How to use Claude to implement automation specs
A consultancy owner specifies client automations in Markdown, delegates implementation to Claude Code, and reviews and tests the results, reporting roughly halved delivery time.
news.ycombinator.com
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Implement [automation] from [Markdown specification]. Follow its inputs, outputs, edge cases, integration points, API response formats and state transitions. Identify missing requirements and provide work for review and testing.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source poster / publisher
- evara-ai
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- The author reports roughly halved delivery time for typical automation projects and describes well-specified first passes as 80–90% production-ready.
- Human involvement
- The author writes the specifications and explicitly reviews, tests and iterates on Claude's implementations.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Write detailed Markdown specifications before implementation.
- Have Claude Code implement the specified automation.
- Review and test the generated implementation.
The author runs a small AI consultancy and reports using Claude Code daily for client projects, including chatbots, automation pipelines and API integrations. Their workflow begins with about 30 minutes spent writing detailed Markdown specifications. Those specifications cover inputs, outputs, edge cases and integration points before Claude starts implementation.
The author then reviews, tests and iterates on the generated work. They emphasize explicit error-handling expectations, API response formats and state transitions as useful specification content. In their experience, vague specifications lead to debugging that can erase the time saved, while good specifications produce work they describe as 80–90% production-ready on the first pass.
For a typical automation project, illustrated by a WhatsApp booking bot connected to a client's CRM, they report delivery time roughly halving compared with manual implementation. They also report finding two or three focused agents more reliable than larger competing groups, assigning backend, frontend and tests separately to reduce conflicts and consistency work.
Who it's for: Consultants delivering bounded integrations who can define behavior precisely and validate generated implementations.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Hacker News · evara-ai
Frequently asked questions
What goes into the author's Markdown specifications?
Inputs, outputs, edge cases and integration points, with explicit expectations for error handling, API response formats and state transitions.
How many agents did the author find reliable?
They report preferring two or three focused agents, such as separate backend, frontend and testing agents, over six to eight competing agents.
More details
Remaining steps
- Iterate on the work using the specified behavior and integration requirements.
About this account
Claude Code handles implementation from detailed specifications while the author remains involved in review, testing and iteration.
Platform: Hacker News. Attribution: https://news.ycombinator.com/item?id=47230248. Prompt: Suggested (written by Agents In Use, not the author's). Sourced by the Agents In Use pipeline.
Credit: evara-ai on Hacker News
Source published
Added to Agents In Use
Source checked
Tools and topics
Claude Code · Markdown
Automation · Specifications · Client projects
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.