Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
77 changes: 77 additions & 0 deletions docs/fork/security-sync.md
Original file line number Diff line number Diff line change
Expand Up @@ -1529,3 +1529,80 @@ required modules, neither called and neither raised by Dependabot.
| `yarn npm audit --severity moderate`, the CI gate | no suggestions | no suggestions |
| `yarn npm audit --severity moderate --recursive` | braces, follow-redirects x2, smol-toml, sprintf-js | braces, smol-toml, sprintf-js |
| osv-scanner on `yarn.lock` | 2 high, both follow-redirects | no issues |

## 2026-10-24 - next 16.4.2, storybook 10.6.2, ip-address 10.7.4

Seven alerts. next is pinned in `security/overrides.yaml`. storybook and
ip-address move by lockfile inside their declared ranges, and the storybook move
takes @vitest/mocker out of the tree. Nothing is dismissed.

### next - pinned at ^16.4.1, resolving 16.4.2

| Alert | Advisory | Severity | Needs | Here |
| ----- | ------------------- | -------- | --------------------------------------------------------- | --------------------------------- |
| 363 | GHSA-2464-fhr4-w9xw | critical | sharp below 0.35.5 under the image optimiser, on glibc | optimiser on, glibc, sharp 0.35.5 |
| 366 | GHSA-673q-mjm6-r5xr | critical | an App Router `use cache` function returning typed arrays | Pages Router, no `use cache` |
| 362 | GHSA-p3m5-59mp-jv5r | high | an App Router Server Action | Pages Router, no Server Action |

363 has a sink in the image. The root image is trixie-slim, which is glibc, and
`images.unoptimized` sits only in the static-export branch of `next.config.mjs`
(`:180`), so `/_next/image` loads sharp. The lockfile has resolved sharp 0.35.5,
the fix, since the 2026-10-09 re-resolve, and 16.4.1 raises next's own floor to
`sharp ^0.35.5`. `packages/app` has no `app/` directory and no `use cache` or
`use server`, so 366 and 362 reach nothing.

Pinned rather than re-resolved because the merge driver takes upstream's
lockfile, which is how 2.40.0 took next 16.3.6 back to 16.3.4. Upstream has the
bump open as hyperdxio/hyperdx#3460. `npmMinimalAgeGate` holds 16.4.3
(2026-10-23), so the range takes 16.4.2 (2026-10-15). 16.4.2 still vendors the
pre-fix image-size, so the 2026-09-27 verdict stands.

### storybook 10.2.13 -> 10.6.2, and @vitest/mocker gone

| Alert | Advisory | Severity | Needs |
| ----- | ------------------- | -------- | ---------------------------------------------------------- |
| 360 | GHSA-f2ph-cmg5-wxc8 | critical | a malicious website open while `storybook dev` runs |
| 365 | GHSA-px85-f4q5-jh77 | high | a host on the developer's network reaching `storybook dev` |
| 361 | GHSA-qxg6-89mg-rqv8 | moderate | a client on a Vite dev server running `interceptorPlugin` |

Dev only. storybook is an app devDependency (`packages/app/package.json:151`),
@vitest/mocker 3.2.4 came only from `@storybook/builder-webpack5` 10.1.4's exact
pin, and none of the family is in the standalone trace. Fixed anyway, since 360
and 365 target the developer running `yarn storybook` and every range admits the
fix.

```sh
yarn up -R storybook @storybook/nextjs @storybook/addon-docs @storybook/addon-links @storybook/addon-themes
```

The family moves together because `@storybook/nextjs` pins its builder exactly,
and builder-webpack5 10.6.2 drops @vitest/mocker. A newer babel preset and
probe-image-size make the lockfile diff about 2,900 lines, every added version
reached only through the new storybook packages. `storybook build` completes on
10.6.2. `osv-scanner.toml` loses GHSA-82fw-gwwq-j7x9, which the scanner no
longer reports.

### ip-address 10.7.3 -> 10.7.4

| Alert | Advisory | Severity | Needs |
| ----- | ------------------- | -------- | ----------------------------------------------------------------- |
| 364 | GHSA-rxp7-6hxx-x8q5 | moderate | many `%` then a line terminator, in a string handed to `Address6` |

In the image via `@hyperdx/api`, express-rate-limit 8.7.1 and socks 2.8.10. Both
of our callers hand `isPrivateIp` a `new URL()` hostname
(`utils/validators.ts:161`, `routers/api/clickhouseProxy.ts:105`), and Node's
URL parser refuses a host holding `%` or a line feed. Moderate, so below THE
BAR, and every range admits 10.7.4: `yarn up -R ip-address`. The verdict in
`scripts/dismiss-triaged-alerts.py` now names `^10.7.1` and 10.7.4, not
`^10.3.1`.

