THE MEDIA FOR AI THAT ACTUALLY HELPS.English edition · Understand & take action
Spotted on the web

Prepare recurring infrastructure applications by an agent on Slack

Use described publicly by Blog technique de leboncoin (29 September 2026), summarised by the desk. Not tested by iadujour.

The result to aim for

The Leboncoin infrastructure team built BotISS, an agent that responds on its Slack support channel, completes DNS requests and opens the code change itself, then validated by an engineer.

What was done

According to Leboncoin's technical blog (29 September 2026), about ten engineers from the Infrastructure Self Service team are responding to the demands of about 700 people from the tech via a Slack channel, with two "firefighters" on a daily basis. To ease this burden, the team began with a narrow and repetitive case: creating DNS recordings. BotISS, written in Python with the SDK Strands of AWS and Claude Sonnet 4.5 via Amazon Bedrock, passes each message through an orchestrator agent who determines whether the request is within his reach and then entrusts it to a specialized agent. He reads the internal documentation in Markdown, asks for missing information (registration type, IP address), changes Terraform code, opens a pull request and summarizes the change; A firefighter will read it again.Other specialized agents now cover the whitelisting of IP addresses and documentation. Each agent's procedures are written in natural language in Markdown files, and sessions are kept in S3. The team is considering extending the tool to other teams, which raises governance questions.

Before you start

Tool cited by the source: BotISS (leboncoin); other tools cited: Claude, Amazon Bedrock and Strands Agents. Gather the information the prompt asks for. Strip confidential data that is not needed and check your tool’s access terms. Features and plans can change.

The method, in three steps.

  1. Gather the context

    Gather the information described in the prompt.

  2. Adapt the prompt

    Copy the prompt below, replace the bracketed items, then paste it into Claude.

  3. Reread before using

    Reread the result, correct it if needed and keep the final decision. Caveats are detailed below.

Your starting point.

Replace the bracketed items, then adapt the proposal to your situation. This prompt is written by the iadujour desk It is not a quotation from the source.

The prompt to customise
You're the support assistant for our infrastructure team. This is our procedure for

When asked comes:
1. Verify that it falls within this procedure; If not, say it and hand it to a human.
2. List the Missing Information and Ask Questions.
3. Proposes the exact code change according to our conventions, with a summary for the editor.
Do not deploy anything: the modification will be reread and validated by an engineer.

What deserves your attention

The human rereading of each pull request remains the lock: the article evokes the idea of a self-validation for trivial cases, to be considered only with strict rules and simple reverse returns.
Origin of the case. Use described publicly by Blog technique de leboncoin (29 September 2026).
Proof level: Testimony published by an identified source. Source checked on 10 November 2026. Use described by the source, not tested by the desk. Original summary written by iadujour.fr: figures and claims are the source’s; the prompt is a desk proposal, not a quotation.
TO GO FURTHER

Other ideas to put to work.

The whole Dev theme
Dev

Make a pull request by an agent before the human review

A first automated passage to identify probable bugs, missing tests and safety risks, so that the human review focuses on the essentials.

Tool: Claude CodeType: Practical case
DISCUSSION

Comments.

A field note, a question, a precision on this use case? Comments are reviewed before publication.

Published comments

Loading comments…

    Leave a comment

    Moderation charter. Every comment is reviewed before publication. We refuse hate speech, insulting or discriminatory language, advertising, spam and promotional links, off-topic posts and other people’s personal data.