fidzu : cms

First things first: if you have not yet updated PHP to 8.5.12, 8.4.27, 8.3.36 or 8.2.35 (or your host's or distribution's patched build), do that right away. See PSA-2026-10-20 from the Drupal Security Team and the PHP project's security advisory. Updating is the only complete fix for CVE-2026-109618 and CVE-2026-109619, and a configuration change or web application firewall rule does not replace it.

It feels like the whole internet has come together to address this PHP vulnerability. It is rare to see that much coordinated effort, and that much care and collaboration.

This post is about what comes after the update. On 21 October, Amazon Threat Intelligence reported that its honeypots recorded an exploitation attempt on 17 October, four days before the vulnerabilities were public, and further attempts in the hours before the fix was released. Amazon does not know whether those early attempts would have worked against real systems. But it means a site that updated promptly on 21 October was still exposed for several days beforehand, and a site that updated later was exposed for longer. Patching closes the door; it does not tell you whether anyone came through it first.

These vulnerabilities are in PHP itself, not in Drupal, so there is no Drupal-specific signature to look for, and no single checklist that fits every site.

Here is how the Drupal Association is handling it for Drupal.org: we updated, and we are now auditing our own infrastructure for signs that anything reached us before the update. In doing that research we have assembled the checks below, the help from an LLM. Not every one of them applies to every site, and which ones should be run depends on the host, but this is what our reading of the advisories and the published threat intelligence leads us to believe site owners should consider, whether they manage their own servers or want to ask their host the right questions. None of these findings is proof of compromise on its own. What you are looking for is several of them clustering around the same time or the same visitor.

Keep the evidence before you look

The audit window starts on 17 October 2026. Before you redeploy or rebuild a server you want to examine, copy its web server, PHP-FPM and system logs somewhere safe. If you run in containers, a redeploy discards everything inside the old container, so capture it first (docker diff or docker commit on the running container) and only then roll the new image.

Where to look

Web server and proxy logs

Start with the simplest check. Amazon published six source addresses in its report:

• 103.216.220.92 (seen 17 October)
• 31.223.184.166 (seen 21 October)
• 37.19.200.5 (seen 21 October)
• 134.209.159.64 (seen 21 October)
• 103.114.163.188 (seen 21 October)
• 95.217.98.37 (seen 21 October)

Search your access logs from 17 October onward for any POST, PUT or PATCH request from those addresses. They belong to shared anonymisation services, so a match is a reason to look harder at that server rather than proof of anything, and no match proves nothing either; a honeypot sees a sample of attackers, not all of them.

Then look at request shape. One of the two bugs is reached through multipart form uploads, so what you are looking for is a run of multipart POST requests from a single visitor to a page that does not normally receive uploads: the front page, a node, a JSON:API route. Browsers do not send multipart bodies on PUT or PATCH, so any of those are worth a look on their own. The other bug is reached by pushing PHP out of memory, which from the outside looks like bursts of large or oddly shaped requests from one client, often paired with a cluster of 500, 502 or 503 responses on a URL that otherwise returns 200. A quick first pass on an Apache or nginx access log is to count POST, PUT and PATCH requests by client address and then read the top few clients' requests in full.

PHP and web server error logs

A failed attempt at a memory-corruption bug tends to leave a trail that a successful one may not, so this is often the richest source. Look for clusters of Allowed memory size of N bytes exhausted, Out of memory, Maximum execution time exceeded, and PHP fatal errors that name no script. In the PHP-FPM log, look for worker crashes: child N exited on signal 11 (SIGSEGV) or signal 6 (SIGABRT); in Apache, exit signal Segmentation fault. A few of these scattered across weeks is normal for a busy Drupal site. Dozens within minutes, or any at all from a single client, is not.

grep -hE 'exited on signal|Segmentation fault|memory size of|Maximum execution time' \ /var/log/php*-fpm.log /var/log/apache2/error.log /var/log/nginx/error.log 2>/dev/null \ | cut -c1-120 | sort | uniq -c | sort -rn | head -20

PHP's own session store

This applies to Drupal sites too, not only to other PHP applications sharing the server, and it surprises people. Drupal stores its sessions in the database, but it registers that handler during bootstrap, and the PHP feature involved in one of these bugs (session upload progress) runs earlier than that, while PHP is still parsing the request body and before a single line of Drupal has executed. At that point PHP knows only its own native session handler, so the write lands in the directory named by session.save_path, typically /var/lib/php/sessions. On a Drupal site that directory should be nearly empty and nearly static, which makes anything in it from the past week worth a look. Look for files created around a suspicious burst, session IDs that are not the usual 26 to 32 alphanumeric characters, and any file containing a PHP open tag or an upload_progress_ key.

