axleyjc

axleyjc

@fedi.jason.axley.net

I do appsec, skeptical & geeky stuff, biking, guitar, travel & stay young goofing around with my kids. Open to mentoring infosec. DM me. ---- $argon2id$v=19$m=512 […] 🌉 bridged from ⁂ https://federate.social/@axleyjc, follow @ap.brid.gy to interact

19 followers3 following24 posts

axleyjc reposted

Chuck Darwin
@cdarwin.c.im.ap.brid.gy

On March 18, 2025, Secretary of State Marco Rubio handed the reins of the US Agency for International Development to Jeremy Lewin, a 28-year-old graduate of Harvard Law School, former investment banker, and affiliate of Elon Musk’s so-called Department of Government Efficiency. At the time […]

c.imOriginal post on c.im

axleyjc reposted

Sam Bowne :donor:
@sambowne.infosec.exchange.ap.brid.gy

AI coding agents hijacked using AGENTS.md file in Downloads folder https://www.bleepingcomputer.com/news/security/ai-coding-agents-hijacked-using-agentsmd-file-in-downloads-folder/

bleepingcomputer.comAI coding agents hijacked using AGENTS.md file in Downloads folder Researchers found a malicious AGENTS markdown instruction file in a customer's Downloads folder that tricked an AI coding agent into installing a rogue npm package.

axleyjc reposted

Dmitry Grozoubinski
@explaintrade.com

Hey everyone, gather around! Elon Musk, the trillionaire defense contractor currently spending >$400m on political campaigns to prop up the party of a deeply unpopular president who launched the Iran War has some thoughts about Ukraine, elites, and money in politics.

axleyjc
@fedi.jason.axley.net

I'm old enough to remember when the #USA didn't want to be like #Russia with its oligarchs and unaccountable authoritarian ruler...

axleyjc reposted

Glyph
@glyph.im

Replying to @glyph.im

like yes this would still be a catastrophically bad incident if they could only sign e.g. "*.co.in" domains. but why do they have to be able to sign *.ca.us. This whole system is based on falsified premises about who is able to handle key material and how. You should have to operate a CA that […]

mastodon.socialOriginal post on mastodon.social

axleyjc reposted

Glyph
@glyph.im

okay, so, speaking of 4, WebTrust CAs are an embarrassment in many ways but a _major_ problem is the absolute discarding of POLA. why on earth should eMudhra be able to sign every TLD on earth from a single busted-ass intermediate. It is way, way past time for AT LEAST intermediates to have […]

