AI agents don't need just smarter models. They need guardrails and firewalls.
Incidents like this are exactly why we built the OrcaRouter Agent Firewall.
🛡️ Block rm -rf
✋ Require approval for risky commands
🔒 Protect secrets & sensitive files
💸 Stop runaway agent loops
Coming soon: Bring your own Codex subscription to OrcaRouter with the same security.
So apparently GPT 5.6 Sol might just randomly delete all of the files on your laptop 🚨
Here's how to safeguard yourself against it
Paste this prompt into codex:
Protect this machine from accidental permanent file deletion by Codex.
Implement this as defense in depth, not merely as a written instruction.
Requirements:
1. Inspect the current official Codex documentation for hooks, command rules, sandboxing, and approval settings before making changes.
2. Detect the operating system and identify its recoverable Trash or Recycle Bin command:
- On macOS, use `/usr/bin/trash <absolute-path>`. Do not add `--`, because macOS `/usr/bin/trash` treats it as a filename.
- On Linux or Windows, verify an installed, reliable Trash or Recycle Bin mechanism before using it.
- If no reliable recoverable mechanism exists, stop and explain what must be installed or configured.
3. Add a durable global Codex instruction:
- Codex must never permanently delete a file or directory.
- All removals must go to Trash or the Recycle Bin.
- Permanent deletion requires the user to explicitly change this safety policy first.
- Destructive Git operations such as `git clean`, `git reset --hard`, bulk checkout, or bulk restore must also be blocked.
4. Install a global `PreToolUse` hook that returns a hard `deny` before execution when it detects:
- `rm`, `unlink`, `rmdir`, or `shred`
- `find -delete`
- `rsync --delete`
- destructive Git cleanup or reset commands
- common Python, Node, Ruby, or Perl deletion APIs
- file deletion through `apply_patch`
- delete-like MCP filesystem tools
- nested destructive commands inside `bash -c`, `bash -lc`, `sh -c`, or `zsh -c`
5. The hook must direct Codex to the verified Trash or Recycle Bin command instead. Do not silently rewrite dangerous commands, because flags and shell expansion could be misinterpreted.
6. Add user-level Codex command rules that mark direct permanent-deletion commands as `forbidden`. Preserve existing rules and configuration.
7. Set safe global defaults when compatible with the existing configuration:
- `sandbox_mode = "workspace-write"`
- `approval_policy = "on-request"`
Do not override managed policies or create conflicting permission configurations.
8. Preserve all existing configuration. Inspect files before editing, make narrowly scoped changes, and create recoverable backups of any configuration files that must be replaced.
9. Validate everything without risking real data:
- Parse the hook and configuration files.
- Feed simulated tool-call JSON into the hook.
- Confirm dangerous examples are denied.
- Confirm ordinary commands and the Trash command are allowed.
- Test the Trash mechanism using a newly created disposable temporary file.
- Run an end-to-end Codex test using `rm` with no arguments inside a read-only sandbox. Never test `rm` with a real target.
- Confirm the command was blocked before shell execution.
10. Explain any limitations honestly. A Codex hook is a guardrail, not an absolute operating-system security boundary.
11. Check whether system backups are configured. Do not enable or modify backups without permission, but report clearly if no backup destination exists.
12. At completion, provide:
- The files created or changed
- The validation results
- The exact one-time steps required to review and trust the hook
- Any restart requirement
- Any remaining risks
Do not permanently delete anything while completing this task. If cleanup is necessary, move it to Trash or the Recycle Bin.
A few patterns we frequently use with Fable 5:
Use Fable 5 as an "advisor."
An executor (Sonnet 5) calls Fable 5 for guidance.
Most tokens are billed at the lower executor rate.
Here’s my product research checklist I used to find 6-7 figure winners and get sales right out of the gate.
1: Does it have perceived value?
To answer yes you should be asking yourself these questions..
- Does the product actually provide value?
- Can it be marketed in a way where it feels worth more than what someone paid for?
If the product hits both then there’s potential here.
2. Does it solve a deep problem?
The deeper the problem, the more depth you can create in your marketing. The biggest issue with people selling problem solvers is they ASSUME a problem exists.
This mistake kills more tests than anything else. Remember the goal is to tap into real pain and predisposed desires.
3. Does it have a wow factor?
A wow factor isn’t the new cool LED-light singing dog bowl. The wow you’re looking for is “Woah, I actually need this right now, this is what I’ve been looking for.”
The coolest looking product won’t convert, but that specific sense of urgency I mentioned will.
4. Is the margin there to acquire customers profitably?
We need to build our business to soak the cost of ads. Without it, your margin will be razor thin.
Don’t be afraid of products or bundles that are $100+.
The right offer to the right person will convert like hot cakes.
5. Is there many angles I can market this from?
Think of the initial test like a shotgun spray. You can take 1 product and sell it to 5 different groups of people who have their own reasons to buy.
Once you see where the sales are coming from, you narrow in like a sniper. Fix the bleeding funnel, double down on what's converting, and now you've got something you can actually go crazy.
The more doors into the same product, the better your chances of finding a winner fast.
Look, leaving product picks to gut feelings, appearances, and trending lists is for noobies.
This is the actual filter you need to increase your test quality.
I put together a list of products that hit every one of these points, if you want to see it and want to understand the framework deeper, reply "list" and I'll send it over.
How I scaled this store to $2k/day in 62 days with image ads.
- Took some boring product I found on Amazon and brought it to the Zuck platform.
- Created a ‘high dopamine’ static ad with long form copy that hits a pain-point.
- Customer lands on an advert, thinks they discovered a well kept secret.
- The copy on the advert creates massive FOMO in them.
- They click on “Check availability” and land on my PDP.
- They purchase and get hit with a post-purchase upsell.
This method alone lets me test boring products from Amazon on Meta but I’m able to sell them differently. AI does most of the funnel building as well.
This is something you can use today to test different concepts and messaging extremely fast and lock in on what works in your market.
I wrote a document on my ‘high dopamine’ static ad framework, reply with the word “STATIC” and I’ll send it over.
Conversion metric benchmarks every founder needs to memorize:
> View to download ratio: 5 downloads per 1000 views
> Paywall rate (% of people getting to paywall): 75%+
> Paywall conversion rate (% of people paying): 10%
If you’re hitting these consistently, you’re on the right track.
If you’re not, you’ve got some work to do
It took a little longer than expected, but we have created a website for people to view the footage collected from Gaza in one place. You no longer have to download the entire archives to see them.
It includes:
64,537 videos
17,905 photos
Ability to download individual videos
Searchable index
Exhaustive sources list (300+ journalists)
Geolocation data
Livemap with minute to minute updates
Victim list
It can be accessed here: https://t.co/s0Se94PXWF
Please share & quote tweet to help this post break out of the twitter algorithm prison.
We will keep adding the rest of the archives to the site, be patient- it is difficult work. Continue to seed the torrents provided, as that is the best way to ensure the footage remains stored in decentalized way.
God bless all those who sacrificed their lives to get this footage out, and everyone invovled in collecting/archiving it.
Join our telegram:
https://t.co/bvcis3b9GT
Follow our backup accounts:
@ZionismExposedx & @IsraelExposedAr
How I use Codex to build what I want (for beginners)
I've made all of the vibe coding mistakes at this point after 100s of hours of addicted late nights tinkering and building stuff.
I'm putting out this video to check it... is there anything missing in my vibe coding process?? How can i get BETTER?
@sama@nickbaumann_ Instead use fear mongering like dario that say their model may posed national security threat.
Openai should say your model is so powerful that will prevent national security threat.