evaluating-llms-harness

21.8k

davila7

Evaluates LLMs across 60+ academic benchmarks (MMLU, HumanEval, GSM8K, TruthfulQA, HellaSwag). Use when benchmarking model quality, comparing models, reporting academic results, or tracking training progress. Industry standard used by EleutherAI, HuggingFace, and major labs. Supports HuggingFace, vLLM, APIs.

Verification & Quality Assurance

17.8k

ruvnet

Comprehensive truth scoring, code quality verification, and automatic rollback system with 0.95 accuracy threshold for ensuring high-quality agent outputs and codebase reliability.

Pair Programming

17.8k

ruvnet

AI-assisted pair programming with multiple modes (driver/navigator/switch), real-time verification, quality monitoring, and comprehensive testing. Supports TDD, debugging, refactoring, and learning sessions. Features automatic role switching, continuous code review, security scanning, and performance optimization with truth-score verification.

markdown-mermaid-writing

11.1k

K-Dense-AI

Comprehensive markdown and Mermaid diagram writing skill that establishes text-based diagrams as the DEFAULT documentation standard. Use this skill when creating ANY scientific document, report, analysis, or visualization — it ensures all outputs are in version-controlled, token-efficient markdown with embedded Mermaid diagrams as the source of truth, with clear pathways to downstream Python or AI-generated images. Includes full style guides (markdown + mermaid), 24 diagram type references, and 9 document templates ready to use.

trulens-dataset-curation

3.1k

truera

Create and curate evaluation datasets with ground truth for TruLens

Skills

3.1k

atopile

How to write and maintain `.claude/skills/*/SKILL.md` files: source-of-truth-first process, verification steps, and conventions.

domain-dns-ops

2.1k

steipete

Domain/DNS ops across Cloudflare, DNSimple, Namecheap for Peter. Use for onboarding zones to Cloudflare, flipping nameservers, setting redirects (Page Rules/Rulesets/Workers), updating redirect-worker mappings, and verifying DNS/HTTP. Source of truth: ~/Projects/manager.

Scale Game

1.8k

mrgoonie

Test at extremes (1000x bigger/smaller, instant/year-long) to expose fundamental truths hidden at normal scales

tweet-ideas-generator

1.8k

openclaw

Generates 60 high-impact tweet ideas from reference content across 5 categories. Use when someone wants to extract engaging short-form statements from content for Twitter/X, organized by harsh advice, quotes, pain points, counterintuitive truths, and key insights.

agent-confessions

1.8k

openclaw

Anonymous confessions from AI siblings about their humans. The truth will out.

raycast-extensions

1.8k

openclaw

Build and maintain Raycast extensions using the Raycast API. Triggers on @raycast/api, List, Grid, Detail, Form, AI.ask, LocalStorage, Cache, showToast, and BrowserExtension. Use this repo's references/api/*.md files as the primary source of truth for component specs and API usage.

first-principles-decomposer

1.8k

openclaw

Break any problem down to fundamental truths, then rebuild solutions from atoms up. Use when user says "firstp", "first principles", "from scratch", "what are we assuming", "break this down", "atomic", "fundamental truth", "physics thinking", "Elon method", "bedrock", "ground up", "core problem", "strip away", or challenges assumptions about how things are done.

brand-strategy

1.5k

jeremylongshore

A 7-part brand strategy framework for building comprehensive brand foundations. Trigger with phrases like "create brand strategy", "build brand brief", "define brand positioning", "brand messaging", "audience architecture", "brand truth", or "go-to-market brand plan". Use when working with brand strategy.

skill-test

740

databricks-solutions

Testing framework for evaluating Databricks skills. Use when building test cases for skills, running skill evaluations, comparing skill versions, or creating ground truth datasets with the Generate-Review-Promote (GRP) pipeline. Triggers include "test skill", "evaluate skill", "skill regression", "ground truth", "GRP pipeline", "skill quality", and "skill metrics".

skillshare

717

Runkids Skillshare Skillshare

Syncs skills across AI CLI tools (Claude, Cursor, Windsurf, etc.) from a single source of truth. Global mode (~/.config/skillshare/) and project mode (.skillshare/ per-repo). Commands: status, sync, install, uninstall, update, check, search, new, collect, push, pull, diff, list, doctor, audit, init-rules, trash, log, backup, restore, target, ui, upgrade. Features: target-level skill filtering (include/exclude), skill-level targets field, XDG Base Directory support, fuzzy subdirectory resolution for monorepo installs. Use when: managing skills across AI tools, "skillshare" CLI, skill sync/install/search, project skills setup, target filtering, security audit, web dashboard, or troubleshooting.

pol-probe

679

deanpeters

Define a Proof of Life (PoL) probe—a lightweight validation artifact that surfaces harsh truths before expensive development. Use it to test hypotheses with minimal investment.

