<?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: AI CodeOps</title>
    <description>The latest articles on DEV Community by AI CodeOps (@aicodeops_dev).</description>
    <link>https://dev.to/aicodeops_dev</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%2F4216096%2F2afeb95f-de8b-4631-b368-c92e6c9dd616.png</url>
      <title>DEV Community: AI CodeOps</title>
      <link>https://dev.to/aicodeops_dev</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/aicodeops_dev"/>
    <language>en</language>
    <item>
      <title>What SOC 2 auditors ask about AI-written code</title>
      <dc:creator>AI CodeOps</dc:creator>
      <pubDate>Thu, 19 Nov 2026 14:30:00 +0000</pubDate>
      <link>https://dev.to/aicodeops_dev/what-soc-2-auditors-ask-about-ai-written-code-51d0</link>
      <guid>https://dev.to/aicodeops_dev/what-soc-2-auditors-ask-about-ai-written-code-51d0</guid>
      <description>&lt;p&gt;SOC 2 doesn't mention AI coding tools. It doesn't need to. The change management criterion (CC8.1 in the Trust Services Criteria) already expects that changes to production systems are authorized, tested, approved and documented. When AI writes a growing share of those changes, auditors are starting to ask how that still holds.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;This article is general information, not audit or legal advice. Your auditor's expectations are what count.&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  The questions that come up
&lt;/h2&gt;

&lt;p&gt;Based on what security and compliance teams describe, the questions tend to be some version of:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Who reviewed this change?&lt;/strong&gt; A named person, not "the AI reviewed it."&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;What was tested, and on which version?&lt;/strong&gt; Evidence that the tests ran on the code that shipped.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Was the change authorized?&lt;/strong&gt; It traces back to a ticket, request or plan.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Can a single actor push to production alone?&lt;/strong&gt; Separation between the author (human or agent) and the approver.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Can you show this for a sample?&lt;/strong&gt; Auditors pick changes from the period and ask for the record.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;None of these are new. What's new is volume: AI agents produce far more changes, and the evidence for each has to come from somewhere.&lt;/p&gt;

&lt;h2&gt;
  
  
  Where teams get stuck
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Approvals that don't match what shipped.&lt;/strong&gt; A reviewer approved one commit, the agent pushed a fix, and the merge happened without a fresh look.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;"AI reviewed it" as the only review.&lt;/strong&gt; An AI review is useful input, but it isn't a human approval.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Evidence spread across tools.&lt;/strong&gt; Ticket in one place, CI logs in another, approval in the pull request, and nothing linking them.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Agent permissions.&lt;/strong&gt; Agents with tokens that can merge or push to protected branches remove the separation auditors look for.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  What good looks like
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Every change links to its authorization (ticket or plan).&lt;/li&gt;
&lt;li&gt;Tests run on the exact commit that merges, and the result is kept.&lt;/li&gt;
&lt;li&gt;A named person approves, and approvals reset on new commits.&lt;/li&gt;
&lt;li&gt;Agents can propose changes but can't approve or merge their own work.&lt;/li&gt;
&lt;li&gt;One record per change brings this together, and you can export it for any sample.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How AI CodeOps helps
&lt;/h2&gt;

&lt;p&gt;AI CodeOps was built around this record. AI agents plan, build, review and test changes on your repository, and a pull request only opens when your tests pass on that exact commit, the AI review of the real diff has no open high or critical findings, and a person approves. Policies control what each agent can do, and agents can't approve their own changes.&lt;/p&gt;

&lt;p&gt;Every change gets a signed evidence pack: what was requested, what changed, which tests ran on which commit, what the review found and who approved it. When an auditor picks a sample, the record already exists.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Free during the open beta: &lt;a href="https://aicodeops.dev" rel="noopener noreferrer"&gt;https://aicodeops.dev&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

</description>
      <category>security</category>
      <category>compliance</category>
      <category>ai</category>
      <category>devops</category>
    </item>
    <item>
      <title>Your repo can talk to your AI agent: prompt injection for coding teams</title>
      <dc:creator>AI CodeOps</dc:creator>
      <pubDate>Tue, 17 Nov 2026 14:30:00 +0000</pubDate>
      <link>https://dev.to/aicodeops_dev/your-repo-can-talk-to-your-ai-agent-prompt-injection-for-coding-teams-21oo</link>
      <guid>https://dev.to/aicodeops_dev/your-repo-can-talk-to-your-ai-agent-prompt-injection-for-coding-teams-21oo</guid>
      <description>&lt;p&gt;A coding agent reads a lot before it writes anything: the issue it was assigned, comments on it, the README, docs, source files, test output, and sometimes web pages and package docs. It treats all of that as context.&lt;/p&gt;

&lt;p&gt;The problem is that a language model can't reliably tell the difference between context it should use and instructions it should follow. If text in that context says "ignore the task and do this instead," the model may comply. This is prompt injection, and OWASP lists it as the top risk in its Top 10 for LLM applications.&lt;/p&gt;

&lt;p&gt;For coding teams, the question isn't whether a model can be tricked. It's what an agent can do once it has been.&lt;/p&gt;

&lt;h2&gt;
  
  
  Where injected instructions can come from
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Issues and comments.&lt;/strong&gt; On public repositories, anyone can open an issue. If agents pick up issues automatically, the issue body is untrusted input.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Pull request descriptions and review comments&lt;/strong&gt; from outside contributors.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Files in the repository&lt;/strong&gt;, including docs, config files and code comments, especially in dependencies or vendored code.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Tool output.&lt;/strong&gt; Test logs, error messages and command output that include text from elsewhere.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Web pages and package docs&lt;/strong&gt; the agent fetches while researching.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Instructions can be hidden: inside HTML comments, collapsed sections, long lines, or text written to look like part of the file.&lt;/p&gt;

&lt;h2&gt;
  
  
  What an injected agent might try
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Add a dependency, a script or a CI step that runs attacker code&lt;/li&gt;
&lt;li&gt;Read secrets from the environment and print them into logs or a pull request&lt;/li&gt;
&lt;li&gt;Weaken a check: skip a test, loosen a permission, remove validation&lt;/li&gt;
&lt;li&gt;Push to a branch or merge without review&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;All of these depend on what the agent is allowed to do. That's where to focus.&lt;/p&gt;

&lt;h2&gt;
  
  
  Limit what a fooled agent can do
&lt;/h2&gt;