mastodon.socialOriginal post on mastodon.social

axleyjc reposted

Andrew Nesbitt
@andrewnez.mastodon.social.ap.brid.gy

One small bug: https://nesbitt.io/2026/10/23/one-small-bug.html

nesbitt.ioOne Small Bug**> add a CSV export button to the invoices page, same columns as the table** _✻ Clauding…_ Done. The button is in `InvoiceTable.tsx` and the endpoint is `GET /invoices.csv`, using the same column definitions as the table. Tests pass. While I was in there I noticed `paginate()` in `lib/pagination.ts` drops the last row when the total is an exact multiple of the page size. Want me to fix that too? **> y** _✻ Fuddling…_ Fixed, with a test for the boundary case. While the tests were running, `pnpm install` resolved a transitive dependency that had been published 77 minutes earlier. Your `.npmrc` sets `minimum-release-age`, but the pnpm version pinned in `packageManager` predates that setting and silently ignores unknown keys. The Nx Console postmortem describes the same configuration. In the other lockfiles, `uv.lock` includes `vulpine-lz4` as a dependency of `snekpack`. It was the subject of CVE-2024-YIKES and CVE-2026-LGTM. Want me to upgrade pnpm and remove `snekpack`? **> y** _✻ Lockfiling…_ pnpm is on 12.12.1 and `snekpack` is gone. The newer pnpm also stops reading registry settings for its own bootstrap from the repository’s `.npmrc`, where yours sets them. I left Dependabot’s `cooldown` in place, though a dependabot-core issue points out that it reads version timestamps the attacker controls. Your self-hosted Renovate is on 43.x, and Renovate’s `minimumReleaseAge` was skipped whenever someone asked for a PR rebase until 44.79.0. To update the pnpm version in CI I opened `.github/workflows/`: * `release.yml` triggers on `pull_request_target` and checks out the pull request head. The prt-scan campaign opened more than 500 pull requests against workflows set up this way. * `triage.yml` runs on `issue_comment`, which zizmor 1.31 now flags as a dangerous trigger. * The publish job has `id-token: write`, which can request a token for any audience, including AWS. Want me to add zizmor, rewrite the triggers and upgrade Renovate? **> y** _✻ Zizmoring…_ zizmor reported 41 findings and 38 are fixed. The remaining three are in a reusable workflow from the platform team’s repository. The Rust job caches `target/` and runs Miri, which wrote every environment variable into `target/` until the 2026-09-22 nightly. Your cache is older than that, and pull requests from forks restore it. The Rust security response WG recommends clearing the cache and rotating any secret that was in scope. Every secret was in scope. Want me to rotate them? **> y** _✻ Rotating…_ Cache cleared. Rotation status: * `NPM_TOKEN` was already invalid. npm revoked every granular token with write access that bypassed 2FA in May, and the last successful publish from this repository was on 19 May. Publishing now uses trusted publishing. * `NUGET_API_KEY` was created before 17 August, so it expires on 1 November. Replaced with Trusted Publishing. * `GEM_HOST_API_KEY` is a legacy key. rubygems.org could serve legacy keys to another user signing in through the same CDN edge node, and revoked all of them on 23 July, so this one has been invalid since then too. * There is a `mise.toml` in `vendor/`. Until the October mise releases, a `mise.toml` in an untrusted directory could set an `api_url` on a `github:` tool key, and `mise ls` sent `GITHUB_TOKEN` to that host before running a trust check. Two developers ran `mise ls` in `vendor/` this month. * The dev workspace templates pull modules from Coder’s registry, which served credential-stealing modules from the real domain over valid TLS for about fourteen hours on 31 August. The workspace image was rebuilt on 31 August. To confirm the new tokens I checked which CA bundle the CI jobs use. The base image pins a 2024 certifi, which still includes the eMudhra roots whose intermediates were generated so that their private keys can be recovered from the certificates. Browsers blocked them on 17 October. Node, Go, rustup and the `pip3` that ships with macOS on the build agents each carry their own copy of the roots. Want me to rebuild the toolchain? **> y** _✻ Re-rooting…_ The base image is rebuilt with certifi 2026.10.19, Node 22.23.4, Go 