: 0GitHub stars: 523
- Skilladam OptimizeOptimize an existing SKILL.md from a usage-intent description with SkillAdam. Use when a user asks to test, improve, patch, review, validate, or iteratively optimize an Agent Skill, with optional selective hunk review.Votes: 0GitHub stars: 79
- Skilladam OptimizeOptimize an existing SKILL.md from a usage-intent description with SkillAdam. Use when a user asks to test, improve, patch, review, validate, or iteratively optimize an Agent Skill, with optional selective hunk review.Votes: 0GitHub stars: 79
- Skilladam OptimizeOptimize an existing SKILL.md from a usage-intent description with SkillAdam. Use when a user asks to test, improve, patch, review, validate, or iteratively optimize an Agent Skill, with optional selective hunk review.Votes: 0GitHub stars: 79
- Skilladam OptimizeOptimize an existing SKILL.md from a usage-intent description with SkillAdam. Use when a user asks to test, improve, patch, review, validate, or iteratively optimize an Agent Skill, with optional selective hunk review.Votes: 0GitHub stars: 79
- Alfworld1. **Parse the goal**: Identify the exact target object class, required quantity, required transform (heat/cool/clean) if any, the transform appliance, and the final destination receptacle or interaction (desklamp for examine-in-light tasks). Treat the destination as the final target, not as the first place to search unless it is also a likely source.Votes: 0GitHub stars: 79
- Docvqa1. **Read the question for answer shape before scanning**: Decide whether it asks for a year/date, count, percentage, money amount, table cell value, category name, identifier number, page indicator, or handwritten note. This determines whether the answer should be a bare number, a copied phrase, a category label, a yes/no token, or a full multi-part written value.Votes: 0GitHub stars: 79
- Lmb1. **Isolate the precise theorem conclusion**: Identify every clause of the claim — all hypotheses, quantifiers, boundary cases, and every conclusion clause marked by words like "moreover" or "in fact." List every distinct feature the theorem actually proves; you will need this complete list in steps 3 and 6.Votes: 0GitHub stars: 79
- Officeqa1. **Locate the exact source before reading broadly**: If the task context already provides parsed Treasury or federal source excerpts, extract any usable rows from those excerpts first, then use tools only to fill missing dates, rows, or verification spans. Use `glob` for bulletin date/page patterns, then `grep` targeted phrases from the question: publication month/year, table title, entity, security type, maturity class, and period label. If the question names a publication date such as June 1Votes: 0GitHub stars: 79
- SearchqaChoose concise canonical answer forms for clue-style question answering.Votes: 0GitHub stars: 79
- Spreadsheetbench1. **Load for mutation, optionally load for values**: Use `wb = load_workbook(INPUT_PATH)` for writing. If existing formula results are needed, also open `wbv = load_workbook(INPUT_PATH, data_only=True)` and read cached values from `wbv`, but mutate only `wb`. 2. **Target exact sheets, cells, and qualifying rows**: Pick sheets by visible names when present and write only the requested output range. Avoid broad rewrites unless explicitly requested. For highlight/color-based conditions, inspect thVotes: 0GitHub stars: 79