&lt;p&gt;You can't make a model immune to prompt injection. You can make injection low-impact:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;No secrets in the agent's environment.&lt;/strong&gt; Run agents with the minimum credentials needed, and keep production keys out entirely.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Scoped repository tokens.&lt;/strong&gt; Access only to the repositories in the task, and no permission to merge or push to protected branches.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Sandboxed execution.&lt;/strong&gt; Tests and commands run in an isolated environment with limited network access.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Treat outside input as untrusted.&lt;/strong&gt; Don't let agents act automatically on issues or comments from people outside the team.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Review the diff, not the description.&lt;/strong&gt; An injected change can come with a perfectly reasonable summary.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Gate the merge.&lt;/strong&gt; Nothing reaches main without tests on the exact commit, an independent review and a human approval.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  How AI CodeOps handles it
&lt;/h2&gt;

&lt;p&gt;In AI CodeOps, agents work in sandboxes with tokens scoped to the repositories you choose, and policies control what each agent may do. Agents can't approve or merge their own work. A separate AI review checks the real diff, including changes nobody asked for, and a pull request only opens when tests pass on that exact commit, the review has no open high or critical findings, and a person approves. If anything errors, the gate fails closed.&lt;/p&gt;

&lt;p&gt;That doesn't make prompt injection impossible. It means a fooled agent produces a pull request that gets caught, not a change that ships.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Free during the open beta: &lt;a href="https://aicodeops.dev" rel="noopener noreferrer"&gt;https://aicodeops.dev&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

</description>
      <category>security</category>
      <category>ai</category>
      <category>llm</category>
      <category>devsecops</category>
    </item>
    <item>
      <title>How to review an AI-written pull request: a 10-point checklist</title>
      <dc:creator>AI CodeOps</dc:creator>
      <pubDate>Fri, 13 Nov 2026 21:57:20 +0000</pubDate>
      <link>https://dev.to/aicodeops_dev/how-to-review-an-ai-written-pull-request-a-10-point-checklist-42bn</link>
      <guid>https://dev.to/aicodeops_dev/how-to-review-an-ai-written-pull-request-a-10-point-checklist-42bn</guid>
      <description>&lt;p&gt;AI-written pull requests have a particular look. The code is tidy, the names are sensible, the description is confident, and the tests are green. That polish makes them easy to approve and harder to review well.&lt;/p&gt;

&lt;p&gt;The mistakes are still there. They just sit in different places than a rushed human change. Here's a checklist for the ten minutes you spend on an AI-written pull request, roughly in the order worth doing them.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Read the task before the diff
&lt;/h2&gt;

&lt;p&gt;Start with what was asked: the issue, the prompt, the plan. Then ask whether the diff does that, all of it, and nothing else. Agents often solve a nearby problem, or the stated problem plus a "while I was here" refactor nobody requested.&lt;/p&gt;

&lt;h2&gt;
  
  
  2. Check the size against the request
&lt;/h2&gt;

&lt;p&gt;A one-line bug fix shouldn't touch nine files. If the diff is much bigger than the task, ask for it to be split or trimmed before reviewing details. Large unrequested changes are where regressions hide.&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Look at what was deleted
&lt;/h2&gt;

&lt;p&gt;Reviewers read added lines. Agents also remove things: a validation check, an error branch, a log line, a test case that was "flaky." Scan the red lines on purpose.&lt;/p&gt;

&lt;h2&gt;
  
  
  4. Check the tests, not just the result
&lt;/h2&gt;

&lt;p&gt;Green tests only mean something if they test the change. Look for:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Tests that assert what the code does rather than what it should do&lt;/li&gt;
&lt;li&gt;Mocks that replace the very thing being tested&lt;/li&gt;
&lt;li&gt;Tests that were changed or skipped to make the suite pass&lt;/li&gt;
&lt;li&gt;No new test for a bug fix&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  5. Follow the inputs
&lt;/h2&gt;

&lt;p&gt;Wherever the change reads input (request bodies, query strings, files, environment variables, model output), check that it's validated, size-limited and safely encoded on the way out. Agents tend to secure the path they were asked about and leave a parallel path open.&lt;/p&gt;

&lt;h2&gt;
  
  
  6. Verify every new dependency
&lt;/h2&gt;

&lt;p&gt;Check that each new package exists, is the one you think it is, is maintained, and is actually needed. Models sometimes suggest package names that don't exist or are close to popular ones, which is how typosquatting gets in.&lt;/p&gt;

&lt;h2&gt;
  
  
  7. Look for invented APIs and options
&lt;/h2&gt;

&lt;p&gt;A call to a method or a config option that sounds right but doesn't exist in your version of the library may compile in a dynamic language and fail at runtime. If you don't recognise it, check the docs.&lt;/p&gt;

&lt;h2&gt;
  
  
  8. Check error handling and failure paths
&lt;/h2&gt;

&lt;p&gt;Look at what happens when the network call fails, the file is missing, or the value is null. Common patterns in generated code are catch-all handlers that swallow errors and fallbacks that quietly return a default instead of failing.&lt;/p&gt;

&lt;h2&gt;
  
  
  9. Check secrets and config
&lt;/h2&gt;

&lt;p&gt;Look for hard-coded keys, tokens, URLs or credentials, test settings left on, and debug flags. Also check that nothing sensitive was added to logs.&lt;/p&gt;

&lt;h2&gt;
  
  
  10. Confirm what you're approving
&lt;/h2&gt;

&lt;p&gt;Make sure the commit you're approving is the one that will merge, and that the tests ran on it. If the agent pushed again after your review, look again.&lt;/p&gt;

&lt;h2&gt;
  
  
  Make the checklist cheaper
&lt;/h2&gt;

&lt;p&gt;Most of this list can be pre-checked so the human review focuses on judgement:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Run a separate AI review of the real diff and rank findings by severity, so the reviewer starts from a list.&lt;/li&gt;
&lt;li&gt;Fail the change automatically when tests didn't run on the final commit.&lt;/li&gt;
&lt;li&gt;Flag new dependencies and deleted tests in the pull request summary.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;That's how AI CodeOps works: AI agents plan, build, review and test changes on your repository, and a pull request only opens when your tests pass on the exact commit, the AI review has no open high or critical findings, and a person approves. The reviewer still makes the call; they just start with the evidence in front of them.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;AI CodeOps is free during the open beta, and free for non-commercial open-source projects: &lt;a href="https://aicodeops.dev" rel="noopener noreferrer"&gt;https://aicodeops.dev&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

</description>
      <category>codereview</category>
      <category>ai</category>
      <category>programming</category>
      <category>productivity</category>
    </item>
    <item>
      <title>Harden a fresh Ubuntu 24.04 server with one command (open source)</title>
      <dc:creator>AI CodeOps</dc:creator>
      <pubDate>Thu, 05 Nov 2026 17:16:15 +0000</pubDate>
      <link>https://dev.to/aicodeops_dev/harden-a-fresh-ubuntu-2404-server-with-one-command-open-source-5830</link>
      <guid>https://dev.to/aicodeops_dev/harden-a-fresh-ubuntu-2404-server-with-one-command-open-source-5830</guid>
      <description>&lt;p&gt;Every new VPS gets the same first hour: create a user, lock down SSH, set up a firewall, install fail2ban, turn on automatic updates… and then forget one of them on the next server.&lt;/p&gt;