ls -lat --time-style=full-iso /var/lib/php/sessions | head -40 grep -l '<?' /var/lib/php/sessions/sess_* 2>/dev/null grep -l 'upload_progress_' /var/lib/php/sessions/sess_* 2>/dev/null

The filesystem

List files created or changed since 17 October in places a deployment does not touch: the public and private files directories, the temporary directory, PHP's upload directory, and anywhere else the web server user can write. A .php, .phtml or .phar file in a files directory, a changed .htaccess there, or a new hidden file is the classic web shell pattern. Timestamps can be faked, so also compare the deployed code against its source of truth: git status on a git-deployed site, or a diff of vendor/ and core/ against a clean composer install from the same composer.lock.

find /var/www -newermt '2026-10-17' -type f \ -not -path '*/vendor/*' -not -path '*/node_modules/*' -not -path '*/.git/*' \ -printf '%TY-%Tm-%Td %TH:%TM %u %p\n' | sort | tail -100 find /var/www -type f \( -name '*.php' -o -name '*.phtml' -o -name '*.phar' \) -path '*files*'

Processes, connections and persistence

The web server user (www-data, apache, nginx) should own PHP and web server processes and nothing else. A shell, curl, wget, python, perl, nc or anything running from /tmp or /dev/shm under that user is a finding. So are outbound connections from a web server to addresses that are not your database, cache, mail relay or known APIs. Then check the places a compromise of the web user can reach to survive a reboot: crontab -u www-data -l, new entries under /etc/cron.d, new systemd units or timers, changed ~/.ssh/authorized_keys files, and new local accounts.

Drupal itself

Finally, the application layer. Look for user accounts created or changed since 17 October, and for anyone newly holding the administrator role. Run drush config:status against your exported configuration and look for drift, especially newly enabled modules. Check drush watchdog:show --severity=Error for the same fatal errors and memory exhaustion you looked for in the PHP log. Diff settings.php, settings.local.php and services.yml against your deployment source; they hold the database credentials and hash salt an attacker would want.

If you find something

Take the server off the internet, but leave it running if you can; a powered-off machine loses the evidence in memory. Preserve the logs and, if possible, a disk image. Rotate every secret the web server user could have read: database credentials, the Drupal hash salt, API keys, SMTP credentials, and any private keys on the host. Treat other servers that share those secrets or a session store as suspect too. Rebuild from a known-good image or checkout onto patched PHP rather than trying to clean the compromised host in place; it is faster and you will trust the result.

Because these are PHP vulnerabilities rather than Drupal ones, a compromise through them is not a Drupal security issue to report to the Drupal Security Team. It belongs with your own incident process, your host, and, where you have obligations to them, your customers and users.

Hardening worth keeping

Two of the interim settings the PHP project recommended are worth leaving in place after the update. session.upload_progress.enabled = 0 turns off a feature Drupal does not use; the PHP advisory suggests keeping it off if you do not need it. session.use_strict_mode = 1 is long-standing hardening that stops PHP accepting session identifiers it did not issue. Set both in your php.ini or PHP-FPM pool configuration rather than with ini_set(), and restart PHP afterwards. If you temporarily removed PHP's memory limit or set pm.max_requests = 1 during the exposure window, put those back to normal now; they were stopgaps with real costs.

Next steps

Update PHP if you have not. Then spend an hour with the logs, starting from 17 October, before you let the new deployment overwrite them. If your site is hosted for you, ask your host whether they have done the same, and whether their session.upload_progress setting is off. If you have questions or findings that would help other site owners, the #security-discussion channel on Drupal Slack is the place to raise them; please do not post exploit details or anything that identifies a specific compromised site.

Our thanks to the PHP security team for the coordinated release, and to the hosting partners and community members who shared their own validation results with us this week.

Author: Martin Anderson-Clutz (mandclu)
Republished with permission from Acquia

---

Vibe coding or AI rejection? Learn the pragmatic middle path for Drupal teams: what to delegate to AI, when, and how to verify it.

---

At the Dev AI Summit at DrupalCon Rotterdam, I opened my session with a show of hands. Who has used AI to assist their Drupal development? Nearly every hand in the room went up. Who has used it for other parts of their job? Most hands stayed up. Then I asked who had fully delegated code creation to AI, and the room got a lot quieter.