### The audit, before and after

| Run | main | This change |
| ------------------------------------------------- | ---------------------------- | --------------- |
| `yarn npm audit --severity moderate`, the CI gate | not run | no suggestions |
| `yarn npm audit --severity moderate --recursive` | not run | smol-toml 1.6.1 |
| osv-scanner on `yarn.lock` | 3 critical, 6 high, 2 medium | 4 high, all zod |

The four left are GHSA-mrhx-c7xc-m3w5 against zod, fixed only in 4.6.6. Not
touched here, and Dependabot has raised no alert for it yet.
15 changes: 8 additions & 7 deletions docs/fork/what-we-changed.md
Original file line number Diff line number Diff line change
Expand Up @@ -619,14 +619,15 @@ header check is what holds when a deployment authenticates some other way.
touched the file four times, and the conflict is trivial either way: take
ours, then re-run `yarn ratchet:update` after the sync so the numbers match
the merged tree.
- `package.json` `resolutions` - two generated fork pins,
- `package.json` `resolutions` - three generated fork pins,
`systeminformation ^5.31.7`, raising upstream's own `^5.24.0`, and
`follow-redirects ^1.16.1`, which upstream does not pin. Both are generated
from `security/overrides.yaml` by `scripts/security-override.py --apply`, so
edit the register, never these lines. The vectors are in the register entries.
`--check` judges the pin against upstream's own lockfile at our merge base,
never ours: the pin makes our lockfile resolve above its own floor, so ours
would call every live pin redundant.
`follow-redirects ^1.16.1` and `next ^16.4.1`, which upstream does not pin.
All three are generated from `security/overrides.yaml` by
`scripts/security-override.py --apply`, so edit the register, never these
lines. The vectors are in the register entries. `--check` judges the pin
against upstream's own lockfile at our merge base, never ours: the pin makes
our lockfile resolve above its own floor, so ours would call every live pin
redundant.

## Docs

Expand Down
4 changes: 0 additions & 4 deletions osv-scanner.toml
Original file line number Diff line number Diff line change
Expand Up @@ -33,10 +33,6 @@ reason = "@opentelemetry/instrumentation-*: tolerable risk. Only mongoose has a
id = "GHSA-45rx-2jwx-cxfr"
reason = "@opentelemetry/propagator-jaeger: runs only when OTEL_PROPAGATORS names jaeger, which nothing here, in dfe-infra or in dfe-docker sets; the default is tracecontext and baggage."

[[IgnoredVulns]]
id = "GHSA-82fw-gwwq-j7x9"
reason = "@vitest/mocker: under @storybook/builder-webpack5, not in the image."

[[IgnoredVulns]]
id = "GHSA-2g4f-4pwh-qvx6"
reason = "ajv 6.12.6: schema-utils 3 under webpack build plugins, not in the image, which ships ajv 8.20.0."
Expand Down
3 changes: 2 additions & 1 deletion package.json
Original file line number Diff line number Diff line change
Expand Up @@ -97,7 +97,8 @@
"fast-xml-parser": "^4.5.6",
"systeminformation": "^5.31.7",
"@asyncapi/specs": "6.11.1",
"follow-redirects": "^1.16.1"
"follow-redirects": "^1.16.1",
"next": "^16.4.1"
},
"resolutionsNotes": {
"@stryker-mutator/core/minimatch": "Dev tooling, not a CVE override. Added when a blanket `brace-expansion` pin forced v2 (CJS) while minimatch v10 needs v5's named exports. That blanket pin is gone — it broke `nx affected` the same way — so brace-expansion now resolves per consumer and Stryker would most likely work on minimatch v10. The pin stays because no CI job runs `yarn dev:mutation`, so nothing would catch it if that guess is wrong. Drop it once someone has run the mutation suite on v10. See CONTRIBUTING.md#mutation-tests."
Expand Down
5 changes: 3 additions & 2 deletions scripts/dismiss-triaged-alerts.py
Original file line number Diff line number Diff line change
Expand Up @@ -288,8 +288,9 @@
# Already above the range on the copy that matters.
"ip-address": (
"not_used",
"packages/api/package.json declares ^10.3.1 and the lock resolves "
"10.3.1, above the <= 10.3.0 range. isPrivateIp uses that copy. "
"packages/api/package.json declares ^10.7.1 and the lock resolves "
"10.7.4, above the <= 10.3.0 and <= 10.7.3 ranges. isPrivateIp uses "
"that copy. "
"RE-CHECK EVERY SYNC - a merge dragging it below 10.3.1 makes this a "
"real unauthenticated SSRF.",
),
Expand Down
17 changes: 17 additions & 0 deletions security/overrides.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -76,6 +76,23 @@ overrides:
exact 1.16.0 pin, dev only, moves with it.
upstream: none
added: 2026-10-21
- package: next
range: '^16.4.1'
advisory: GHSA-2464-fhr4-w9xw
severity: critical
vector: >-
The image runs Next's standalone server with the image optimiser on, since
images.unoptimized is set only in the static-export branch of
packages/app/next.config.mjs:180. GHSA-2464-fhr4-w9xw is librsvg under the
sharp the optimiser loads. The lockfile already resolves sharp 0.35.5, the
fix, and 16.4.1 raises next's own sharp floor to ^0.35.5 so the fix no
longer rests on a re-resolve. GHSA-673q-mjm6-r5xr needs an App Router `use
cache` function and GHSA-p3m5-59mp-jv5r an App Router Server Action, and
packages/app is Pages Router with neither. Pinned rather than re-resolved
because the sync merge driver takes upstream's lockfile, which is how the
2.40.0 sync took next 16.3.6 back to 16.3.4.
upstream: https://github.com/hyperdxio/hyperdx/pull/3460
added: 2026-10-24
# Example of the shape (keep commented - an empty register is the goal state):
#
# - package: some-parser
Expand Down
Loading
Loading