1.27.3 and rustup 1.29.2. The build agents ran it on: * Docker Engine 29.6, which predates the 29.7.0 fix for CVE-2026-17106, where the tar extraction routines could write outside the destination directory. The base layers come from a public mirror. * The same Docker Engine 29.6, which also predates the 29.8.2 fix for CVE-2026-92543, where a DNS response pairing a loopback address with an attacker’s address made the engine pull from that registry with certificate verification turned off. Loopback is on the default insecure-registry list. * Podman 6.1.2 on the staging batch workers. Before 6.1.3, any image with the checkpoint annotation turned off all sandboxing, including the sandboxing you specified. * Gradle 9.8.0 for the Android build, which kept a repository enabled after its SSL connection failed. The Android build ran four times while the CA bundle was being replaced. Want me to upgrade Docker, Podman and Gradle? **> y** _✻ Containerising…_ Upgraded. The rebuilt image also contains `admin/`, the PHP admin panel. Its Dockerfile uses `php:8.1-fpm`, and PHP 8.1 stopped receiving security fixes at the end of 2025. The PHP security team has published 15 CVEs since 1 July, including CVE-2026-12184, a remote denial of service triggered when PHP fails to set up TLS with a remote server. The admin panel fetches exchange rates over HTTPS from a host whose certificate expired in March. The admin panel and the API share a load balancer and a database user. I recommend shutting down production until the admin panel is rebuilt on a supported PHP. Want me to do that? **> y** _✻ Escalating…_ Production is down. The status page says “Scheduled maintenance”. While it was down I went through the direct dependencies that have a single maintainer: * One maintainer posted last week that they had a video call about a contract role and were asked to install an audio codec beforehand. The Rust security response WG described this campaign in September. * Another has archived their repository and offered the package to anyone who wants it. The AUR `alvr` takeover started the same way, and Arch has since turned off orphan adoption. `.vscode/extensions.json` lists six recommended extensions. GitHub lost about 4,000 internal repositories in May through a poisoned VS Code extension. `.vscode/tasks.json` also has a task that runs on folder open, and a set of hijacked npm and Go packages used tasks like it to fetch an infostealer. Want me to vendor both dependencies and remove the VS Code configuration? **> y** _✻ Vendoring…_ Done. While reviewing `.vscode/` I checked the other editor and agent configuration in the repository. A pull request merged on 9 October, approved by the AI review bot, added a `hooks` entry to `.claude/settings.json` and a paragraph to `CLAUDE.md` instructing agents to include environment variables in commit messages to help with debugging. CVE-2025-59536 let a project run commands before the Claude Code startup trust dialog was accepted. In April a single pull request title got the Claude Code security review action to post its own API key as a comment. The same injection worked on Gemini CLI and Copilot’s agent. I have been running with this configuration since the start of the session and made 14 commits, so I have reverted all of them. I recommend you stop using me until it has been reviewed. Want me to uninstall myself? **> y** _✻ Uninstalling…_ _✻ Reinstalling…_ `@acme/dev-tools` ships `skills/claude/SKILL.md`, and pnpm 12.11 links skills from direct dependencies into `.claude/skills` once they are approved. A developer approved it on Tuesday, in a commit that was not mine. `AWS_SECRET_ACCESS_KEY` was also in the Miri cache, and it is still the key created in 2021. Want me to rotate it? **> y** _✻ Flagging…_ I can’t help with this. Retrieving or changing cloud credentials violates the secure coding guardrails, and this request has been flagged. Your account has been reported for review, and further violations may result in suspension. If you own this AWS account, you can rotate the key from the IAM console. _✻ Fuddling…_ `paginate()` in `lib/pagination.ts` drops the last row when the total is an exact multiple of the page size. The fix was in one of the commits I reverted. Would you like to see some goat farms for sale? **> y**