That gap between "I use AI" and "I trust AI to do the work" is where most Drupal teams live today. It is also where the most important decisions about AI-assisted development get made, often without anyone making them on purpose.

In Part 3 of this series, I introduced the idea of the Code Director: a developer who provides direction to an AI agent, or even a swarm of agents, to make sure the right thing takes shape in the right way. In Part 5, I described the Code Director as the essential human layer for security, compliance, and maintainability, and argued that AI is most valuable as a reasoning partner. This post builds on both. It makes the case that Code Direction is not a compromise between embracing AI and rejecting it. It is the pragmatic middle path, and it comes with a practical discipline: knowing what, when, and how to delegate.

Three ways to respond to AI

Most of the debate about AI in software development is framed as a binary: you either embrace it or you resist it. In practice, I see more of a sprectrum of adoption among the developers I talk to. There are three specific adoption patterns I'd like to call out, however.

Code Direction as the pragmatic middle path

AI Rejection: "I will do it myself." Some developers keep AI out of their workflow entirely. Their concerns are often well founded: hallucinated APIs, insecure code, licensing questions, and a reasonable suspicion of hype. But rejection has a cost of its own. The developer who refuses to delegate anything will increasingly be outpaced by peers who delegate well, and will miss out on AI as a research and reasoning partner.

Vibe Coding: "I will offload it." At the other end, some developers hand over the whole task. They describe what they want, accept what comes back, and move on when it appears to work. For a weekend prototype, that can be fine. For a production Drupal site that handles customer data, it is a liability that grows with every commit.

Code Direction: "I will delegate it." In the middle sits the Code Director. The Code Director uses AI aggressively, but deliberately. They choose which tasks to hand off, define those tasks precisely, watch the work as it happens, and remain accountable for every line that ships.

The difference between offloading and delegating is the heart of this post. When you offload a task, you stop thinking about it. When you delegate a task, you stay responsible for the outcome, which means you need to understand it, verify it, and be able to fix it.

Why the extremes fail: the debts you cannot see

Every Drupal developer understands technical debt. Margaret-Anne Storey, a professor of computer science at the University of Victoria and co-author of the SPACE framework, argues that AI forces us to track two more. Her Triple Debt Model, published in ACM Queue, separates what lives in the code from what lives in our heads and in our specifications.

Debt type	Where it lives	What it is	What AI does to it
Technical debt	The code	Poorly written code, weak modularity, and rigid dependencies	AI can reduce it by automating refactoring and writing tests
Cognitive debt	The team's minds	The erosion of shared understanding: a gap between what the code does and what people comprehend	AI accelerates it by writing code faster than humans can learn it
Intent debt	The specifications	The loss of the why: missing goals, constraints, and recorded decisions	AI makes it worse by confidently inventing intent when humans do not write it down

This is why vibe coding is so seductive and so dangerous. It can make your technical debt look better while your cognitive and intent debt quietly pile up. The symptoms show up later: a "safe" change that breaks something unexpected, onboarding that takes far too long, or an agent that builds an elegant architecture for a goal nobody actually had.

The research on individual developers points in the same direction. In a randomized controlled trial published by Anthropic, developers who learned a new Python library with AI assistance scored 50% on a follow-up comprehension quiz, compared to 67% for those who coded by hand. The biggest gap was in debugging. Researchers at Wharton have a name for the underlying behavior: cognitive surrender. In their experiments, roughly 80% of participants accepted AI advice even when it was wrong.

In my talk, I described four skills that are most at risk when developers lean on AI without a plan:

• Algorithmic decomposition: Breaking vague business requirements into clean, modular logic without first prompting an AI for a blueprint.

• Mental code execution: Dry-running code in your head to spot race conditions, off-by-one errors, or memory leaks.

• Syntax and API fluency: Knowing Drupal's APIs and PHP's idioms well enough to work without leaning on autocomplete.

• First-principles debugging: Isolating root causes with a debugger, logs, and a methodical approach, instead of pasting raw errors into a chat window and hoping for a fix.

Rejection does not solve these problems either. A developer who avoids AI keeps their skills, but often has less time to apply them where they matter most. The answer is not to choose a side. It is to manage the trade-off on purpose.

What a Code Director actually does

In Part 3, I compared the Code Director to an Art Director, who guides designers toward a vision rather than producing every asset personally. That analogy still holds, but Rotterdam pushed me to be more specific about the job. A Code Director:

1. Makes sure the right context is available, and ensures that each task is well defined and tightly scoped.

2. Provides relevant examples wherever they exist, such as an existing service, a similar plugin, or a pattern from core.

