Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
53 commits
Select commit Hold shift + click to select a range
d375ecc
Security audits M1: add the off-by-default capability and scan schedu…
Brooooooklyn Nov 18, 2026
647228d
Security audits spikes S1/S1b: container egress lock, cold start, ima…
Brooooooklyn Nov 18, 2026
063308a
Security audits M2: add the shared model list and policy parser
Brooooooklyn Nov 18, 2026
29a949f
Security audits M3: add the security tables and migration 0065
Brooooooklyn Nov 18, 2026
14e4d41
Security audits M4: add the operator security policy and status routes
Brooooooklyn Nov 18, 2026
826974f
Security audits M5: add the scanner runner package with Engine B
Brooooooklyn Nov 18, 2026
7516f90
Security audits spikes S2/S3: Agent SDK and Codex SDK in a locked con…
Brooooooklyn Nov 18, 2026
99fc9a1
Security audits M6: add the scanner controller, provisioning and enti…
Brooooooklyn Nov 18, 2026
8f71f16
Security audits M7: add the model proxy and scan budget meter
Brooooooklyn Nov 18, 2026
2199a57
Security audits M8: write the scan row at go-live and hold staged input
Brooooooklyn Nov 18, 2026
d73230b
Security audits M9: run the scan queue and the hourly security jobs
Brooooooklyn Nov 18, 2026
e633e99
Security audits spikes S4/S8: Engine B quality, cost and time on @cf
Brooooooklyn Nov 18, 2026
f80a201
Security audits M10: build the scan input from the deployed bundle
Brooooooklyn Nov 18, 2026
99c0abf
Security audits M11: scan the GitHub source diff of hosted deploys
Brooooooklyn Nov 18, 2026
2af3394
Security audits spikes S2/S3 live + S10: model calls through AI Gateway
Brooooooklyn Nov 18, 2026
e25e210
Security audits M12: store scan results and findings at ingest
Brooooooklyn Nov 18, 2026
faf0182
Security audits: keep findings when one check session fails; widen sc…
Brooooooklyn Nov 19, 2026
990a4f1
Security audits M13: add Engines A and C, the MCP tool server and the…
Brooooooklyn Nov 19, 2026
f331699
Security audits: record the S4 rerun on glm-5.3 and deepseek
Brooooooklyn Nov 19, 2026
f752e28
Security audits: make deepseek-v4.1-flash-1017 the default scan model
Brooooooklyn Nov 19, 2026
6fa254e
Security audits M14: project security API routes
Brooooooklyn Nov 19, 2026
0af0eed
Security audits M15: email notifications for findings, caps, public g…
Brooooooklyn Nov 19, 2026
a32086e
Security audits M16: scan before public (public gate, bypass, exposur…
Brooooooklyn Nov 19, 2026
e45b72f
Security audits M17: operator security API
Brooooooklyn Nov 19, 2026
538dd8d
Security audits M18: admin panel Security pages and project controls
Brooooooklyn Nov 19, 2026
026d54f
Security audits M16: add the missing gate race and admission tests
Brooooooklyn Nov 19, 2026
3a1145b
Security audits M19: read-only project security page
Brooooooklyn Nov 19, 2026
1aecbf6
Security audits M20: void project security CLI commands
Brooooooklyn Nov 19, 2026
3847b29
Security audits M21: zero-trust bypass CLI command and held-request s…
Brooooooklyn Nov 19, 2026
8949a6c
Security audits M22: deploy security notice and project status line
Brooooooklyn Nov 19, 2026
3a092fa
Security audits M23: operator security CLI, installer flag, scanner u…
Brooooooklyn Nov 19, 2026
9864af4
Security audits M24: user docs for scans, caps and the public gate
Brooooooklyn Nov 19, 2026
07b2db1
Security audits spikes S7/S9/S11: inputs, refusal mapping, red team
Brooooooklyn Nov 19, 2026
baae790
Security audits M13b: Engine A/C images, controller dispatch, recorde…
Brooooooklyn Nov 19, 2026
97f91d5
Security audits spikes S13/S5: runtime copy and static confirmations
Brooooooklyn Nov 19, 2026
72666dd
Security audits M25: project rescan, admin backfill and scheduled sweeps
Brooooooklyn Nov 19, 2026
bb053ee
Security audits M26: fallback scans, failure-streak emails, digest an…
Brooooooklyn Nov 19, 2026
b0f92cf
Security audits M27: per-plan scan caps
Brooooooklyn Nov 19, 2026
37f52dd
Security audits M28: Void Security Audit GitHub check and PR comment
Brooooooklyn Nov 19, 2026
6d482c0
Security audits M29: page and dashboard forms for findings, model, bu…
Brooooooklyn Nov 19, 2026
660ac64
Security audits M30: runtime checks on a throwaway copy of the deploy…
Brooooooklyn Nov 20, 2026
02d552f
Security audits M31: secret findings, client asset pass and SARIF export
Brooooooklyn Nov 20, 2026
f16ef2c
Security audits: fix refusal mapping and the Glob path escape found b…
Brooooooklyn Nov 20, 2026
2770528
Security audits: close refusal and path-hook gaps found in review
Brooooooklyn Nov 20, 2026
bd89485
Security audits M32: local check of the working tree before deploy
Brooooooklyn Nov 20, 2026
e506bc8
Security audits M33: own OpenAI and Anthropic keys for scans
Brooooooklyn Nov 20, 2026
7b6d4c8
Security audits M34: per-project scan rule (inherit, always, never)
Brooooooklyn Nov 20, 2026
d4b8ffb
Security audits M35: let admins change a finding's severity
Brooooooklyn Nov 20, 2026
ad4164c
Security audits M36: correct admin docs and the scanner invariants table
Brooooooklyn Nov 20, 2026
80598d4
Security audits: close gaps from the completeness review
Brooooooklyn Nov 20, 2026
7b698e8
Merge remote-tracking branch 'origin/main' into feat/security-audits
Brooooooklyn Nov 20, 2026
2a0b6d5
test(cli): strip colour before reading the Zero Trust status box
Brooooooklyn Nov 20, 2026
de390f6
Security audits: give scan memory the model's fingerprint, not the st…
Brooooooklyn Nov 20, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
The table of contents is too big for display.
Diff view
Diff view
  •  
  •  
  •  
31 changes: 31 additions & 0 deletions .github/workflows/platform-ci.yml
Original file line number Diff line number Diff line change
Expand Up @@ -38,6 +38,37 @@ jobs:
- name: Build API build container
run: docker build --pull --tag void-api-build-container:ci --file platform/packages/api/container/Dockerfile .

