Skip to content

Tool hub: detekt and security-code-scan not routed; devskim SARIF shape stub-only - #2330

Closed
remigiusz-antczak wants to merge 2 commits into
mainfrom
feat/hub-kotlin-dotnet-2
Closed

remigiusz-antczak wants to merge 2 commits into
mainfrom
feat/hub-kotlin-dotnet-2

Conversation

@remigiusz-antczak

Copy link
Copy Markdown
Collaborator

Documents two candidate tools as not routed in the tool hub, each with its evidence row:

  • detekt: network-free behavior is unverified, so it is not routed.
  • security-code-scan: LGPL-3.0, and it needs a full project build.

Also states that the devskim SARIF shape is tested only against a stub.

test-review-checks-hub-kd.sh asserts both rows exist in tool-hub.md, and fails if either is removed. It also requires valid JSON before the negated jq check.

Replaces #2329, which had review findings. Stacked on #2325. Merge #2325 first.

Tests: hub-kd 5/5, hub-c 13/13, review-checks-tool-hub 108/108.

🤖 Generated with Claude Code

remigiusz-antczak and others added 2 commits November 12, 2026 21:10
Each tool is NOT CHECKED with an install hint when missing, counts as wired
only once installed, and is not run on a crash, empty output, wrong-shape
output or a non-zero exit. Changed files only, no network.

- flawfinder: --sarif, batch (GPLv2+).
- cppcheck: --xml-version=2, XML read from stderr, first <location> is the
  primary one (GPLv3).
- devskim: analyze --source-code accepts a single file, run once per file
  with --file-format sarif (MIT). Its own source has no HTTP, update or
  telemetry code; its NuGet dependencies were not audited, so dependency
  egress is marked unverified in tool-hub.md.

Flags, shapes and licenses were verified against man pages, manuals and
cloned source fetched 2026-11-12 (docs/standards-index.md). Fixtures use the
cppcheck manual sample verbatim; the flawfinder SARIF follows the tool's
source shape; the devskim SARIF is stub-only. Not routed: none.

size-budget-raise: .claude/skills/deep-code-review/references/tool-hub.md 43180→46217 three hub rows with their verification and egress evidence

Tests: scripts/test-review-checks-hub-c.sh 13/13. Affected ci-tests suites
otherwise pass except pre-existing failures that fail identically on clean
origin/main 49a330c: the 3 scan-summary suites (host semgrep cache,
"FAIL default excludes") and the size-gate test
test_index_has_a_size_budget_and_the_gate_fires_when_it_is_exceeded
(low-context-index, train-budget).

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…pe stub-only

detekt (Apache-2.0, OSI-listed): the CLI docs and Main.kt/CliArgs.kt show no
network code, but the analysis core was not read, so no network-free claim
holds; it is also a code-quality linter. security-code-scan (LGPL-3.0) is a
Roslyn analyzer that runs inside a project build with a NuGet restore, so it
cannot scan changed files only. Both get not-routed rows with evidence
fetched 2026-11-12 (docs/standards-index.md); devskim stays the routed .NET
scanner. The devskim row now says its SARIF shape is stub-tested only.

Test: scripts/test-review-checks-hub-kd.sh also greps both not-routed rows in tool-hub.md, asserts out is valid JSON before the negated jq, and has stubs of both tools that would
emit findings and asserts neither runs, with a devskim control that does.

size-budget-raise: .claude/skills/deep-code-review/references/tool-hub.md 46217→47544 two not-routed rows with their evidence and the devskim wording fix

Tests: scripts/test-review-checks-hub-kd.sh 5/5, hub-c 13/13, review-checks-tool-hub 108/108 (review-skips-index and review-load-bytes exit 0). Not run here: the size-gate test and scan-summary suites that fail identically on clean main per the base commit.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
@remigiusz-antczak

Copy link
Copy Markdown
Collaborator Author

Superseded by #2344.

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