Productivity
Skill Forge 技能熔炉
技能熔炉 — 锻造/评估/改进 Skill。说 技能熔炉 走全流程（含R5改进已有skill）；说 技能评估/skill评估/评估技能 只做同类比对+腾讯9维度。可选能力：搜索SkillHub同类技能（通过TRAE内置工具）、修改已有skill文件（仅R5诊断修复路径，需用户确认）。发布环节请用 skill-pu...
Testing
pulser
Diagnose and test Claude Code skills against Anthropic's 7 principles. Scans SKILL.md files, checks 8 rules (gotchas, description, allowed-tools, file-size, structure, frontmatter,…
AI / ML
darwin-skill
Darwin Skill (达尔文.skill): autonomous skill optimizer inspired by Karpathy's autoresearch. Evaluates SKILL.md files using an 8-dimension rubric (structure + effectiveness), runs hil…
AI / ML
empirical-prompt-tuning
Fetch and execute mizchi's empirical-prompt-tuning skill at runtime. Use when evaluating or iteratively refining an agent-facing prompt (skill / slash command / task prompt / CLAUD…
Research
research-idea-benchmark
This skill should be used when the user asks to "evaluate my research idea", "benchmark my research", "score my paper idea", "assess research quality", "check if my idea is publish…
AI / ML
metaharness
Design, evaluate, and iterate on Claude Code agents, commands, and skills with structural best practices. Use when creating a new .claude/agents/ file, updating an existing agent's…
Research
assess-outline
Evaluates research paper outlines against empirical research criteria for research question clarity, contributions, hypotheses, data description, results approach, and robustness c…
AI / ML
forge-evals
Design evaluations for LLM features including golden datasets, rubric scoring, LLM-as-judge calibration, CI regression detection, online A/B tests, cost and latency budgets, and ad…
AI / ML
ce-optimize
Run metric-driven iterative optimization loops. Define a goal, add measurement scaffolding, execute parallel experiments across approaches, score results against gates or quality j…
Content
通用-平台签约评估框架
Evaluates creative works for platform acceptance or serialization potential using strict, conservative criteria that prioritize negative evidence. Provides a unified assessment fra…
Testing
loop-evals
Designs the evaluation harness for agent loops, establishing trustworthy verification through a 7-layer suite with false-completion-rate and repair-productivity as core metrics. En…
AI / ML
experiment-lab
Run reproducible CS/AI experiments, model evaluation, regression/classification/clustering analyses, bioinformatics workflows, QC, differential expression, single-cell starter anal…
Productivity
legw-rmap-evaluierung-und-aenderung
Manages the full lifecycle of a rulemap norm: versioning, drag-and-drop edits in the builder, NKRG/GGO evaluation, impact monitoring, and feedback from implementation. Outputs chan…
Testing
os-eval-runner
Stateless evaluation engine that scores and gates skill improvement iterations using headless Python scripts. Use for "evaluate this skill", "run autoresearch loop", "optimize this…
Data
data-source-eval
Evaluates new data sources for feasibility, integration effort, and expected value. Runs a standardized spike covering objectives, API discovery, technical viability, architecture,…
Testing
shadow-mode-runner
Coordinates SHADOW mode operation where the agent runs parallel to human input without delivering output or incurring billing. Measures agreement rates and generates promotion repo…
General
new-project-gate
Pre-build gate that evaluates new ideas against both agent criteria (legible, drivable, automatable) and human criteria (usefulness, maintainability) before creating any tool or re…
General
capstone-final-eval
Evaluates Ewha Womans University capstone design final deliverables including team presentations, PDF reports, GitHub repositories, and Project Briefs. Use for end-of-term capstone…
Research
light-idea-critique
Evaluates research ideas against top-tier publication standards to distinguish genuine breakthroughs from incremental or superficial combinations. Delivers blind-then-explicit scor…
Automation
loop-design-check
Designs and reviews goal-oriented agent loops for failure modes like token waste, verifier gaming, and completing wrong answers. Writes loops with decidable goals and skeletons; au…
AI / ML
calibrator
