Scattered context
Repository decisions and rules end up split across files, chats, and prompts.
lemon-code · practical kit
Pre-launchTemplates and routines for delegating coding tasks, reviewing work across models, and closing each change with a local quality gate.
The problem
Different agents need explicit context, boundaries, and completion criteria. Without them, teams repeat instructions and discover missing work late.
Repository decisions and rules end up split across files, chats, and prompts.
A second pass helps when it checks concrete risks and points to evidence.
Lint, types, tests, and build checks belong in the workflow before delivery.
Inside the kit
Editable materials to adapt to your repository and the tools you already use.
01
AGENTS.md and CLAUDE.md templates for scope, conventions, and execution boundaries.
02
Structures for splitting tasks, supplying context, and setting acceptance criteria.
03
Checklists for an independent second pass across models and tools.
04
A practical sequence for lint, typecheck, tests, and build before delivery.
05
Checks for connecting MCP servers with clear scope and configuration.
06
A routine for keeping parallel tasks isolated and diffs easy to review.
Who it is for
From the lemon-code series
The kit organizes practices discussed in the lemon-code series into templates and checklists. Explore the posts before joining the list.
Browse lemon-code postsFrequently asked questions
Not yet. This page is in pre-launch mode and the list is for launch updates.
No. Signup is free and does not start a purchase or subscription.
Examples are designed for Claude Code, Codex, and similar coding-agent workflows.
No. The templates are starting points to adapt to your repository's rules, tools, and criteria.