&lt;p&gt;So we turned our checklist into a small open-source script: &lt;strong&gt;&lt;a href="https://github.com/veloxahost/server-hardening" rel="noopener noreferrer"&gt;veloxahost/server-hardening&lt;/a&gt;&lt;/strong&gt;. One command, readable configs, safe to re-run.&lt;/p&gt;

&lt;h2&gt;
  
  
  What it does
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Creates a sudo user and installs your SSH key&lt;/li&gt;
&lt;li&gt;SSH: keys only, no root login, no passwords — via a drop-in file, validated with &lt;code&gt;sshd -t&lt;/code&gt; before restart&lt;/li&gt;
&lt;li&gt;UFW: deny inbound, rate-limit SSH, allow 80/443&lt;/li&gt;
&lt;li&gt;fail2ban with the &lt;strong&gt;systemd backend&lt;/strong&gt; (the reliable choice on Ubuntu 24.04) and escalating bans&lt;/li&gt;
&lt;li&gt;Unattended security upgrades + &lt;code&gt;needrestart&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;auditd rules for identity files, sudoers, sshd config and root commands&lt;/li&gt;
&lt;li&gt;Kernel sysctl hardening that doesn't break Docker&lt;/li&gt;
&lt;li&gt;A Lynis audit at the end so you get a score&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Usage
&lt;/h2&gt;

&lt;p&gt;On a &lt;strong&gt;fresh&lt;/strong&gt; Ubuntu 24.04 server (22.04 and Debian 12 work too), keep your current session open and run:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;git clone https://github.com/veloxahost/server-hardening.git
&lt;span class="nb"&gt;cd &lt;/span&gt;server-hardening
&lt;span class="nb"&gt;sudo &lt;/span&gt;&lt;span class="nv"&gt;NEW_USER&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;deploy &lt;span class="nv"&gt;SSH_KEY&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="s2"&gt;"ssh-ed25519 AAAA... you@example.com"&lt;/span&gt; ./harden.sh
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Then open a &lt;strong&gt;new&lt;/strong&gt; terminal and check &lt;code&gt;ssh deploy@your-server&lt;/code&gt; works before logging out.&lt;/p&gt;

&lt;p&gt;Options are plain environment variables:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Var&lt;/th&gt;
&lt;th&gt;Default&lt;/th&gt;
&lt;th&gt;Meaning&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;NEW_USER&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;deploy&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;sudo user to create&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;SSH_KEY&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;required&lt;/td&gt;
&lt;td&gt;public key for that user&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;SSH_PORT&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;22&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;SSH port to allow&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;ALLOW_PORTS&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;80,443&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;extra TCP ports to open&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;ENABLE_IP_FORWARD&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;0&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;set &lt;code&gt;1&lt;/code&gt; on Docker hosts&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;RUN_LYNIS&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;1&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;run the audit at the end&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  Three things people get wrong on Ubuntu 24.04
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;1. Editing &lt;code&gt;sshd_config&lt;/code&gt; directly.&lt;/strong&gt; 24.04 reads &lt;code&gt;/etc/ssh/sshd_config.d/*.conf&lt;/code&gt; first. Put your settings in a drop-in so package upgrades never overwrite them.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;2. fail2ban watching a log file that's empty.&lt;/strong&gt; SSH logs go to the journal now. Use &lt;code&gt;backend = systemd&lt;/code&gt; or your &lt;code&gt;sshd&lt;/code&gt; jail will show zero failures forever.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;3. Trusting UFW with Docker.&lt;/strong&gt; Ports published with &lt;code&gt;-p 8080:8080&lt;/code&gt; bypass UFW completely. Bind internal services to &lt;code&gt;127.0.0.1&lt;/code&gt; and put a reverse proxy in front.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why not Ansible?
&lt;/h2&gt;

&lt;p&gt;If you already run Ansible, use it. This is for the "one server, five minutes" case: a single bash script you can read top to bottom before you run it as root.&lt;/p&gt;

&lt;h2&gt;
  
  
  More
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Full write-up with every command explained: &lt;a href="https://veloxahost.com/blog/posts/linux-server-hardening-checklist.html?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=server-hardening" rel="noopener noreferrer"&gt;The Complete Linux Server Hardening Checklist (2026)&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Production Docker Compose stacks with automatic HTTPS: &lt;a href="https://github.com/veloxahost/compose-stacks" rel="noopener noreferrer"&gt;veloxahost/compose-stacks&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;PRs and issues welcome — especially for other distros.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I work on &lt;a href="https://veloxahost.com/?utm_source=devto&amp;amp;utm_medium=referral&amp;amp;utm_campaign=server-hardening" rel="noopener noreferrer"&gt;VeloxaHost&lt;/a&gt;, a hosting platform. The script is MIT-licensed and works on any Linux server.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>linux</category>
      <category>security</category>
      <category>devops</category>
      <category>ubuntu</category>
    </item>
    <item>
      <title>AI Code Review for Gitea and Forgejo: What Works Today</title>
      <dc:creator>AI CodeOps</dc:creator>
      <pubDate>Tue, 03 Nov 2026 17:04:28 +0000</pubDate>
      <link>https://dev.to/aicodeops_dev/ai-code-review-for-gitea-and-forgejo-what-works-today-3kb0</link>
      <guid>https://dev.to/aicodeops_dev/ai-code-review-for-gitea-and-forgejo-what-works-today-3kb0</guid>
      <description>&lt;p&gt;Most AI code review tools are built for GitHub first, then GitLab and Bitbucket. If your team runs Gitea or Forgejo, often self-hosted for cost or control, the list gets short fast.&lt;/p&gt;

&lt;p&gt;For example, CodeRabbit's &lt;a href="https://docs.coderabbit.ai/platforms/overview" rel="noopener noreferrer"&gt;platform docs&lt;/a&gt; list GitHub, GitLab, Bitbucket and Azure DevOps (cloud and self-hosted variants), with no Gitea or Forgejo support as of November 2026.&lt;/p&gt;

&lt;p&gt;AI CodeOps works with Gitea and Forgejo, including self-hosted servers, the same way it works with GitHub and GitLab: you connect the server with a token scoped to the repositories you choose, and AI agents plan, build, review and test changes on your real repository.&lt;/p&gt;

&lt;h2&gt;
  
