3,748 tools found
MarketFish simulates a product launch with 128+ AI consumer agents shopping across 30 rounds, grounded in 6 academic multi-agent-market papers.
CLI where an AI agent writes and self-maintains a linked wiki for your codebase or personal knowledge, with a visualizer for humans.
Z-Jail is an ~81 KiB, dependency-free multi-layer sandbox that runs native Linux binaries behind seven ordered isolation layers.
nanoeuler is a GPT-2-class language model written from scratch in C and CUDA, covering tokenizer, pretraining, SFT and hand-verified backward passes.
Bash4LLM+ is a dependency-free Bash CLI and TUI for calling OpenAI-compatible LLM APIs, with an optional encrypted key vault.
Reconstructs the workflow behind a past coding-agent artifact by mining the local ax graph of commits, sessions, and skills.
Runs a structured interview to fully understand a project before any code - writes the result as a self-contained prompt.md for later.
Run a scoped adversarial static bug hunt and maintain a persistent bugs.md report with exact locations, triggers, severity, and fix suggestions.
Ship polished web UIs with mandatory render critique, overflow gates, and design-system rigor to avoid generic AI output.
Shared vocabulary skill for designing deep modules - small interface, large implementation - with strict glossary and tests.
Build a cited competitor comparison with scoped research, explicit unknowns, a feature and pricing matrix, and local HTML and CSV reports.
cron-doctor is a free Claude Code skill that catches impossible or misleading cron schedules before they ship to production.
