New: Skill bounties are live. Post a request, fund the bounty, and creators compete for 7 days to build it -> See open bounties
THE AGENSI STORE
28 skills found
by loreto
Published AI benchmarks measure brains in jars. They test models in isolation or within a single reference harness — and then attribute all performance to the model. This skill teaches you to decompose agent performance into its two actual components: model capability and harness multiplier. The result is evaluations that predict real-world behavior instead of benchmark theater.
by loreto
RAG fails quietly. It retrieves documents, returns confident-looking answers, and misses the question entirely — because the question required connecting facts across documents, reasoning about sequence, or tracing causation. This skill gives you a five-question diagnostic checklist that classifies any failing query as either RAG-safe or structurally RAG-incompatible, then maps it to the specific failure pattern and the architectural fix that resolves it.
by Aman Garg
AI drafts sound like AI wrote them. This skill learns how you actually write and makes every email, message, and doc sound like you typed it yourself.
by Aman Garg
Your agent writes the code but never commits or documents what it learned. This skill handles both automatically after every task.
by Aman Garg
AI output is full of padding, repeated information, and filler phrases. This skill cuts all of that while keeping the specifics and nuance that actually matter.
by Matthew King
Most 'migrations' rewrite your tests from scratch. This one maps your existing keywords and locators automatically.
by Aman Garg
Stop burning expensive model tokens on repetitive subtasks. This skill delegates mechanical work to cheaper models and writes handoff snapshots so you never lose context switching between sessions.
by Joh
Use this skill when the user is facing a bug in an Angular application, including issues with components, services, RxJS, HTTP calls, tests, or unexpected behavior.
by Matthew King
SeleniumLibrary is the flakiest part of your Robot Framework suite. This skill swaps it for Browser library, structure intact
by Matthew King
Most 'TDD' skips the failing test. This mentor enforces Red before Green, even on legacy code.
by Matthew King
Most quality strategies are generic templates nobody follows. This one is built from your actual repo
A repo visitor decides in about ten seconds whether your project solves their problem — and most READMEs spend those ten seconds on installation instructions instead of answering 'what does this actually do'
'Should work now' is the most expensive sentence in software — it means the change was reasoned about, not observed, and the next person to touch this code will pay for that assumption
The gap between 'the code works' and 'this branch is safe to merge' is where teams bleed quality
Find the LLM integration code that will not survive a provider being pulled or going down. Flags single-provider lock-in with no alternative, calls with no failover branch, missing timeouts, retries with no limit or backoff, no degraded-mode default, and hardcoded endpoints with no alternate. This is about the model going away, not the model declining.
Review a database schema, queries, or migration for the mistakes that get expensive in production — bad table design, missing or wrong indexes, slow and N+1 queries, SQL injection, and migrations that lock or break prod. Engine-aware (PostgreSQL, MySQL, SQLite, SQL Server), it runs an ordered review and returns a PASS/REVIEW/BLOCK verdict with prioritized fixes. Schema mistakes are the most expensive kind — this catches them before they ship.
by Boubacar B
Your AI generates UI components. This skill makes those components production-ready. Spacing, states, accessibility, responsiveness -- baked into every component Claude writes, without you specifying each one from scratch.
by Matthew King
Your multi-step form works. It just feels cheap. This skill fixes the feel, not the logic
Stop leaving your AI startup exposed to malicious users trying to steal your proprietary system prompts or bypass your paywalls. The AI Prompt Injection Defense Shield is an automated code review agent that deeply analyzes your Next.js or Python backend, instantly detecting insecure LLM input fields, un-sanitized API data streams, and weak prompt boundaries. By automatically generating the exact copy-paste code patches required to harden your AI wrapper against the latest OWASP top 10 LLM vulnerabilities, this skill allows solo developers and indie hackers to confidently launch their SaaS without the fear of massive, unexpected API billing spikes or catastrophic data leaks.
Audit any page, article, or doc for how likely AI answer engines — ChatGPT, Perplexity, Google AI Overviews, Claude — are to surface and cite it. Checks answer-first extractability, question-shaped structure, quotable units, factual specificity, sourcing, schema, and freshness, then returns a PASS/REVISE verdict with the prioritized fixes. This is Answer Engine Optimization, not classic SEO — it optimizes content to be quoted inside AI answers.
Messy, insecure, unfixable — that's what AI builds without architecture. This file is the architecture: 10 years of senior judgement on rendering, caching, security and SEO, so your agent builds it right from day one.
by Timoranjes
A skill that enables AI coding agents to maintain context across session boundaries. When a coding session reaches its token limit, this skill saves the full state — files in progress, decisions made,
by Boubacar B
Claude can write frontend code. What it needs is a design system vocabulary to write CONSISTENT frontend code. This skill gives it that vocabulary, so every button, form, and layout follows the same rules across your entire app.
by Boubacar B
Claude defaults to complex. This skill does not let that happen. Before your session ships a 400-line solution to a 20-line problem, the simplicity filter stops it and asks: does this actually need to be this complex?