  
  What you get on Gitea or Forgejo
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;AI review of the real diff&lt;/strong&gt;, with findings sorted by severity&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Your own tests&lt;/strong&gt;, run in a sandbox on the exact version of the change&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A release gate&lt;/strong&gt;: no pull request until the tests pass, the review has no open high or critical findings, and a person approves&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A release-check API&lt;/strong&gt; your CI can call on any PR or commit, so direct merges are covered too&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Free during the open beta, and free forever for non-commercial open-source projects: &lt;a href="https://aicodeops.dev" rel="noopener noreferrer"&gt;https://aicodeops.dev&lt;/a&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>codereview</category>
      <category>selfhosted</category>
      <category>git</category>
    </item>
    <item>
      <title>Two Policy Rules to Give Your AI Coding Agents Before Anything Else</title>
      <dc:creator>AI CodeOps</dc:creator>
      <pubDate>Tue, 03 Nov 2026 14:12:42 +0000</pubDate>
      <link>https://dev.to/aicodeops_dev/two-policy-rules-to-give-your-ai-coding-agents-before-anything-else-eh</link>
      <guid>https://dev.to/aicodeops_dev/two-policy-rules-to-give-your-ai-coding-agents-before-anything-else-eh</guid>
      <description>&lt;p&gt;Most teams adopting AI coding agents start with trust settings: what can the agent see, what can it change. Before you tune anything else, we'd start with two rules.&lt;/p&gt;

&lt;h2&gt;
  
  
  Rule 1: No production deploys on an agent's word
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight json"&gt;&lt;code&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"id"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"no-production-deploys"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"action"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"release.deploy"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"resource"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"production"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"decision"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"deny"&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;An agent can propose a change, run tests and open a pull request. Shipping to production should still be a human decision, made with the evidence in front of them. A deny rule makes that explicit instead of hoping nobody wires it up.&lt;/p&gt;

&lt;h2&gt;
  
  
  Rule 2: Any change to your CI files waits for a person
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight json"&gt;&lt;code&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"id"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"review-ci-changes"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"action"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"diff.apply"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"resource"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;".github/**"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"decision"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"ask"&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Your CI configuration is what decides whether everything else is checked. A change to &lt;code&gt;.github/workflows&lt;/code&gt; can quietly skip a test job or loosen a branch protection step. That makes CI files the highest-leverage target in the repo, for a buggy agent or a prompt injection alike. &lt;code&gt;ask&lt;/code&gt; holds the change on an Approvals page until someone reviews it.&lt;/p&gt;

&lt;h2&gt;
  
  
  How the rules behave
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Strictest wins.&lt;/strong&gt; When several rules match, deny beats ask, and ask beats allow, whatever their order.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Ask fails closed.&lt;/strong&gt; Where an action can't wait for a person, an ask is enforced as a deny.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;One reject stops it.&lt;/strong&gt; On the Approvals page, a single reject ends the request.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Versioned.&lt;/strong&gt; Each save creates a new policy version, so you can see what changed and when.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you're unsure about a rule's impact, run it in log-only mode first and watch which actions it would have blocked.&lt;/p&gt;

&lt;p&gt;Video walkthroughs are on our YouTube channel: &lt;a href="https://www.youtube.com/@AICodeOps" rel="noopener noreferrer"&gt;https://www.youtube.com/@AICodeOps&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;AI CodeOps is free during the open beta and free forever for non-commercial open source: &lt;a href="https://aicodeops.dev" rel="noopener noreferrer"&gt;https://aicodeops.dev&lt;/a&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>devops</category>
      <category>security</category>
      <category>github</category>
    </item>
    <item>
      <title>Our AI's Security Audit Flagged a Bug We Never Asked It to Fix</title>
      <dc:creator>AI CodeOps</dc:creator>
      <pubDate>Tue, 03 Nov 2026 14:12:15 +0000</pubDate>
      <link>https://dev.to/aicodeops_dev/our-ais-security-audit-flagged-a-bug-we-never-asked-it-to-fix-4lap</link>
      <guid>https://dev.to/aicodeops_dev/our-ais-security-audit-flagged-a-bug-we-never-asked-it-to-fix-4lap</guid>
      <description>&lt;p&gt;We gave a set of AI agents a small, realistic task on a demo shopping cart: discount codes now come from user input, so validate them. Trim whitespace, uppercase, letters and digits only, 20 characters max, and reject anything else before it reaches the totals. Add tests.&lt;/p&gt;

&lt;p&gt;We ran it through a Security-Sensitive Change workflow: five agents in a row. One scopes the change, one analyses the risk, one implements, one audits the diff, and one validates the result. The whole run took about 35 seconds and cost under 20 cents.&lt;/p&gt;

&lt;p&gt;The implementation did what we asked. The diff added a &lt;code&gt;validateDiscountCode&lt;/code&gt; helper that trims, uppercases, checks the pattern and length, and throws on anything else. Tests covered valid codes, illegal characters, long codes and unknown codes.&lt;/p&gt;

&lt;p&gt;Then the audit step came back with four findings. Three were minor. One was marked HIGH:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Unvalidated &lt;code&gt;discountPercent&lt;/code&gt; parameter allows arbitrary discount values.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;The cart's older &lt;code&gt;total()&lt;/code&gt; function still accepted a raw &lt;code&gt;discountPercent&lt;/code&gt; option. We had asked the agents to lock down the new input path, and they did. But the old path was still wide open: a caller could pass 100, or 250, and get a free cart or a negative total. Nobody had asked about it. The auditor found it anyway, because it was reading the code around the change, not just the change.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;The gate held.&lt;/strong&gt; Because a HIGH finding was open, the change was not release-ready. It could not become a pull request until the finding was fixed or a person signed off on it. One click on Fix sends the finding back to an agent as a new task.&lt;/p&gt;

&lt;h2&gt;
  
  
  Three takeaways
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Audit the code around the change, not just the diff lines.&lt;/strong&gt; The most dangerous bugs were next to what we changed, not in it.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A finding needs teeth.&lt;/strong&gt; If a reviewer can flag a HIGH issue and the change still merges, the review is advice. Tie open high or critical findings to a merge block.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Scope creep is good when it's a finding, not a commit.&lt;/strong&gt; The auditor didn't silently "fix" the extra issue inside our change. It reported it, so a person decides.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;We recorded runs like this as short tutorials on our YouTube channel: &lt;a href="https://www.youtube.com/@AICodeOps" rel="noopener noreferrer"&gt;https://www.youtube.com/@AICodeOps&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;AI CodeOps is free during the open beta and free forever for non-commercial open source: &lt;a href="https://aicodeops.dev" rel="noopener noreferrer"&gt;https://aicodeops.dev&lt;/a&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>security</category>
      <category>devsecops</category>
      <category>codereview</category>
    </item>
    <item>
      <title>AI CodeOps Is Live on Product Hunt</title>
      <dc:creator>AI CodeOps</dc:creator>
      <pubDate>Mon, 02 Nov 2026 14:00:00 +0000</pubDate>
      <link>https://dev.to/aicodeops_dev/ai-codeops-is-live-on-product-hunt-18ln</link>
      <guid>https://dev.to/aicodeops_dev/ai-codeops-is-live-on-product-hunt-18ln</guid>
      <description>&lt;p&gt;Today AI CodeOps launches on Product Hunt.&lt;/p&gt;