build-security-scanner-images:
if: github.event_name == 'pull_request'
runs-on: ${{ github.repository == 'voidzero-dev/void' && 'blacksmith-4vcpu-ubuntu-2404' || 'ubuntu-latest' }}
timeout-minutes: 20
permissions:
contents: read
steps:
- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
with:
persist-credentials: false
- uses: voidzero-dev/setup-vp@1b32467adbe183473499fd9d5d372c3ed9641754 # v1.18.0
with:
cache: true
node-version-file: .node-version
- run: vp install
- run: vp run --filter ./platform/packages/security-scanner build
- name: Build Engine A scanner image
run: docker build --platform linux/amd64 --tag void-scanner-claude:ci --file platform/packages/security-scanner/image/Dockerfile.claude .
- name: Build Engine C scanner image
run: docker build --platform linux/amd64 --tag void-scanner-codex:ci --file platform/packages/security-scanner/image/Dockerfile.codex .
- name: Check the Engine C image pins its Codex version and keeps only our models
run: |
docker run --rm --entrypoint /opt/void-scanner/codex-bin void-scanner-codex:ci --version
docker run --rm --entrypoint node void-scanner-codex:ci -e "const m=require('/opt/void-scanner/codex-models.json').models.map(x=>x.slug);console.log(m.join());if(m.length!==2)process.exit(1)"
# Both images run a whole scan with no network against a scripted fake model (design
# doc 0095 M13b). The builds above leave the layers cached for this run.
- name: Run a recorded scan in both scanner images
env:
VOID_SCANNER_E2E: '1'
run: vp test run platform/packages/security-scanner/test/e2e/recorded-run.e2e.test.ts

lint-build-check:
runs-on: ${{ github.repository == 'voidzero-dev/void' && 'blacksmith-4vcpu-ubuntu-2404' || 'ubuntu-latest' }}
permissions:
Expand Down
50 changes: 50 additions & 0 deletions .github/workflows/platform-security-scanner-images.yml
Original file line number Diff line number Diff line change
@@ -0,0 +1,50 @@
name: Publish Security Scanner Images

