<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: samuccaya</title>
    <description>The latest articles on DEV Community by samuccaya (@samuccaya).</description>
    <link>https://dev.to/samuccaya</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4252951%2F2aa9dfd1-f9e7-48b0-82a7-dd9d17054419.jpg</url>
      <title>DEV Community: samuccaya</title>
      <link>https://dev.to/samuccaya</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/samuccaya"/>
    <language>en</language>
    <item>
      <title>Turn your AI agent's checks into tests your team can rerun</title>
      <dc:creator>samuccaya</dc:creator>
      <pubDate>Thu, 19 Nov 2026 05:03:31 +0000</pubDate>
      <link>https://dev.to/samuccaya/turn-your-ai-agents-checks-into-tests-your-team-can-rerun-jmb</link>
      <guid>https://dev.to/samuccaya/turn-your-ai-agents-checks-into-tests-your-team-can-rerun-jmb</guid>
      <description>&lt;p&gt;When an AI agent verifies something ("POST /orders works, the row is there"), that knowledge usually dies with the chat. Next sprint, someone asks again.&lt;/p&gt;

&lt;p&gt;The fix is boring and effective: &lt;strong&gt;make the agent save what it verified, as files in the repo.&lt;/strong&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  The loop
&lt;/h2&gt;

&lt;p&gt;With the &lt;a href="https://testapilocally.com/mcp/" rel="noopener noreferrer"&gt;SignOff Studio MCP server&lt;/a&gt; connected (Claude Code, Cursor, VS Code Copilot or Gemini CLI):&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Verify.&lt;/strong&gt; &lt;code&gt;verify_api_effect&lt;/code&gt; runs a request plus database checks bound to its response, and returns one verdict.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Save.&lt;/strong&gt; &lt;code&gt;save_request&lt;/code&gt; and &lt;code&gt;save_db_check&lt;/code&gt; write the request, its assertions and the database check into the workspace (&lt;code&gt;signoff-workspace.json&lt;/code&gt;, collections, &lt;code&gt;database/&lt;/code&gt;). These are plain files.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Commit.&lt;/strong&gt; They go in the same pull request as the code.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Rerun.&lt;/strong&gt; &lt;code&gt;run_collection&lt;/code&gt; (from the agent), the desktop app, the VS Code extension or the CLI in CI runs them again. Failures show expected vs actual.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Report.&lt;/strong&gt; &lt;code&gt;export_evidence&lt;/code&gt; writes an HTML/Markdown report plus JUnit, and a short summary to paste in the PR.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Prompts that work
&lt;/h2&gt;

&lt;p&gt;The server ships slash-command prompts in Claude Code and VS Code:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;verify-endpoint&lt;/code&gt;: call it, check it, save it.&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;verify-migration&lt;/code&gt;: check the schema and data after a migration.&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;regression-check&lt;/code&gt;: rerun the saved collection and summarise failures.&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;explore-api&lt;/code&gt;: import an OpenAPI/Postman/Insomnia/Bruno/cURL file and try the main endpoints.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Why files, not a SaaS dashboard
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Reviewers see the checks in the diff.&lt;/li&gt;
&lt;li&gt;CI runs them with no account and no network dependency.&lt;/li&gt;
&lt;li&gt;Nothing leaves your machine: requests and queries run locally, and results go only to your agent.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Getting started
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;npx -y @signoff-studio/mcp init
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Node 22+, or use the free Windows app / VS Code extension, which ship the same server (Help → Connect AI agents writes the config, with a preview of every file).&lt;/p&gt;

&lt;p&gt;&lt;a href="https://testapilocally.com/mcp/" rel="noopener noreferrer"&gt;testapilocally.com/mcp&lt;/a&gt; · &lt;a href="https://github.com/samuccaya/signoff-studio-mcp" rel="noopener noreferrer"&gt;GitHub&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I build SignOff Studio.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>testing</category>
      <category>ci</category>
      <category>mcp</category>
    </item>
  </channel>
</rss>