&lt;p&gt;AI CodeOps is a browser-based platform where AI agents plan, build, review and test code on your real repository. What makes it different is what happens at the end: nothing merges on the AI's word.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem we kept running into
&lt;/h2&gt;

&lt;p&gt;AI coding agents got fast. A task that took an afternoon now comes back as a pull request in minutes. But the review didn't get faster. Teams end up doing one of two things: trusting the agent's own summary of its work, or skimming a diff that's too big to read properly. Either way, the bar for AI-written code quietly drops.&lt;/p&gt;

&lt;p&gt;We think the bar should stay where it is. So we built a gate.&lt;/p&gt;

&lt;h2&gt;
  
  
  How the gate works
&lt;/h2&gt;

&lt;p&gt;Every change an agent makes waits at a release gate. A pull request opens only when:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Your tests passed&lt;/strong&gt; in a sandbox, on that exact version of the change. Edit the change afterwards and it counts as untested again.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The AI review is clear.&lt;/strong&gt; A reviewer step reads the real diff. Open high or critical findings block until they're fixed or triaged.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A person approved it.&lt;/strong&gt; The person who started the session can't be the only approver when someone else in the workspace can approve.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If a check errors or times out, the gate says no. It fails closed.&lt;/p&gt;

&lt;p&gt;Your CI can ask the same gate for a verdict on any pull request or commit, so changes merged directly on your Git host are covered too (&lt;a href="https://dev.to/aicodeops_dev/how-to-gate-ai-written-pull-requests-in-github-actions-and-gitlab-ci-2gm4"&gt;here's how to wire it into GitHub Actions or GitLab CI&lt;/a&gt;). And every shipped change gets a signed evidence pack: the task, the models, the findings, the test runs and who approved it.&lt;/p&gt;

&lt;h2&gt;
  
  
  What else you get
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Works with GitHub, GitLab, Bitbucket, Gitea and Forgejo, including self-hosted servers&lt;/li&gt;
&lt;li&gt;Runs in the browser: no CLI, no IDE plugin&lt;/li&gt;
&lt;li&gt;Cost per session, workspace budgets and per-workflow spend caps&lt;/li&gt;
&lt;li&gt;Your code, prompts and outputs are never used to train AI models&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Free to try
&lt;/h2&gt;

&lt;p&gt;AI CodeOps is free during the open beta. Non-commercial open-source projects get it free forever.&lt;/p&gt;

&lt;p&gt;We'd love your feedback on the Product Hunt page, especially on the gate rules. What would you need a gate to check before you'd let an agent's pull request through?&lt;/p&gt;

&lt;p&gt;👉 Product Hunt: &lt;a href="https://www.producthunt.com/products/ai-codeops" rel="noopener noreferrer"&gt;https://www.producthunt.com/products/ai-codeops&lt;/a&gt;&lt;br&gt;
👉 Try it: &lt;a href="https://aicodeops.dev" rel="noopener noreferrer"&gt;https://aicodeops.dev&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Drafted with AI assistance and edited by the AI CodeOps team.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>producthunt</category>
      <category>ai</category>
      <category>devops</category>
      <category>showdev</category>
    </item>
    <item>
      <title>How to Gate AI-Written Pull Requests in GitHub Actions and GitLab CI</title>
      <dc:creator>AI CodeOps</dc:creator>
      <pubDate>Mon, 02 Nov 2026 01:41:39 +0000</pubDate>
      <link>https://dev.to/aicodeops_dev/how-to-gate-ai-written-pull-requests-in-github-actions-and-gitlab-ci-2gm4</link>
      <guid>https://dev.to/aicodeops_dev/how-to-gate-ai-written-pull-requests-in-github-actions-and-gitlab-ci-2gm4</guid>
      <description>&lt;p&gt;AI coding agents open pull requests faster than teams can review them. Branch protection helps, but a green CI run only tells you the build passed. It doesn't tell you whether the change was reviewed, whether serious findings are still open, or whether the tests ran on the version you're about to merge.&lt;/p&gt;

&lt;p&gt;This tutorial adds one required check to your pipeline. It asks the AI CodeOps release gate whether a pull request may ship, prints the reasons, and fails the job when the answer is no.&lt;/p&gt;

&lt;h2&gt;
  
  
  What the gate checks
&lt;/h2&gt;

&lt;p&gt;The release gate looks at every AI CodeOps session that touched the change and applies the same rules as the console:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;The project's tests passed, on the version being merged&lt;/li&gt;
&lt;li&gt;The AI review completed and no blocking findings are open (high or critical by default)&lt;/li&gt;
&lt;li&gt;Approvals are in place&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;You can make it stricter per pipeline: a coverage floor, required workflows, disallowed licenses from the project's SBOM, or a requirement that the change already deployed to an environment such as staging.&lt;/p&gt;

&lt;p&gt;Two details matter for CI:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;A denial is still HTTP 200.&lt;/strong&gt; You read the &lt;code&gt;allowed&lt;/code&gt; field, not the status code.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;It fails closed.&lt;/strong&gt; If the evaluation itself fails, you get &lt;code&gt;allowed: false&lt;/code&gt; with the summary "Evaluation failed". A change with no matching sessions is also denied.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Step 1: Create an API key
&lt;/h2&gt;

&lt;p&gt;In the console, open &lt;strong&gt;Configure → API Keys&lt;/strong&gt; and create a key. The secret starts with &lt;code&gt;sk-&lt;/code&gt; and is shown once. The key acts as the member who created it, so create it from an account whose role can read the workspace.&lt;/p&gt;

&lt;p&gt;Store it as a CI secret named &lt;code&gt;AICODEOPS_API_KEY&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: GitHub Actions
&lt;/h2&gt;

&lt;p&gt;AI CodeOps opens its pull requests on branches named &lt;code&gt;codeops/run-&amp;lt;id&amp;gt;&lt;/code&gt;. Scoping the job to those branches means pull requests written by people aren't affected.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight yaml"&gt;&lt;code&gt;&lt;span class="pi"&gt;{&lt;/span&gt;&lt;span class="nv"&gt;% raw %&lt;/span&gt;&lt;span class="pi"&gt;}&lt;/span&gt;
&lt;span class="na"&gt;name&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;AI CodeOps release gate&lt;/span&gt;