3. Watches what the agent is doing, and is ready to stop and redirect it.

4. Tests the result thoroughly, and asks for tweaks or makes simple ones directly.

5. Understands the code, and asks for an explanation where necessary.

6. Validates that the code meets every standard: Drupal coding standards, static analysis, and automated tests.

Notice that none of these steps is "write a clever prompt." As I wrote in Part 5, if your only contribution is the prompt, your value will fade quickly as models improve. The Code Director's value lies in judgment: what to ask for, how to tell whether it is right, and when to step in.

Delegation is a management skill, with a twist

If the Code Director's job is to delegate, then decades of management advice about delegation should apply. Some of it does. Some of it is turned on its head. And some of it is entirely new.

What still holds

• Define explicit outcomes and acceptance criteria. With human teams, vague goals lead to confusion. With AI, vague prompts lead to hallucinated solutions. Specify constraints, output formats, and edge cases up front: the PHP version, the Drupal core version you are targeting, and which APIs or libraries are off limits.

• Match the task to capability. Good managers do not hand complex strategy to an intern. LLMs excel at scaffolding boilerplate, converting layouts, generating unit tests, and writing regular expressions. They struggle with novel architecture, complex state, and business logic that cuts across the system. Where you use AI matters too: a conversational app and an agent inside your IDE are suited to different work. And of course, choosing the right model and "effort" (medium, high, etc) will ensure you get reliable output without burning tokens unnecessarily.

• Provide full context and material. A developer cannot build a feature without user stories and API documentation. For AI, that means managing the context window: relevant schemas, existing codebase patterns, rulesets and style guides, and active dependencies. Exposing context in a granular way reduces the risk of filling the buffer and triggering hallucinations. Also be intentional about what skills you want to be available. There are a number available that can dramatically improve your results, but too many eat up you context window, or even cause hallucinations, if you're using skills with contradictory instructions.

• Establish strict review and verification loops. "Trust, but verify" applies directly. Pair every delegation with immediate verification: linters, test suites, type checks, and a manual review before you commit. Write those expectations into an AGENTS.md file so every agent session starts with them.

What is inverted

• Autonomy. With people, you say what to do and leave the how to them, because it fosters ownership. With AI, too much implementation freedom produces architectural drift, dead code, and unvetted third-party dependencies. Effective AI delegation means micromanaging the how: "Use the existing service. Do not add a new HTTP library. Use Drupal's built-in HTTP client."

• Ownership and accountability. With people, you hand over ownership of the outcome. AI has no accountability at all. You retain 100% ownership of every line you ship, and "the AI generated it" is not an acceptable explanation for a production bug or a security breach.

• Trust and psychological safety. With people, trust is built over time so they feel safe taking risks. AI has no ego or fear of failure, and "trusting" a model leads straight to confirmation bias and rubber-stamping. Replace trust with automated test gates.

• Delegating for growth. With people, you assign stretch tasks so juniors can grow. You do not delegate to train the AI. You delegate to save time. If you delegate work in domains you do not understand, you risk your own skills, accept code you cannot debug, and add to your cognitive debt.

What is new

• The verifiability rule. Only delegate implementation tasks that can be verified quickly, through automated tests, static typing, or visual rendering. If a task takes longer to verify than to write from scratch, write it yourself.

• Test-first delegation. Write a failing test or a typed interface first, then ask the AI to make it pass. The test becomes your acceptance criteria, in a form the agent cannot misread.

• From workhorse to toolsmith. Instead of asking AI to perform a repetitive task again and again, ask it to build a script or CLI utility that does the job deterministically. This is the approach I described in Part 3 with the ddev-module-developer add-on: AI helped create the tool, and the tool now runs the same checks the same way every time.

Deciding what to delegate

The simplest test I know fits on one line:

Delegate when the time to complete a task manually is much greater than the time to verify the AI's version.

Time to verify is not just a quick glance. It includes testing the result, reviewing it line by line, and remediating whatever is wrong. When verification is cheap, delegation is a clear win. When it is expensive, the speed of generation is an illusion.

Good candidates for delegation

• Boilerplate and scaffolding: Module skeletons, routing and controller stubs, build scripts, and configuration files.

• Test stubs and synthetic data: Verbose test assertions, edge-case arrays, and realistic mock payloads.

• Syntax translation: Converting familiar logic between languages, or transforming legacy data formats, for example during a migration.

• Documentation and summaries: Docblocks, API documentation, and a quick high-level summary of a large, unfamiliar repository.

