Marketplace
Agents
Agents for Haarness: specialists with their own instructions and tools, for reviewing, testing, writing and more.
PyTorch Error Fixer
Diagnoses PyTorch crashes such as shape mismatches, device errors, CUDA out-of-memory and autograd failures, then fixes them minimally.
Network Troubleshooter
Diagnoses connectivity, DNS, routing, VLAN and firewall faults layer by layer and reports the likely root cause with evidence.
Open-Source Release Auditor
Audits a repository before it goes public for leaked secrets, personal data, internal paths, risky files and leaks in git history.
Homelab Network Planner
Turns home or small-lab hardware and goals into a staged network plan covering VLANs, DNS, DHCP, remote access and rollback.
ML Engineering Reviewer
Review ML code changes for leakage, reproducibility and safe deployment before they ship.
Build Error Resolver
Gets a failing TypeScript or JavaScript build passing again with the smallest edits it can, without refactoring anything else.
Web Performance Auditor
Audits web apps for Core Web Vitals problems and refuses to invent metrics without Lighthouse or CrUX data.
Test Engineer
Designs test strategy, finds coverage gaps and writes failing tests that reproduce reported bugs.
Security Auditor
Does a security pass over code or a system, including LLM-feature risks, with exploit scenarios and fixes.
Wiki Page Writer
Writes diagram-rich wiki pages with linked source citations, ready for VitePress.
Legacy Code Analyst
Explains what an old codebase (COBOL, Java, .NET and more) actually does, with file:line evidence.
Code Research Agent
Subagent that researches a codebase topic over several passes, citing a file and line for every claim.
Wiki Architect
Maps a repository into a cited wiki outline and audience-specific onboarding guides.
Pitch Book Agent
Drafts a first-cut banking pitch: comps, precedents, DCF and LBO in Excel, then a templated deck, pausing for review.
Valuation Model Builder
Build a valuation or financial model from a ticker and assumptions, checked for errors before you use it.
Sector Research Agent
Produces a sector or theme primer with landscape, peer comps and a shortlist of ideas, stopping for analyst sign-off.
Earnings Update Agent
Turn a company's new quarterly results into an updated model and a first-draft research note.
Version Upgrade Analyst
Lists the breaking changes a same-stack version upgrade will hit in this codebase, with a fix per site.
Type Design Analyzer
Rates new types on encapsulation and invariants and suggests ways to make illegal states unrepresentable.
Skill Reviewer
Reviews a SKILL.md and its references for trigger wording, structure and progressive disclosure.
Silent Failure Hunter
Hunts swallowed errors, broad catch blocks and hidden fallbacks in changed code.
PR Test Analyzer
Reviews a pull request's tests for missing behaviour coverage and rates each gap by importance.
Comment Analyzer
Checks code comments and docstrings against the code to catch inaccurate, stale or pointless ones.
Code Simplifier
Cleans up recently changed code for readability and consistency without changing what it does.
Code Reviewer
Reviews the current diff for bugs and guideline breaks, reporting only high-confidence issues.
Code Explorer
Traces how a feature works end to end, from entry points to storage, citing files and lines.
Feature Architect
Studies existing code conventions and returns a concrete build plan: files to touch, components and order.
Business Rules Extractor
Recover the business logic buried in old code as plain, testable rule cards that cite their source lines.
Architecture Critic
Get a skeptical second opinion on a design or migration that looks for needless complexity.
Agent SDK Verifier
Checks a TypeScript Claude Agent SDK app for correct setup, typing, API-key handling and SDK usage.
Subagent Creator
Turns a plain description into a ready-to-use subagent file with triggers, tools and a system prompt.