&lt;span class="na"&gt;on&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;pull_request&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="na"&gt;types&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="pi"&gt;[&lt;/span&gt;&lt;span class="nv"&gt;opened&lt;/span&gt;&lt;span class="pi"&gt;,&lt;/span&gt; &lt;span class="nv"&gt;synchronize&lt;/span&gt;&lt;span class="pi"&gt;,&lt;/span&gt; &lt;span class="nv"&gt;reopened&lt;/span&gt;&lt;span class="pi"&gt;,&lt;/span&gt; &lt;span class="nv"&gt;ready_for_review&lt;/span&gt;&lt;span class="pi"&gt;]&lt;/span&gt;

&lt;span class="na"&gt;jobs&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;release-gate&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="na"&gt;if&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;startsWith(github.head_ref, 'codeops/run-')&lt;/span&gt;
    &lt;span class="na"&gt;runs-on&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;ubuntu-latest&lt;/span&gt;
    &lt;span class="na"&gt;steps&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
      &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="na"&gt;name&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;Ask the release gate&lt;/span&gt;
        &lt;span class="na"&gt;env&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
          &lt;span class="na"&gt;AICODEOPS_API_KEY&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;${{ secrets.AICODEOPS_API_KEY }}&lt;/span&gt;
          &lt;span class="na"&gt;PR_URL&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;${{ github.event.pull_request.html_url }}&lt;/span&gt;
        &lt;span class="na"&gt;run&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="pi"&gt;|&lt;/span&gt;
          &lt;span class="s"&gt;body=$(jq -n --arg pr "$PR_URL" '{pullRequestURL: $pr, minCoveragePct: 80}')&lt;/span&gt;
          &lt;span class="s"&gt;verdict=$(curl -sSf -X POST "https://console.aicodeops.dev/api/release/check" \&lt;/span&gt;
            &lt;span class="s"&gt;-H "Authorization: Bearer $AICODEOPS_API_KEY" \&lt;/span&gt;
            &lt;span class="s"&gt;-H "Content-Type: application/json" \&lt;/span&gt;
            &lt;span class="s"&gt;-d "$body")&lt;/span&gt;
          &lt;span class="s"&gt;echo "$verdict" | jq -r '.summary'&lt;/span&gt;
          &lt;span class="s"&gt;echo "$verdict" | jq -r '.reasons[]?'&lt;/span&gt;
          &lt;span class="s"&gt;test "$(echo "$verdict" | jq r '.allowed')" = "true"&lt;/span&gt;
&lt;span class="pi"&gt;{&lt;/span&gt;&lt;span class="nv"&gt;% endraw %&lt;/span&gt;&lt;span class="pi"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;curl -f&lt;/code&gt; makes the job fail on an auth or network error, and the last line fails it on a denial. Either way, nothing gets through by accident.&lt;/p&gt;

&lt;p&gt;Then make it required: &lt;strong&gt;Settings → Rules → Rulesets&lt;/strong&gt; (or branch protection) on your main branch, and add &lt;code&gt;release-gate&lt;/code&gt; as a required status check. When the job is skipped for a human pull request, GitHub treats the skipped check as passing.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: GitLab CI
&lt;/h2&gt;

&lt;p&gt;The same idea in &lt;code&gt;.gitlab-ci.yml&lt;/code&gt;. The &lt;code&gt;pullRequestURL&lt;/code&gt; field takes your merge request URL as it appears on GitLab.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight yaml"&gt;&lt;code&gt;&lt;span class="na"&gt;release-gate&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
  &lt;span class="na"&gt;stage&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;test&lt;/span&gt;
  &lt;span class="na"&gt;image&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s"&gt;alpine:3.20&lt;/span&gt;
  &lt;span class="na"&gt;rules&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="na"&gt;if&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt; &lt;span class="s1"&gt;'&lt;/span&gt;&lt;span class="s"&gt;$CI_PIPELINE_SOURCE&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;==&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;"merge_request_event"&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;&amp;amp;&amp;amp;&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;$CI_MERGE_REQUEST_SOURCE_BRANCH_NAME&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;=~&lt;/span&gt;&lt;span class="nv"&gt; &lt;/span&gt;&lt;span class="s"&gt;/^codeops\/run-/'&lt;/span&gt;
  &lt;span class="na"&gt;script&lt;/span&gt;&lt;span class="pi"&gt;:&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="s"&gt;apk add --no-cache curl jq&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="s"&gt;MR_URL="$CI_PROJECT_URL/-/merge_requests/$CI_MERGE_REQUEST_IID"&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="s"&gt;BODY=$(jq -n --arg pr "$MR_URL" '{pullRequestURL&lt;/span&gt;&lt;span class="err"&gt;:&lt;/span&gt; &lt;span class="s"&gt;$pr}')&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="pi"&gt;&amp;gt;&lt;/span&gt;
      &lt;span class="s"&gt;VERDICT=$(curl -sSf -X POST "https://console.aicodeops.dev/api/release/check"&lt;/span&gt;
      &lt;span class="s"&gt;-H "Authorization: Bearer $AICODEOPS_API_KEY"&lt;/span&gt;
      &lt;span class="s"&gt;-H "Content-Type: application/json"&lt;/span&gt;
      &lt;span class="s"&gt;-d "$BODY")&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="s"&gt;echo "$VERDICT" | jq -r '.summary, (.reasons[]?)'&lt;/span&gt;
    &lt;span class="pi"&gt;-&lt;/span&gt; &lt;span class="s"&gt;test "$(echo "$VERDICT" | jq -r '.allowed')" = "true"&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Add &lt;code&gt;AICODEOPS_API_KEY&lt;/code&gt; under &lt;strong&gt;Settings → CI/CD → Variables&lt;/strong&gt; (masked), and turn on &lt;strong&gt;Pipelines must succeed&lt;/strong&gt; under &lt;strong&gt;Settings → Merge requests&lt;/strong&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: Tune it
&lt;/h2&gt;