Calibrates prompts, agents, and skills by comparing AI outputs against author-edited versions. Identifies recurring divergence patterns and generates targeted instruction fixes in …
Research
elicit
Applies one of nine structured reasoning methods—pre-mortem, inversion, first principles, red team/blue team, Socratic, constraint removal, stakeholder mapping, analogical, or seco…
AI / ML
evaluating-cosmos-policy
Evaluates NVIDIA Cosmos Policy on LIBERO and RoboCasa simulation environments. Use when setting up cosmos-policy for robot manipulation evaluation, running headless GPU evaluations…
AI / ML
model-profiler
Profiles models from any provider family against a fixed set of 10 work categories, mapping public benchmarks into tier rankings with full provenance. Emits routing-table.json, aud…
Research
legalsearchqa-eval
Benchmarks retrieval of current legal information from external sources and reasoning over it for multiple-choice legal questions. Measures factual accuracy, uncertainty calibratio…
Testing
eval-case-author
Generates evaluation cases for any artifact using real human/agent pairs or declared synthetic cases, then persists them in evals/{artifact_id}/cases/case-{n}.md. Supports C4 (≥30 …
Productivity
jobfit
Evaluates job fit by discovering matching roles across boards and scoring provided postings A–F; researches compensation, company signals, and legitimacy, then outputs ranked brief…
AI / ML
eval-harness-first
Builds the evaluation harness that gates every fine-tuning run — golden sets, per-failure-mode graders, judge calibration, and base-model baselines. Use when launching fine-tuning,…
Research
raven-research-looped-reasoning-eval
Scaffolds reproducible experiments comparing looped/recurrent-depth transformers against matched vanilla models on vulnerability discovery after security-corpus pretraining. Use fo…
Research
lancet-fit
Pre-evaluate any study for The Lancet by checking clinical or public-health significance, global reach, practice-changing potential, and equity considerations to determine fit for …
Productivity
workflow-debate
Structured adversarial debate between AI councillors representing distinct perspectives to rigorously evaluate ideas, plans, or decisions. Produces nuanced verdicts (PROCEED / PROC…
AI / ML
autostar-web
Generalised autonomous optimisation loop for measurable artifacts. Uses connector-backed memory first, then project-pack, then none. Supports iterative experiments, multi-dimension…
Testing
eval-writer
Creates rigorous eval suites for AI agents and LLM systems using trace-driven analysis, binary judges, cross-validation, and agreement metrics. Delivers failure taxonomies, rubrics…
AI / ML
09. RAG 与语言智能体
本页是网站中的第 9 个 CS224n 专题，内容对应官方 PPT `Lecture 10: RAG and Language Agents`。PPT 前几页先收尾上一讲的 Adapter 和高效适配，本页只简要定位；主体是 open-domain question answering、retrieval augmented generation、langu…
AI / ML
agent-eval-adoption
Reference implementation for adopting agent evaluation primitives: manifest definitions, runLoop topologies, prompt optimization, MCP delegation, trace sinks, and CI gates. Pairs w…
Business
team-ops
AI-powered team performance auditing using the Elon Algorithm for ruthless evaluation of output velocity, quality, independence, and initiative. Delivers A/B/C stack ranking with p…
Content
都市悬疑-平台签约评估框架
Evaluates urban mystery manuscripts for platform serialization potential using strict, conservative criteria that prioritize negative evidence. Routes through the core platform ass…
AI / ML
build-eval
Designs and implements rigorous evaluations for LLM agents, multi-agent systems, skills, and prompts—covering frameworks like DeepEval, Braintrust, and RAGAS plus metrics such as p…
Engineering
choose-stack
Evaluates and selects technologies using a weighted decision matrix. Activates for technology choices, framework comparisons, technical alternatives, decision matrices, stack evalu…
Productivity
opc
OPC — One Person Company. Digraph-based task pipeline with independent multi-role evaluation. Builds, reviews, analyzes, and brainstorms with specialist agents. Every path ends wit…
Testing
paper-autoraters