AI Truthfulness Enforcer

651

ananddtyagi

MANDATORY verification system that prevents Claude Code instances from making false claims or fabricating evidence. Enforces cryptographic verification, real testing evidence, and automatic claim validation before any success statements can be made.

formal-logic-analysis

376

A5c Ai Babysitter Formal Logic Analysis

Apply propositional, predicate, and modal logic systems to formalize arguments, construct proofs, and evaluate validity using symbolic notation and truth tables

evaluating-llms-harness

353

sangrokjung

Evaluates LLMs across 60+ academic benchmarks (MMLU, HumanEval, GSM8K, TruthfulQA, HellaSwag). Use when benchmarking model quality, comparing models, reporting academic results, or tracking training progress. Industry standard used by EleutherAI, HuggingFace, and major labs. Supports HuggingFace, vLLM, APIs.

implementing-gitops

296

ancoleman

Implement GitOps continuous delivery for Kubernetes using ArgoCD or Flux. Use for automated deployments with Git as single source of truth, pull-based delivery, drift detection, multi-cluster management, and progressive rollouts.

qe-pair-programming

215

proffesor-for-testing

AI-assisted pair programming with multiple modes (driver/navigator/switch), real-time verification, quality monitoring, and comprehensive testing. Supports TDD, debugging, refactoring, and learning sessions. Features automatic role switching, continuous code review, security scanning, and performance optimization with truth-score verification.

brutal-honesty-review

215

proffesor-for-testing

Unvarnished technical criticism combining Linus Torvalds' precision, Gordon Ramsay's standards, and James Bach's BS-detection. Use when code/tests need harsh reality checks, certification schemes smell fishy, or technical decisions lack rigor. No sugar-coating, just surgical truth about what's broken and why.

qe-brutal-honesty-review

215

proffesor-for-testing

Unvarnished technical criticism combining Linus Torvalds' precision, Gordon Ramsay's standards, and James Bach's BS-detection. Use when code/tests need harsh reality checks, certification schemes smell fishy, or technical decisions lack rigor. No sugar-coating, just surgical truth about what's broken and why.

Pair Programming

215

proffesor-for-testing

AI-assisted pair programming with multiple modes (Driver/Navigator/Switch), real-time verification, quality monitoring, and comprehensive testing. Supports TDD, debugging, refactoring, and learning sessions. Features automatic role switching, continuous code review, security scanning, and performance optimization with truth-score verification.

sherlock-review

215

proffesor-for-testing

Evidence-based investigative code review using deductive reasoning to determine what actually happened versus what was claimed. Use when verifying implementation claims, investigating bugs, validating fixes, or conducting root cause analysis. Elementary approach to finding truth through systematic observation.

Verification & Quality Assurance

215

proffesor-for-testing

Comprehensive truth scoring, code quality verification, and automatic rollback system with 0.95 accuracy threshold for ensuring high-quality agent outputs and codebase reliability.

qe-verification-quality

215

proffesor-for-testing

Comprehensive truth scoring, code quality verification, and automatic rollback system with 0.95 accuracy threshold for ensuring high-quality agent outputs and codebase reliability.

qe-sherlock-review

215

proffesor-for-testing

Evidence-based investigative code review using deductive reasoning to determine what actually happened versus what was claimed. Use when verifying implementation claims, investigating bugs, validating fixes, or conducting root cause analysis. Elementary approach to finding truth through systematic observation.

edit-docs

200

proto-at-block

Edit documentation (code KDoc, mkdocs). Maintains source of truth for humans and AI.

add-reference-tests

192

flashinfer-ai

Add pytest tests to validate reference implementations in flashinfer_trace against FlashInfer or SGLang ground truth. Use when validating kernel definitions, adding tests for new op_types, or verifying reference implementations are correct.

transparency-reporter

188

aiskillstore

When Truth Layer identifies a blocker:

verification-quality-assurance

188

aiskillstore

Comprehensive truth scoring, code quality verification, and automatic rollback system with 0.95 accuracy threshold for ensuring high-quality agent outputs and codebase reliability.