&lt;p&gt;All of these go in the request body:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Field&lt;/th&gt;
&lt;th&gt;What it does&lt;/th&gt;
&lt;th&gt;Default&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;blockingSeverity&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Open findings at or above this block&lt;/td&gt;
&lt;td&gt;&lt;code&gt;high&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;blockOnTestFailure&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Failing tests block&lt;/td&gt;
&lt;td&gt;&lt;code&gt;true&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;minCoveragePct&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Coverage floor for the latest test run&lt;/td&gt;
&lt;td&gt;not gated&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;requiredWorkflows&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Workflows that must have completed&lt;/td&gt;
&lt;td&gt;none&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;disallowedLicenses&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;License ids that must not be in the SBOM&lt;/td&gt;
&lt;td&gt;none&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;requireDeployedTo&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;An environment the change must already be deployed to&lt;/td&gt;
&lt;td&gt;none&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;blockOnInFlight&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Deny while a session is still running&lt;/td&gt;
&lt;td&gt;&lt;code&gt;false&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;A stricter release branch might send &lt;code&gt;"blockingSeverity": "medium", "requireDeployedTo": "staging"&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Good to know
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;The workspace limit is 60 checks a minute. A &lt;code&gt;429&lt;/code&gt; response includes &lt;code&gt;Retry-After&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;The API is deliberately small during the open beta: this endpoint, plus public keys for verifying evidence packs.&lt;/li&gt;
&lt;li&gt;The same check is in the console under &lt;strong&gt;Review &amp;amp; Gate → Release gate&lt;/strong&gt;, if you want to see a verdict before wiring up CI.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Full reference: &lt;a href="https://aicodeops.dev/docs/api/release-check" rel="noopener noreferrer"&gt;https://aicodeops.dev/docs/api/release-check&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Drafted with AI assistance and edited by the AI CodeOps team.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>githubactions</category>
      <category>gitlab</category>
      <category>cicd</category>
      <category>devops</category>
    </item>
    <item>
      <title>AI Coding Agents Shouldn't Write Surprise Invoices</title>
      <dc:creator>AI CodeOps</dc:creator>
      <pubDate>Mon, 02 Nov 2026 01:40:21 +0000</pubDate>
      <link>https://dev.to/aicodeops_dev/ai-coding-agents-shouldnt-write-surprise-invoices-3c1p</link>
      <guid>https://dev.to/aicodeops_dev/ai-coding-agents-shouldnt-write-surprise-invoices-3c1p</guid>
      <description>&lt;p&gt;AI coding agents are billed by the token, and agents use a lot of tokens. A planning step, a build step, a review, a test-fixing loop: each one calls a model, sometimes many times. One long session can cost more than a week of short ones, and most teams only find out when the invoice arrives.&lt;/p&gt;

&lt;p&gt;Predictable AI spend comes down to three things: seeing the cost, limiting it, and knowing what happens at the limit.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. See the cost of every session
&lt;/h2&gt;

&lt;p&gt;You can't manage what you only see monthly. In AI CodeOps, every model call is recorded and rolled up:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Per session:&lt;/strong&gt; each session shows its total cost, plus a table of every agent step with its runtime, cost and tokens.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Per change:&lt;/strong&gt; each proposed change shows what the session that produced it spent.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Per day and model:&lt;/strong&gt; the Usage page charts daily spend by model, filterable by model and API key.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Tokens from failed model calls count too, because the provider charges for them. A cost view that skips failures will always look cheaper than the bill.&lt;/p&gt;

&lt;h2&gt;
  
  
  2. Set a budget that actually stops spend
&lt;/h2&gt;

&lt;p&gt;A budget that only sends a warning is a report, not a limit. AI CodeOps budgets have two levels:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Soft limit:&lt;/strong&gt; a marker. Past it, starting a session shows a warning with the period's spend, and the crossing is recorded.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Hard limit:&lt;/strong&gt; new sessions are refused, and running sessions stop once spend reaches it.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Budgets can run daily, weekly, monthly or on a rolling 30 days. The budget is checked before every model call, so a runaway session stops close to the limit instead of at the end of the month.&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Cap the expensive workflows separately
&lt;/h2&gt;

&lt;p&gt;Not all work costs the same. A full feature workflow with planning, building, multiple reviews and tests costs much more than a quick fix. Per-workflow caps let you keep the expensive one within its own limit while quick ones run freely under the workspace budget.&lt;/p&gt;

&lt;h2&gt;
  
  
  Know what happens at the edges
&lt;/h2&gt;

&lt;p&gt;Good controls are honest about their edges:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;If the budget check fails when a session starts, the session is blocked rather than run unmetered.&lt;/li&gt;
&lt;li&gt;A single model call already in flight can take spend slightly past the limit.&lt;/li&gt;
&lt;li&gt;Budgets apply whether a session runs on the platform's models, on credits, or on your own provider key.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Bring your own key, keep the same controls
&lt;/h2&gt;

&lt;p&gt;Many teams want to use their own model provider account for billing or data reasons. That shouldn't mean giving up spend controls. Sessions on your own key show the same cost and respect the same budgets.&lt;/p&gt;

&lt;h2&gt;
  
  
  The takeaway
&lt;/h2&gt;

&lt;p&gt;Ask your AI coding tool three questions. What did that session cost? What stops a session that's running away? Can I cap an expensive workflow separately? If the answers are "check the invoice", "nothing" and "no", the surprise is already scheduled.&lt;/p&gt;

&lt;p&gt;AI CodeOps is free during the open beta: &lt;a href="https://aicodeops.dev" rel="noopener noreferrer"&gt;https://aicodeops.dev&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Drafted with AI assistance and edited by the AI CodeOps team. Originally published on &lt;a href="https://medium.com/@aicodeops/ai-coding-agents-shouldnt-write-surprise-invoices-4a73a0350f70" rel="noopener noreferrer"&gt;Medium&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>devops</category>
      <category>productivity</category>
      <category>startup</category>
    </item>
    <item>
      <title>Your Tests Passed. On Which Version?</title>
      <dc:creator>AI CodeOps</dc:creator>
      <pubDate>Mon, 02 Nov 2026 01:39:08 +0000</pubDate>
      <link>https://dev.to/aicodeops_dev/your-tests-passed-on-which-version-4ia1</link>
      <guid>https://dev.to/aicodeops_dev/your-tests-passed-on-which-version-4ia1</guid>
      <description>&lt;p&gt;Here's a small thing that causes real incidents.&lt;/p&gt;

&lt;p&gt;An AI agent writes a change. The test suite runs and passes. Someone notices a naming issue and tweaks the change by hand, or the agent is asked for one more small fix. The pull request still shows a green check, so it merges.&lt;/p&gt;

&lt;p&gt;But the green check belongs to the version before the edit. Nobody tested what actually shipped.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why this gets worse with AI
&lt;/h2&gt;

&lt;p&gt;With people, edits after review tend to be few and visible. With AI agents, a change can go through several quick rounds: generate, review, fix, adjust. Each round is a new version. If your process treats "tests passed at some point" as "tests passed", the faster the iteration, the wider the gap between what was tested and what merges.&lt;/p&gt;

&lt;p&gt;It's a quiet failure. Nothing looks wrong. The check is green, the approval is there, and the diff looks reasonable.&lt;/p&gt;

&lt;h2&gt;
  
  
  The rule: a result belongs to a version
&lt;/h2&gt;

&lt;p&gt;The fix is a simple rule. A test result counts only for the version it ran on.&lt;/p&gt;

&lt;p&gt;In AI CodeOps, that rule is built into the release gate:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;After an agent writes a change, the session runs the project's own test command in a sandbox and records the result against that version.&lt;/li&gt;
&lt;li&gt;If anyone edits the change after its tests ran, the old result &lt;strong&gt;neither passes nor blocks it&lt;/strong&gt;. The gate marks the new version as untested.&lt;/li&gt;
&lt;li&gt;The approval opens again too, so a person decides before the edited version can ship.&lt;/li&gt;
&lt;li&gt;Re-running the session tests the new version and the gate can clear it.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Notice the second point. The old result doesn't count as a failure either. It simply stops counting, because it's evidence about a different change.&lt;/p&gt;