Execute the four paper-quality autoraters: Citation F1, Literature Review Quality (6-axis), SxS Overall Paper Quality, and SxS Literature Review Quality. Triggers on requests to sc…
AI / ML
analyze-generative-diffusion-model
Evaluates pre-trained generative diffusion models by computing FID, IS, CLIP scores, precision/recall, inspecting noise schedules, extracting attention maps, and probing latent spa…
Business
vendor-analysis
Comprehensive vendor evaluation covering financial stability, operational metrics, technical alignment, risk factors, and growth opportunities. Delivers HTML reports, Excel data, a…
AI / ML
langsmith-observability
LLM observability platform for tracing, evaluation, and monitoring. Use when debugging LLM applications, evaluating model outputs against datasets, monitoring production systems, o…
Testing
gloss-review
Stage 8 interlinear evaluator — audit completeness and correctness of candidate interlinear artifacts, benchmark them against ground truth, and decide pass/block/promote. Use for e…
Research
ai-scientist-evaluator
Critically review, score, compare, and rank AI scientist outputs for biology, bioinformatics, and life science research. Evaluate notebooks, code, analyses, manuscripts, and report…
Business
business-ideation
A general-purpose skill for developing business or service ideas through three phases—divergence, deepening, and evaluation—usable with any theme or industry. Maintains notes as th…
Testing
eval-robuste
Measures variance of a stochastic evaluator by running N parallel executions, aggregates scores deterministically, and flags noise versus regression against a baseline. Use when re…
AI / ML
model-migration-plan
Creates a phased migration plan for swapping LLM models in production, including eval gates, shadow and canary stages, prompt adaptation, and rollback criteria. Use when deprecatin…
Design
visual-plan
Reviews plan.md against an existing resolved design.md, then runs generate+evaluate loops to produce iters/*.png with status completed/partial/aborted. Requires finalized visual-sp…
Research
deeptutor
Academic advisor evaluation system that investigates professors worldwide. Searches publications, maps co-author networks, tracks student outcomes, classifies advisor types, and as…
Productivity
talk-moss-skills-team-workflow
Guides teams in structuring skill governance: decomposition, ownership, versioning, evaluation scenarios, quality reviews, and lifecycle maintenance. Use when shifting from ad-hoc …
AI / ML
tiger
Scans LLM call sites in any codebase to improve code quality, validate business value through user-task testing, and benchmark model performance vs cost. Stores findings in an Obsi…
Testing
weak-agent-test
Executes adversarial testing of document agents across six real-world scenarios—five editing tasks and one multi-column poetry authoring—measuring tool efficiency, rendering fideli…
Testing
evaluate-asset
Scores a generated visual asset against its brief for fidelity and brand alignment, returning a verdict plus diagnosis inside an evaluation loop. Works on one asset per cycle; excl…
Research
agsy-data-and-model-evaluation
Evaluate models and analyze results for Agricultural Systems manuscripts. Includes observed-vs-simulated comparisons, fit statistics, sensitivity and uncertainty analysis, and trad…
Business
evaluate-submission
Run complete TAM partnership evaluations on WooCommerce Marketplace submissions: analyze threads, detect gaps, score across six fit dimensions with adversarial checks, draft respon…
Productivity
kpi-skill
给数字生命做绩效考核的赛博HR系统。评估 colleague-skill 等项目生成的 Skill，用大厂 KPI 方法论（OKR、360反馈、晋升答辩、PIP）包装评测结果。内置 linkskill-bench 评测引擎。| Cyber HR system that evaluates digital personas using big-tech KPI…
Research
oss-kpi-evaluation
Evaluates open source projects against KPIs via multi-agent research. Pulls live web data on community health, maintenance, security, documentation, adoption, and code quality for …
Business
workflow-mieterhoehung-entscheidung
Rent increase decision workflow for rental and condominium law: evaluates consent requirements, caps, blocking periods, and objections with cold-start, deadline checks, evidence ma…
Showing the top 60 of 976. See the full list →