Agent Onboarding Kit Builder
An online tool that interviews you once about your business, then auto-generates the AGENTS.md, CLAUDE.md, and onboarding docs your AI coding tools need to produce code that matches your stack, conventions, and review standards.
Small software teams and senior engineers who want their AI coding assistant to stop suggesting code that doesn't match the codebase
- 30-question onboarding interview covering stack, conventions, review rigor, deployment, and 'do not touch' zones
- Generated AGENTS.md / CLAUDE.md starters plus editable checklists you can copy into repos
- Style guide distillation — turns your style docs and PR comments into a single bundle
- 'Did the agent follow it?' tracker that lets you spot-check generated code against your own rules
The harness-engineering anthology argues that most agent failures come from missing context — for a small team, building that context the right way used to require a senior engineer; this turns it into a 30-minute setup.
Harness-engineering repo sits at ~304 stars / 15 forks with a v1.0 release, multiple independent guides (agenticoding.ai, deployhq, minibase, tianpan.co) covering AGENTS.md/CLAUDE.md have appeared, but audience is still a narrow slice of senior engineers and small teams — not yet mass-market.lopopolo/harness-engineering — GitHub ↗lopopolo/harness-engineering — trendshift stats ↗Lesson 6: Project Onboarding | Agentic Coding ↗
Web searches surfaced only guide articles and the canonical repo, no purpose-built generator SaaS; however the output is just a markdown file a senior dev can hand-craft or one-shot with an LLM, so defensibility is low.CLAUDE.md, AGENTS.md & Copilot Instructions: Configure Every AI Coding ... ↗AGENTS.md & CLAUDE.md: The New README for AI Coding Agents ↗
Comparable AI dev tooling commands $20–100+/dev/mo, but this is a one-and-done setup artifact, not recurring infrastructure; many buyers will DIY with the free canonical bundle or a ChatGPT prompt, capping willingness to pay.AI coding assistant pricing and ROI guide (2026) ↗Harness engineering: leveraging Codex in an agent-first world ↗
AGENTS.md is now explicitly the vendor-neutral standard across Cursor, Windsurf, Copilot, Claude Code and Codex, with major vendors (OpenAI, Anthropic tooling) committing to instruction-file workflows — the category is structural, not a fad.Lesson 6: Project Onboarding | Agentic Coding ↗Harness engineering: leveraging Codex in an agent-first world ↗
Build is a wizard-style form plus templated markdown emission — a weekend project for one engineer; main risk is content quality of generated prompts, not technical complexity.CLAUDE.md and AGENTS.md: The Configuration Layer That Makes AI Coding ... ↗