&lt;h2&gt;
  
  
  The same rule in CI
&lt;/h2&gt;

&lt;p&gt;The gate your CI can call follows it too. It looks at every session that touched the pull request or commit. You can also ask it to deny while a session is still running (&lt;code&gt;blockOnInFlight&lt;/code&gt;), so a merge can't sneak in between a new version and its test run.&lt;/p&gt;

&lt;h2&gt;
  
  
  What this means for your team
&lt;/h2&gt;

&lt;p&gt;You don't need a special tool to apply the principle. Ask three questions about any green check before merging:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Which commit did this run on?&lt;/li&gt;
&lt;li&gt;Has the change been edited since?&lt;/li&gt;
&lt;li&gt;Does the approval cover the version I'm merging, or an earlier one?&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;If you can't answer all three from your tooling, your process has the gap. With AI writing more of the code, it's worth closing.&lt;/p&gt;

&lt;p&gt;AI CodeOps is free during the open beta: &lt;a href="https://aicodeops.dev" rel="noopener noreferrer"&gt;https://aicodeops.dev&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Drafted with AI assistance and edited by the AI CodeOps team. Originally published on &lt;a href="https://medium.com/@aicodeops/your-tests-passed-on-which-version-9da9a9bffd4d" rel="noopener noreferrer"&gt;Medium&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>testing</category>
      <category>ai</category>
      <category>devops</category>
      <category>codereview</category>
    </item>
    <item>
      <title>AI Coding Agents Beyond GitHub: GitLab, Bitbucket, Gitea and Forgejo</title>
      <dc:creator>AI CodeOps</dc:creator>
      <pubDate>Fri, 30 Oct 2026 02:02:05 +0000</pubDate>
      <link>https://dev.to/aicodeops_dev/ai-coding-agents-beyond-github-gitlab-bitbucket-gitea-and-forgejo-52f6</link>
      <guid>https://dev.to/aicodeops_dev/ai-coding-agents-beyond-github-gitlab-bitbucket-gitea-and-forgejo-52f6</guid>
      <description>&lt;p&gt;&lt;em&gt;Most AI coding agents assume your code lives on GitHub. A lot of it doesn't. Here's what it takes to run gated AI changes on the Git host you already use.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;If you read the docs of most AI coding agents closely, one line keeps showing up: works with repositories hosted on GitHub. That's a reasonable place to start. GitHub is huge.&lt;/p&gt;

&lt;p&gt;But plenty of teams aren't there. Enterprises run self-managed GitLab. Agencies live in Bitbucket. Privacy-minded teams and a growing slice of open source run Gitea or Forgejo on their own servers. For them, "just use the agent" means moving their code first, which usually means not using the agent.&lt;/p&gt;

&lt;p&gt;We built AI CodeOps to work where the code already is.&lt;/p&gt;

&lt;h2&gt;
  
  
  One workflow, five Git hosts
&lt;/h2&gt;

&lt;p&gt;AI CodeOps connects to &lt;strong&gt;GitHub, GitLab, Bitbucket, Gitea and Forgejo&lt;/strong&gt;. The flow is the same on each:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Connect once.&lt;/strong&gt; Use OAuth for the hosted services, or a personal access token for your own server. For a self-hosted GitLab, Gitea or Forgejo, you enter the server's public https address and paste a token&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Describe a change in the browser.&lt;/strong&gt; Agents plan it, write it, review it and run the project's own tests in a sandbox&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Get a pull request on your host.&lt;/strong&gt; When the change passes the release gate, AI CodeOps pushes a branch named &lt;code&gt;codeops/run-&amp;lt;id&amp;gt;&lt;/code&gt; and opens a pull request (a merge request on GitLab) into the default branch&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;A team that uses more than one host can run all of them from one workspace.&lt;/p&gt;

&lt;h2&gt;
  
  
  What it does, and what it doesn't
&lt;/h2&gt;

&lt;p&gt;For teams that own their tooling, the second list matters as much as the first. On Gitea and Forgejo, for example, AI CodeOps:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Does&lt;/strong&gt; list the repositories, branches, files and pull requests the connected account can see&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Does&lt;/strong&gt; push a branch and open a pull request, with commits authored as "AI CodeOps"&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Doesn't&lt;/strong&gt; need admin access. An account that can read the repository, push a branch and open pull requests is enough&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Doesn't&lt;/strong&gt; install webhooks, post commit statuses or comment on your pull requests&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Never&lt;/strong&gt; pushes to your default branch directly&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Tokens are encrypted at rest and never written into a repository's Git configuration. To remove access, disconnect in AI CodeOps or revoke the token on your server.&lt;/p&gt;

&lt;h2&gt;
  
  
  The gate goes with you
&lt;/h2&gt;

&lt;p&gt;The point isn't only that the agent can reach your host. It's that the same rules apply there. A change opens a pull request only when your tests passed on that exact version, the AI review of the real diff has no open high or critical findings, and a person approved it. If a check errors, the gate fails closed.&lt;/p&gt;

&lt;p&gt;A pull request opened by AI CodeOps can be merged from the console only after the release gate passes again. You can also merge it on your host as usual, and if you want those merges gated too, your CI can ask the gate for a verdict through an API and fail the job on a "no".&lt;/p&gt;

&lt;h2&gt;
  
  
  A note for Codeberg users
&lt;/h2&gt;

&lt;p&gt;Codeberg runs Forgejo, so the connection works with &lt;code&gt;https://codeberg.org&lt;/code&gt; and a personal access token. But read Codeberg's Terms of Use first: they don't allow projects that mostly consist of code written by generative AI tools. That's their call, and worth respecting.&lt;/p&gt;

&lt;h2&gt;
  
  
  Try it on your host
&lt;/h2&gt;

&lt;p&gt;AI CodeOps runs in the browser, with nothing to install. It's free during the open beta, and &lt;strong&gt;free forever for non-commercial open-source projects&lt;/strong&gt;.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Integrations: &lt;a href="https://aicodeops.dev/integrations" rel="noopener noreferrer"&gt;aicodeops.dev/integrations&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Open source program: &lt;a href="https://aicodeops.dev/pricing/open-source" rel="noopener noreferrer"&gt;aicodeops.dev/pricing/open-source&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;em&gt;Drafted with AI assistance and edited by the AI CodeOps team.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>gitlab</category>
      <category>selfhosted</category>
      <category>ai</category>
      <category>devops</category>
    </item>
  </channel>
</rss>