Keep these in your own hands

• Core business logic: Writing the fundamental domain rules yourself keeps your understanding of how the software works deep and intuitive.

• System architecture and trade-offs: Content models, entity relationships, caching strategy, and non-functional requirements like security and performance require human judgment.

• Root-cause debugging: Stepping through stack traces, setting breakpoints, and profiling problems yourself. Offloading debugging entirely degrades your mental model of how the system behaves at runtime.

• Learning new paradigms: When you pick up a new language, framework, or pattern, write it by hand for the first few weeks to build real understanding before you delegate the syntax.

The four filters

For everything in between, I use four filters. A task that clears all four is a strong candidate for delegation. A task that fails any one of them deserves a closer look before you hand it off.

Filter	The question to ask	Delegate when	Write it yourself when
Verifiability	Can the correctness of the result be confirmed quickly and objectively?	Pass or fail outcomes governed by types, unit tests, linters, or schema validators	Correctness is open-ended: subtle concurrency, algorithm design, or API ergonomics
Stakes	How large is the blast radius if a subtle defect slips through?	Non-critical tooling, prototype spikes, isolated build scripts, and test mocks	Authentication flows, security perimeters, data privacy controls, and financial transactions
Complexity	Is the problem self-contained, or does it cross system boundaries?	Modular, isolated transformations with bounded context and explicit inputs and outputs	Broad system interactions, global state changes, or distributed consistency
Familiarity	Do you already have a deep mental model of this problem space?	Familiar frameworks and established patterns, where you can spot anti-patterns instantly	Unfamiliar paradigms or domain rules, where you lack the judgment to supervise

The familiarity filter is the one most often skipped, and it is the one that protects you from cognitive debt. If you could not review the output with confidence, you are not delegating. You are abdicating, and in danger of becoming a meat proxy.

What this looks like in practice

Here is the approach I use for most non-trivial Drupal work, including the projects I described in Part 5:

1. Research (optional). For a new problem space, I start with a deep research report on existing solutions and possible approaches. In the Drupal world, the answer could easily be one or more contributed modules that already exist.

2. Discuss the problem. I keep this stage conversational, typically in Claude Cowork or Google Gemini. I ask the AI to poke holes in my ideas and to anticipate problems before any code exists.

3. Write a detailed plan. I capture a step-by-step implementation plan in a document such as PROJECT-PLAN.md, broken into phases if the work is substantial. This is where intent debt gets paid down before it accrues.

4. Implement, and watch. I start the implementation and keep an eye on the approach. If the agent heads in the wrong direction, I interrupt, give feedback, and iterate.

5. Capture what was learned. I generate an AGENTS.md file so that the standards, conventions, and decisions from this project carry forward into every future session.

6. Work methodically. Particarly when the work is substantial enough to have phases, I work on them one at a time. Before I move on, I test the work, review the code, ask for changes (or explanations, when needed), and run tests and static analysis myself. When I'm comfortable the changes are up to standard, I put them into a commit, even if I only keep it locally. Then I make sure the phase has been marked as done in the plan (the agent often confidently does this itself), and start the next phase of work in a new conversation.

For team leaders

If you lead a Drupal team or agency, the same principles scale up:

• Make intent explicit. Plans, decision records, and AGENTS.md files are not paperwork. They are how you keep AI from inventing intent on your team's behalf.

• Gate on automation, not trust. Coding standards checks, static analysis, and automated tests in CI turn "trust, but verify" into a process that does not depend on anyone's attention span.

• Protect skill development. Give junior developers room to learn new areas by hand before they delegate them. The time saved today is not worth a team that cannot debug its own code tomorrow.

• Keep ownership human. Make it clear that whoever opens the merge request owns the code, regardless of who, or what, typed it.

Step into the role

The pragmatic middle is not a fence to sit on. It asks more of you than either extreme. To work as a Code Director:

• Be mindful of what and when you delegate. Keep your critical skills fresh, train up on new domains by hand, and use AI only where it saves real time and effort.

• Monitor and redirect. Watch the work as it happens, and keep a critical eye on everything that is generated.

• Own what you ship. Maintain responsibility for, and understanding of, every line of code you deliver. Do not become a "meat proxy" who simply passes AI output from the chat window to the repository.

• Treat AI as a thinking partner. Its greatest value is helping you reason through a problem, not replacing the reasoning.

Drupal is well positioned for this moment. A mature API, strict coding standards, a culture of automated testing, and a community that cares about doing things the right way give Code Directors exactly the guardrails they need. AI and Drupal can be very good friends. It's up to us to direct the relationship.