synalinks-skills

Why we included this project

Teams building neuro-symbolic programs with Synalinks run into a familiar failure mode: the framework follows Keras conventions closely enough that general-purpose coding agents tend to reuse LangChain or DSPy idioms and produce code that looks right but does not work. This repo ships a single `synalinks` skill in the open Agent Skills format, which any agent that reads SKILL.md, such as Claude Code, Codex, or OpenCode, can auto-activate, keeping the agent on idiomatic Synalinks usage. The skill is a scannable overview, with one section per area of the framework (core ops, programs, modules, control flow, agents, RAG knowledge, training, rewards, optimizers), and every section links to a runnable script and its captured run log. If you write Synalinks programs with an agent, installing these skills reduces how often you have to fix its output by hand; if you are just trying out the framework, those runnable examples also double as a guided tour of what it covers.