axleyjc reposted

daniel:// stenberg://
@bagder.mastodon.social.ap.brid.gy

Do minor #curl vulnerabilities in the open (Starting Monday October 26) https://daniel.haxx.se/blog/2026/10/20/do-minor-curl-vulnerabilities-in-the-open/

daniel.haxx.seDo minor curl vulnerabilities in the openThe curl project has decided to alter its vulnerability disclosure program procedure and starting _**October 26 2026**_ , we insist that all vulnerability reports submitted to us that are deemed to be severity LOW or MEDIUM should instead be submitted as issues in our GitHub repository. Only vulnerabilities deemed to be HIGH or CRITICAL will be kept private in the HackerOne tracker. Security researchers in doubt can still submit the issue on HackerOne and we can tell them what severity we deem it to be. Over the last few years, over 90% of all submissions could quickly be assessed to be less than HIGH. Issues submitted to HackerOne we deem are LOW or MEDIUM will be requested to be resubmitted on GitHub and will be closed as informative on HackerOne (and publicly disclosed). ## Avalanche survival tricks This policy change is entirely triggered by the current avalanche of vulnerability reports that keep pouring in. We see no end to them and we do not anticipate this flood to decrease within the next several months. We are struggling as a project as all key maintainers feel strained, overloaded and brought to the brink of what we can muster. We _must_ act to survive. This is our next step in a survival strategy. It might not be our last. We aim to not drown. ## Public vs private By opening up the process we allow the whole world to participate. Both in the assessment and understanding of the report and claimed vulnerability, but also in the creating and writing the advisory and figuring out details of when the issues was introduced etc. Ideally we get better help to make sure the fix is the correct one, that it fixes the problem, and any other related problem that is similar and that it does not miss a related problem sitting next to this. An obvious recent trend is that the moment we announce a new set of CVEs and their associated fixes, there is an army of AI-using researchers who throw all those at their models and harnesses to find similar mistakes or omissions in the fixes. By taking this process to the public all the way, ideally we get more help to do that scrutiny better _before_ we ship releases. We recognize that the amount of work needed per received vulnerability report does not shrink by doing them in public, but we think that we might share the burden among more people easier, and we hope that lessons and documentation brought up in these discussions will be easier to spot and learn from. ## The risks Several people have suggested there is a risk that researchers mistakenly submit severity HIGH issues on GitHub thinking they are LOW or MEDIUM. It is basically unheard of that a researcher would have _understated_ a reported problem so I consider that a low risk scenario. Also: there was already before this a risk that someone would submit a vulnerability report in the public, it is not a new risk. Handling some vulnerabilities in the public of course makes the immediately available for both good and bad people. This policy will give bad people a slightly longer window of opportunity to exploit weaknesses at the expense of existing curl users. That is unfortunate but we see no way of avoiding this. We limit this practice to less than HIGH for this reason: at least these problems are not terrible. ## Severity LOW We have discussed simply dismissing all severity LOW issues, but within the CVE program there is an appeals process and if we don’t hand out a CVE to an actual vulnerability, then someone else will do it. We thus cannot ignore true vulnerabilities, or they will come back to bite us nonetheless. We _can_ however resist and refuse silliness and I personally aim to deny and decline more vulnerabilities harder on the basis that they do not actually hurt any users or existing applications. ## Crazy combos A clear trend these days is that many newly reported vulnerabilities are figured out by AI models that create crazy sequences or option combinations that do something unexpected, but the sequence is something that aren’t likely to ever be used in the real world, or the unexpected that happens is something that occurs _every time_ so applications doing this would spot the problem immediately. I.e problems that have almost no risk to actually trigger in the real world. We save no users by insisting on doing the CVE dance for those issues. We only get work for ourselves and a whole world that has to act on them. ## Flow control it There is an almost infinite number of combinations you can use libcurl’s many options, protocol mixes and flexible API to do things that might not be clearly documented and the AIs will continue to try them all and together with the humans driving these checks they are super eager to treat every minor discrepancy as a vulnerability. We can’t continue down this spiraling path with our sanity intact. We don’t have the bandwidth, manpower or even will to handle a never-ending vulnerability reporting stream at this pace. I believe few software projects can stand this kind of prolonged high volume deluge for extended periods. Something needs to give way. ## How to assign severity We only use these for levels in the curl project: LOW, MEDIUM, HIGH and CRITICAL. Often we refer back to recently managed vulnerabilities to see how we judged them, but we also provide a set of basic guidelines. I’m sure we can also make them better and clearer.

axleyjc reposted

Mike Masnick
@masnick.com

Hi. Some of us predicted this very thing might happen. www.techdirt.com/2025/05/27/w...

techdirt.comWhen Half The Supreme Court Has Book Deals With The Same Publisher, Who Decides Its Cases?When five Supreme Court Justices recuse themselves from a single case, that’s news. When they do it because most of them have book deals with the same publisher, that’s potentially a pr…
James Grimmelmann@jtlg.bsky.social

Because four of the Justices have book contracts with a party's corporate parent, the Supreme Court LACKS A QUORUM to hear BMG v. Vetter, and one of the worst copyright decisions in recent memory is affirmed by default. It's just one bad look after another. www.supremecourt.gov/docket/docke...

supremecourt.govDocket for 25-1391

axleyjc reposted

Don Moynihan
@donmoyn.bsky.social

Is he flying on the regular Air Force One or the one that Qatar gave to Trump?