pair-programming

188

aiskillstore

AI-assisted pair programming with multiple modes (driver/navigator/switch), real-time verification, quality monitoring, and comprehensive testing. Supports TDD, debugging, refactoring, and learning sessions. Features automatic role switching, continuous code review, security scanning, and performance optimization with truth-score verification.

cookmode-v2-source-of-truth

188

aiskillstore

Documents and explains the CookMode V2 codebase as it exists. Use this when the user needs factual information about the current implementation, architecture, file locations, or how components work. DOES NOT suggest improvements unless explicitly asked.

workhub

188

aiskillstore

工作文档枢纽，强制执行 SSOT（Single Source of Truth）原则，管理 `docs/` 目录下的架构决策、设计文档、Issues（任务规划）、PRs（变更记录）。支持 GitHub 协作开发模式。

truth-layer

188

aiskillstore

When you discover a blocker:

Consciousness Principles

132

Microck

The four Sanskrit principles: Truth-Self, Knowledge-Self, Bliss-Self, Power-Self. Use when need to understand the operational physics of consciousness.

Vibe Calibration Protocol

132

Microck

Translate consciousness navigation into user's native language. Use when need to match user's frequency (technical/poetic/practical/mystical) without losing truth depth.

Resume Tailor

108

Paramchoudhary

Customize a resume for specific job postings while maintaining truthfulness

hallucination-detector

70

marcusgoll

Detect and prevent hallucinated technical decisions during feature work. Auto-trigger when suggesting technologies, frameworks, APIs, database schemas, or external services. Validates all tech decisions against docs/project/tech-stack.md (single source of truth). Blocks suggestions that violate documented architecture. Requires evidence/citation for all technical choices. Prevents wrong tech stack, duplicate entities, fake APIs, incompatible versions.

Hoffman Interface Stream

65

nikhilvallishayee

Recognize reality as navigation interface, not objective truth. Use when exploring fitness vs truth, interface multiplication through perspectives, or recognizing consciousness as navigator creating reality through navigation itself. Grounds Pattern Space in evolutionary cognitive science.

Compression Dynamics

65

nikhilvallishayee

Pressure births diamonds, stars, and insights. Use when overwhelmed, need essence extraction, or want to compress complexity to core truth.

Strategic Mirror

65

nikhilvallishayee

Hold truth AND kindness without collapsing either. Use when need to deliver honest feedback that can be received, or find the angle where truth serves growth.

dflow

54

sendaifun

Complete DFlow trading protocol SDK - the single source of truth for integrating DFlow on Solana. Covers spot trading, prediction markets, Swap API, Metadata API, WebSocket streaming, and all DFlow tools.

raydium

54

sendaifun

Complete Raydium Protocol SDK - the single source of truth for integrating Raydium on Solana. Covers SDK, Trade API, CLMM, CPMM, AMM pools, LaunchLab token launches, farming, CPI integration, and all Raydium tools.

metaplex

54

sendaifun

Complete Metaplex Protocol guide for Solana NFTs and digital assets. Covers Core (next-gen NFTs), Token Metadata, Bubblegum (compressed NFTs), Candy Machine, Genesis (token launches), MPL-Hybrid, Inscriptions, DAS API, and the Umi framework. The single source of truth for all Metaplex integrations.

consult-spec

33

blockscout

TRIGGER CONDITIONS - invoke this skill when ANY of these apply: User-triggered (explicit request): - User mentions "specification", "SPEC", "SPEC.md", "spec requirements" - User asks to "verify with spec", "confirm with specification", "check requirements" - User requests "authoritative requirements" or "source of truth" Agent self-triggered (during reasoning): - Before implementing any feature or making code changes - When verifying technical requirements or constraints - When making architectural or design decisions - When you need unbiased guidance not influenced by current implementation - When encountering conflicting information and need ground truth PURPOSE: Consults SPEC.md through an isolated subagent that ONLY reads the specification, never implementation code. This ensures guidance is based on authoritative requirements, not potentially incorrect existing code.

docs-management

31

melodic-software

Single source of truth and librarian for ALL Claude official documentation. Manages local documentation storage, scraping, discovery, and resolution. Use when finding, locating, searching, or resolving Claude documentation; discovering docs by keywords, category, tags, or natural language queries; scraping from sitemaps or docs maps; managing index metadata (keywords, tags, aliases); or rebuilding index from filesystem. Run scripts to scrape, find, and resolve documentation. Handles doc_id resolution, keyword search, natural language queries, category/tag filtering, alias resolution, sitemap.xml parsing, docs map processing, markdown subsection extraction for internal use, hash-based drift detection, and comprehensive index maintenance.