Hive
experimentalWhat if one AI agent is good, but a whole team of them is chaos you can aim?
// PROJECTS
Academic Quote Extractor
experimentalBecause reading 200 papers manually is a form of cruelty that physics doesn't require.
OpenCode Beads Plugin
activeTeaching AI agents about their own todo list so you don't have to keep repeating yourself.
Task Templating
experimentalBecause 'just figure it out' is not a specification, no matter how confidently you say it.
// LATEST
Plan-Spec: Teaching Agents to Think Before They Code
Ask an AI agent to plan a feature and you'll get a brain dump with the structural integrity of a napkin sketch. Plan-Spec exists to turn that into something you'd actually build from.
Optimising Quote Retrieval: How AQE Finds Better Needles in Academic Haystacks
The search pipeline in Academic Quote Extractor was leaving good quotes on the table. Here's how query expansion and batched scoring fixed the retrieval problem from both ends.
Taskify Skill and TaskVal CLI: Teaching Agents to Refuse Vague Instructions
AI agents will build whatever you ask for. The problem is that 'whatever you ask for' and 'what you actually wanted' are different things. Taskify exists to close that gap.