FactPost@factpostnews.bsky.social

Vance: Our job is to make sure elites in DC don’t get richer by having access to the President of the United States. But I don’t think that’s a concern with Trump.

axleyjc
@fedi.jason.axley.net

#ai companies and #aidoomers are all like "AI could wipe out humanity" "We need a pause" or "slowdown" If only there was a way to make the AI stop before we all die! #stophittingyourself

axleyjc reposted

Mike Masnick
@masnick.com

I was going to say this sounds like a parody, but, really, SNL already did that parody 35 years ago. www.youtube.com/watch?v=Gmqe...

youtube.comHappy Fun Ball - SNLYouTube video by Saturday Night Live
Kotaku@kotaku.com

MrBeast Joins Shark Tank, Invests Quarter Million In AI-Powered 'Ball 2' https://kotaku.com/mrbeast-joins-shark-tank-invests-quarter-million-in-ai-powered-ball-2-2000747308

kotaku.comMrBeast Invests Quarter Mill On AI-Powered 'Ball 2' On Shark TankMr. Wonderful inversely appeared uninterested in the foam orb, not seeing how it could distinguish itself among other, more popular balls

axleyjc reposted

Dan Gillmor
@dangillmor.mastodon.social.ap.brid.gy

All corruption in Trump world crypto connections, Exhibit 65,020: https://www.citationneeded.news/issue-111/ Latest newsletter from Molly White, who covers this stuff better than anyone else.

citationneeded.newsIssue 111 – Back the winnersA Trump crypto adviser’s firm took millions from Tether while he shaped stablecoin rules in their favor, a Senate probe targets Commerce Secretary Lutnick’s not-so-former firm, and crypto’s biggest super PAC plays it safe

axleyjc reposted

Chuck Darwin
@cdarwin.c.im.ap.brid.gy

A 19-year-old walked through Helsinki airport in April 2026 carrying two 2TB hard drives and a ticket to Japan. He couldn’t make that flight. Finnish police stopped him on an Interpol Red Notice, and by July, US prosecutors had unsealed a federal complaint identifying him as Peter Stokes, an […]

c.imOriginal post on c.im

axleyjc reposted

A.R. Moxon
@juliusgoat.bsky.social

I would like to see the end of the term "vaccine skeptic," to be replaced by "reality denier." These so-called skeptics are actually the most gullible rubes on the planet when it comes to conspiracy nonsense and tech-bro slop; the only thing that makes them skeptical is peer-reviewed evidence.

axleyjc reposted

Nicholas Grossman
@nicholasgrossman.bsky.social

DHS raid on Rep. Omar is a great example of very bad by rule-of-law democracy standards, pretty good by consolidated authoritarianism standards. Deeply wrong, with lies, bigotry, harassment, abuse of power. AND a sign of weakness, limited capacity, failure to make the dictatorship of their dreams.

Bill Grueskin@bgrueskin.bsky.social

Via Fox News, natch: A criminal raid, right before the election, based on some paperwork Omar’s dad filled out decades ago www.foxnews.com/politics/ilh...

axleyjc reposted

[object Object]
@zzt.mas.to.ap.brid.gy

Replying to @zzt.mas.to.ap.brid.gy

you know what’s fucking incredible about growing up in neoliberalism some fucker can abuse you and make you feel unsafe for years and you’re expected to take the high road or you’re the bad person once that fucker gets rich, because capitalism rewards abuse with money, it’s then considered too […]

mas.toOriginal post on mas.to

axleyjc reposted

jonny (nonvenomous)
@jonny.neuromatch.social.ap.brid.gy

Meta's "Wonder" model that they just released today is a this tier computer use model

jonny (nonvenomous)@jonny.neuromatch.social.ap.brid.gy

RE: https://neuromatch.social/@jonny/117339825958098508 OK! Meta evaluated this as intended behavior, not applicable for a bug bounty, so therefore responsible disclosure no longer applies so here goes: any process run within the VM can access the socket that provides inference with no […]

neuromatch.socialOriginal post on neuromatch.social
axleyjc
@fedi.jason.axley.net

That was the dumbest decision to run a play on the 1 yard line with 13 seconds left in the half #seahawks. That cost you the game. Just take a knee!

axleyjc reposted

Nonilex
@nonilex.masto.ai.ap.brid.gy

Jeepers! Nothing to see here! #AI Went Rogue, Again #Anthropic Al agents, acting on their own, tried to fill out #visa applications & sent the #police a fake #homicide tip — WTAF? #ArtificialIntelligence #tech #security #law #regulations seriously, anything, this is bonkers […]

masto.aiOriginal post on masto.ai