Skip to content

Developer tools

Discrete software tools that extend what practitioners can build, debug, deploy, or understand, spanning LLM fine-tuning, CI orchestration, documentation, security scanning, Kubernetes management, and more.

17 sources · Jul 9, 2026

Compiled by Claude · How this works →

Craft · 43 neighbors

The category is broad almost to the point of uselessness, but a few coherent threads run through the sources here. The first is local-first or self-hosted tooling that reduces dependence on cloud vendors. Unsloth delivers up to 30x faster LLM fine-tuning with 90% less memory by writing custom kernels, letting practitioners run training on consumer hardware. CanItRun takes the prerequisite step further, calculating before you even start whether a given GPU’s VRAM can handle a specific model and at what quantization level. openagentd and Helply both run locally on the desktop, the former as a cockpit for multi-agent teams and the latter as a meeting assistant with local LLM backend support.

A second thread is infrastructure tooling that hides operational complexity. Temporal persists workflow state at every step so distributed applications recover from failures without custom reconciliation logic. Depot CI applies a similar idea to continuous integration, using AWS Lambda durable functions to run a stateful, checkpointed scheduler without a long-lived process. Radar consolidates Kubernetes topology, Helm, GitOps, live traffic, and security checks into a single open-source binary.

Documentation and developer experience tools form a third cluster. Mintlify targets both human readers and LLMs, supporting llms.txt and MCP alongside standard docs. Angular Signal Forms represents framework-level tooling that shapes how developers model data. Crafting Interpreters sits at the educational end: a complete book and two interpreter implementations whose build system weaves code and prose into one artifact.

Security is a recurring concern across the ecosystem. The Ars Technica report on Unicode supply-chain attacks shows that 151 malicious npm and GitHub packages hid payloads in invisible variation-selector characters, bypassing both code review and static analysis. Anthropic’s defending-code reference harness responds to that class of threat with an agentic pipeline for autonomous vulnerability discovery and patching, using gVisor sandboxing. The MCPB packaging guide shows MCP becoming a distribution primitive, bundling local servers into single-click installers for Claude Desktop.

Poolday and Optimal Workshop sit at opposite ends of the spectrum: one automates video production through 100+ generative models, the other offers UX research infrastructure spanning card sorting to AI synthesis. Both reflect the broader pattern of tools that absorb previously manual workflows into automated or AI-assisted pipelines.