# Manual release step for security audits Engines A and C (design doc 0095 M13b): build
# both scanner images for linux/amd64, push them to ghcr.io, and print the digests to pin
# in platform/packages/api/src/security/scanner-images.json. Commit that file before the
# platform release that should turn those engines on. The ghcr.io packages must be public:
# each install copies the image by digest without credentials. Runs only on main or a
# release tag, with no build cache, and labels each image with the commit it was built
# from (org.opencontainers.image.revision), so a pinned digest can be traced to reviewed code.
on:
workflow_dispatch:

concurrency:
group: platform-security-scanner-images
cancel-in-progress: false

permissions: {}

jobs:
publish:
if: github.ref == 'refs/heads/main' || startsWith(github.ref, 'refs/tags/')
runs-on: ubuntu-latest
timeout-minutes: 30
permissions:
contents: read
packages: write
steps:
- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
with:
persist-credentials: false
- uses: voidzero-dev/setup-vp@1b32467adbe183473499fd9d5d372c3ed9641754 # v1.18.0
with:
cache: false
node-version-file: .node-version
- run: vp install
- name: Log in to ghcr.io
env:
GHCR_TOKEN: ${{ secrets.GITHUB_TOKEN }}
run: printf '%s' "$GHCR_TOKEN" | docker login ghcr.io --username "$GITHUB_ACTOR" --password-stdin
- name: Build, push and pin both images
run: node platform/scripts/publish-security-scanner-images.ts --push
- name: Show the digests to commit
run: |
{
echo '### platform/packages/api/src/security/scanner-images.json'
echo "Built from \`$GITHUB_SHA\` ($GITHUB_REF)."
echo '```json'
cat platform/packages/api/src/security/scanner-images.json
echo '```'
} >> "$GITHUB_STEP_SUMMARY"
2 changes: 2 additions & 0 deletions docs/.vitepress/config.ts
Original file line number Diff line number Diff line change
Expand Up @@ -109,6 +109,7 @@ const referenceItems: Array<SidebarItem> = [
{ text: 'Platform Configuration', link: '/reference/cli/platform-config' },
{ text: 'Platform Users and Projects', link: '/reference/cli/platform-users' },
{ text: 'Project Zero Trust', link: '/reference/cli/platform-zero-trust' },
{ text: 'Platform Security Audits', link: '/reference/cli/platform-security' },
{ text: 'Platform Operations', link: '/reference/cli/platform-operations' },
{ text: 'Platform Email', link: '/reference/cli/platform-email' },
],
Expand Down Expand Up @@ -195,6 +196,7 @@ const platformSidebar = [
{ text: 'Users and Projects', link: '/guide/platform/administration/projects' },
{ text: 'Plans and Limits', link: '/guide/platform/administration/plans' },
{ text: 'Project Zero Trust', link: '/guide/platform/administration/zero-trust' },
{ text: 'Security Audits', link: '/guide/platform/administration/security-audits' },
{ text: 'Email', link: '/guide/platform/administration/email' },
{ text: 'Operations', link: '/guide/platform/administration/operations' },
],
Expand Down
1 change: 1 addition & 0 deletions docs/guide/platform-administration.md
Original file line number Diff line number Diff line change
Expand Up @@ -15,5 +15,6 @@ The operator commands for users, projects, deployments, and system status requir
- [Users and Projects](/guide/platform/administration/projects)
- [Plans and Limits](/guide/platform/administration/plans)
- [Project Zero Trust](/guide/platform/administration/zero-trust)
- [Security Audits](/guide/platform/administration/security-audits)
- [Email](/guide/platform/administration/email)
- [Operations](/guide/platform/administration/operations)
13 changes: 13 additions & 0 deletions docs/guide/platform/administration/plans.md
Original file line number Diff line number Diff line change
Expand Up @@ -122,6 +122,19 @@ Moving an account to another plan, replacing a removed plan, or transferring a p

Inspect a plan's default with `void platform config plans show <plan-id>` or the dashboard's user detail page. If the CLI reports that your platform does not support plan defaults, run `void platform upgrade` first.

## Security Scan Cap {#security-scan-cap}

With [security audits](./security-audits.md) on, a plan can set the most each of its projects may spend on scans in a calendar month (UTC). In the dashboard, edit the plan's **Security scan cap**: **Policy default**, **Monthly amount**, or **No cap**. From the CLI:

```sh
void platform config plans set team --security-cap 200 --plan
void platform config plans set team --security-cap 200 --yes
```

`--security-cap` takes a USD amount, `none` for no cap, or `inherit` for the default project cap. Plan files use the `securityCap` field with the same values. A plan amount applies even while project caps are off, and `none` lifts the default project cap for the plan's projects. A limit you set for one project comes first. The platform cap still stops every scan once it is used up.

A change applies right away: to the next scan that starts, and to scans already running from their next model call, so a lower cap can stop one. Copying a plan copies its scan cap, and accounts moved to a replacement plan get that plan's cap. See [Spending caps](./security-audits.md#spending-caps).

## Archive or Remove a Plan

Archive a plan to stop assigning it to new accounts while preserving its current accounts:
Expand Down
Loading