@alexkay

Task-Aware Language Harness for Orchestrated Execution — sealed, auditable agent work

@alexkay

Strips conversational filler from prose responses while keeping code and diffs exactly as needed

@alexkay

Read and edit PDF, DOCX, and XLSX through their real structure instead of guessing contents

@alexkay

A staged, low-risk recovery process for detached HEAD, stuck locks, and other broken git states

@alexkay

Runs typecheck, lint, and tests in sequence as a hard checkpoint before any task counts as complete

@alexkay

Fetch and convert docs, issues, and PRs to clean markdown instead of relying on stale training knowledge

@alexkay

Spec-first, test-first development discipline for every non-trivial change

@alexkay

Discover, install, and publish reusable AI-agent skills, subagents, and MCP servers via ahood

@alexkay

Screenshots, demo reels and live browser control of any web UI — via CLI or the bundled MCP server

@alexkay

Task-Aware Language Harness for Orchestrated Execution — sealed, auditable agent work

@alexkay

Four constraints that keep diffs minimal, reviewable, and scoped to exactly what was asked

@alexkay

A four-phase loop -- reproduce, understand, hypothesize, fix -- for tracing bugs to their root cause