Build a research brief you can actually check
Turn three source passages into a claim ledger, a restrained recommendation and a one-page brief. Includes a complete teaching packet and reusable prompts.
Work through the example ↗The practical AI decision desk
Useful work.
Clear evidence.
Choose a task, check the evidence and measure the result. Worked guides, an editable cost model and a tool shortlist help you make the next decision.
Start with the job
What are you working on?
A good tool earns its place in your workflow.
Build somethingApps, code & developmentFind an answerResearch & source checkingMake it visualImages, UI & presentationsGet time backWorkflows & integrations
No signup. Save locally. Compare before you commit.
● 832 tools in the directory4 worked guides & templatesLatest source record 13 Oct 2026Our publishing method ↗
Start here / Learn a useful method
Go from a question to a checked result.
Original teaching examples, complete exercises and reusable templates.
Compare the cost of finished work, not the subscription
Build an editable monthly cost model with retries, review time and setup. Follow a hypothetical USD example and calculate a useful break-even point.
Work through the example ↗Run a small AI coding pilot with a reviewable result
Use a two-tool compare action as a bounded coding exercise. Includes acceptance criteria, a reference function, runnable tests and a diff-review checklist.
Work through the example ↗Make a small automation safe to retry
Design one draft per request version, recognize duplicate deliveries and stop conflicting inputs. Includes runnable sample logic and a production handoff checklist.
Work through the example ↗Your assumptions / Your decision
A cheap plan can make expensive work.
Compare recurring charges, review time and setup. Record your acceptance task, source checks and corrections in a worksheet you can download.
EXAMPLE: $220 vs $85 / MONTHOpen the decision lab Fictional inputs. Runs in your browser.
Find candidates. Then run your own trial.
These source briefs are unreviewed discovery records, not tested recommendations. Search, save candidates and verify the original source.
39 tools
ECCBuild & code
Optimizes performance in coding tasks using skills, instincts, memory, and security. Focuses on research-driven development.
AI NotesResearch & learn
Generates source-grounded AI notes with citations from provided text. Open source, built on Hugging Face platform.
histerResearch & learn
Search engine implementation, lightweight, customizable, supports various backends and plugins for expanding functionality.
MarigoldResearch & learn
Generates depth, surface normals, and albedo maps from a single image. Research tool, lacks production validation.
OpenResearchBuild & code
Turns coding agents into research agents. Open-source platform for enhancing development productivity in research contexts.
ClaudeWrite & edit
System prompts extracted from multiple AI models for analysis and testing purposes. Not fully vetted for comprehensive use.
hyperresearchResearch & learn
Agent-based tool for collecting, searching, and organizing web research into a searchable knowledge base.
OpenFluxResearch & learn
Research tool facilitates TCP tunneling; supports pluggable transports for network stack analysis.
MathModelAgentResearch & learn
Agent automates mathematical modeling tasks and generates submission-ready papers. Targets academic and research applications.
deer-flowBuild & code
Open-source tool for extended task management using research, coding, and creation, combined with supportive subagents and sandboxes.
Open ScienceBuild & code
Open-source AI research workbench supporting local execution, model agnosticism, and multi-platform integration.
rare-disease-hackathonResearch & learn
Hackathon focused on real-world applications for rare disease research. Participants develop solutions using collaborative resources.
exploitariumResearch & learn
Archive of public exploit PoCs and vulnerability research. Unreported vulnerabilities. Users can report and claim CVEs.
scientific-agent-skillsResearch & learn
Library of 165 skills for AI agents in scientific domains, leveraging extensive databases for research and development.
academic-research-skillsResearch & learn
Tool for managing academic research process: research, write, review, revise, finalize tasks.
user-scannerResearch & learn
OSINT tool for data extraction from email/username. Analyzes over 465 scan vectors, suitable for security research.
Open SEOResearch & learn
Open source SEO tool suite. Aims to provide alternative functionalities for site analysis, backlink tracking, and keyword research.
marinResearch & learn
Open-source framework targeting foundation model research and development. Active community contributions, but lacks distinct advantages.
openhumanResearch & learn
Local-first AI memory management, orchestrates workflows, and manages agent fleets for personal data and tasks.
OpenSEOResearch & learn
Open-source platform for SEO tasks like keyword research, backlinks, rank tracking, and site audits. Billed by usage.
firecrawlBuild & code
Context API for web search and scraping. Supports large-scale data interaction. Suitable for developers needing web automation.
Lev8Research & learn
Lev8 uses AI agents for real-time web search, data enrichment, intent tracking, and multi-channel outreach.
FlashKDAResearch & learn
FlashKDA implements Kimi Delta Attention kernels focusing on high-performance processing. Suitable for research, not yet proven in production scenarios.
ICML 2026 ReproducerResearch & learn
Reproduces papers from ICML 2026 using an AI agent. Currently in development, functionality untested at scale.
No tools match yet.
Try another search or reset the filters to explore the directory.
Follow the evidence
Each source brief links to its origin. Use the research guide to turn claims into a checked evidence ledger.
Choose on your terms
Compare source descriptions side by side, then use the decision lab to model costs and record a trial.
Try before you trust
Set an acceptance condition before trying a tool. Record review time, errors and the result before adopting it.