Skip to content

chore(toolbox): bump pinned tools to clear the drift gate - #96

Merged
garymike merged 1 commit into
mainfrom
chore/drift-bump-2026-10-30
Oct 30, 2026
Merged

garymike merged 1 commit into
mainfrom
chore/drift-bump-2026-10-30

Conversation

@garymike

Copy link
Copy Markdown
Owner

The weekly drift gate went red again four days after #91 cleared it. Four releases landed upstream in that window.

Tool From To How
snyk-agent-scan 0.6.9 0.6.10 apply-tool-bumps.py
semgrep 1.182.1 1.183.0 apply-tool-bumps.py
checkov 3.3.30 3.4.0 apply-tool-bumps.py
skillspector 2.13.0 2.13.2 by hand

scripts/check-tool-updates.sh --fail-on-drift now exits 0 on all twelve pins.

skillspector is a hand bump by design. Its ARG is SKILLSPECTOR_REF, a commit SHA rather than a version, and apply-tool-bumps.py refuses it on purpose: patch_lock would rewrite toolbox/tools.lock while patch_dockerfiles matched nothing, and the half-bump would be reported as applied. v2.13.2 resolves to abd82935b67f, and the lock and the ARG moved together.

Re-measured, because the drift warning for that pin says to

Built the skill-audit image from these pins and ran the full proof. Nothing moved:

  FIXTURE                  SCORE     SKILLSPECTOR
  gecko-demo               64/100    exit 1, BLOCKS
  gecko-hook-demo          76/100    exit 1, BLOCKS
  config-injection-demo    100/100   exit 1, BLOCKS
  sibling-config-demo      51/100    exit 1, BLOCKS
  memory-poisoning-demo    0/100     exit 0, does NOT block

PROOF-FIXTURE PASSED
16 assertions, 16 PASS, 0 FAIL

Every score is identical to v2.13.0, and semgrep 1.183.0 does not disturb the gate's own rule, including the four assertions added in #93.

Why the memory carrier still reads 0/100. v2.13.2 predates NVIDIA/SkillSpector#893. Checked against the commit graph rather than inferred from release dates: the merge commit is 1196 commits ahead of that tag, and the tag published a day before the merge. So the agent-memory write is still the one carrier this gate uniquely covers, gate-proof.sh assertion 4 holds, and the 0/100 figure in README.md, docs/threat-model.md, docs/references.md, the walkthrough, the threat profile, gate-proof.sh and the Dockerfiles stays accurate.

The release that first carries #893 is the one that moves that fixture to 27/100 and makes those nine references stale. That will be a docs pass, not just a pin bump.

One correction this bump forced

The PyJWT override's comment claimed semgrep pins pyjwt[crypto]~=2.13.0. As of semgrep 1.183.0 it requires <3,>=2.15.0, so this bump made the comment false. Corrected in both Dockerfiles that carry it.

The override itself is left in place deliberately. Its comment names the condition for dropping it and that condition is now met, but removing it un-pins pyjwt in a repo that pins everything on purpose. That trade deserves its own change rather than riding along inside a version bump.

Expect Image pins to go red on merge

This changes toolbox/, so the five pinned images will no longer match the repo's build inputs and the follow-up is the usual pin move once the build publishes. That is the new check doing exactly what it was rebuilt to do in #94, and the first time it will have fired for a real reason rather than because a digest churned.

The gate went red again four days after #91 cleared it; four releases landed upstream in
that window.

| Tool | From | To |
|---|---|---|
| snyk-agent-scan | 0.6.9 | 0.6.10 |
| semgrep | 1.182.1 | 1.183.0 |
| checkov | 3.3.30 | 3.4.0 |
| skillspector | 2.13.0 | 2.13.2 |

The first three went through scripts/apply-tool-bumps.py, so tools.lock and the matching
Dockerfile ARGs moved together. skillspector is a hand bump by design: its ARG is a commit
SHA rather than a version, and that script refuses it on purpose, because patch_lock would
rewrite the lock while patch_dockerfiles matched nothing and the half-bump would report as
applied. v2.13.2 = abd82935b67f. check-tool-updates.sh --fail-on-drift now exits 0 on all
twelve pins.

Re-measured every fixture at 2.13.2, which is what the drift warning for that pin asks for.
Nothing moved: gecko-demo 64/100, gecko-hook-demo 76/100, config-injection-demo 100/100,
sibling-config-demo 51/100, and the persistent-memory write still cleared at 0/100.
gate-proof.sh passes 16 of 16 against a freshly built image carrying 2.13.2 and semgrep
1.183.0.

v2.13.2 predates NVIDIA/SkillSpector#893, checked against the commit graph rather than
inferred from release dates: the merge commit is 1196 commits ahead of that tag. So the
agent-memory write is still the carrier this gate uniquely covers, and the 0/100 figure in
README.md, docs/ and the Dockerfiles stays accurate. The release that first carries #893 is
the one that moves that fixture to 27/100 and makes those references stale.

Also corrects the PyJWT override's comment in both Dockerfiles. It claimed semgrep pins
pyjwt[crypto]~=2.13.0; as of 1.183.0 semgrep requires <3,>=2.15.0, so this bump made the
comment false. The override itself is left alone: removing it un-pins pyjwt in a repo that
pins everything deliberately, and that trade deserves its own change.
@garymike
garymike merged commit 83a3172 into main Oct 30, 2026
9 checks passed
@garymike
garymike deleted the chore/drift-bump-2026-10-30 branch October 30, 2026 19:51
garymike added a commit that referenced this pull request Oct 30, 2026
…blished (#97)

The second half of #96's drift bump, so callers get snyk-agent-scan 0.6.10, semgrep 1.183.0,
checkov 3.4.0 and skillspector 2.13.2 rather than the previous toolchain.

Resolved from each image's `sha-83a3172ac5dd` tag rather than `latest`, because `latest` is
whichever push landed last and two concurrent build runs have already raced it once.
build-toolbox.yml still has no concurrency guard.

Every digest was verified before being written: its build-inputs label was read back from the
registry and compared with the fingerprint this repo computes now, and nothing would have been
written if any of the five disagreed. The five stamps also match what the Image pins run on
83a3172 reported as the repo's current inputs, so two independent sources agree.
scripts/check-image-pins.py now reports all five current.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant