Skip to content

Describe the tool as software composition analysis, not supply chain analysis - #8

Merged
richardmhope merged 1 commit into
mainfrom
claude/supply-chain-docs-update-hp4ybv
Jul 28, 2026
Merged

Describe the tool as software composition analysis, not supply chain analysis#8
richardmhope merged 1 commit into
mainfrom
claude/supply-chain-docs-update-hp4ybv

Conversation

@richardmhope

Copy link
Copy Markdown
Collaborator

"Supply chain analysis" oversold what this does. It inventories declared third-party components and checks them against OSV — software composition analysis. It does not look at build provenance, signatures, or what a downloaded artefact actually contains, and the docs should not suggest otherwise.

Terminology

Renamed the category everywhere it describes the tool:

  • README.md intro
  • website hero eyebrow and site_description in zensical.toml
  • pyproject.toml description, plus software-composition-analysis and dependency-scanning keywords
  • the package docstring and icebergsca --help
  • CLAUDE.md
  • the bundled agent skill — both its frontmatter description (which is what agents select on) and its body

Left "supply chain" where it names a genuine risk rather than the product category: not shelling out to mvn, hand-writing ranges/SARIF/CycloneDX instead of taking dependencies, the CI comments about this repo's own pipeline, and the SARIF supply-chain rule tag. Those statements are still accurate.

Wordmark

icebergsca-horizontal.svg now reads "Software composition analysis". Letter-spacing drops from 2.5 to 1.8 to fit the longer string. Measured in headless Chromium: the subtitle is 279.1px wide against the wordmark's 279.1px, so it sits flush with the text above rather than overflowing the 660px viewBox.

Scope note

Since the rename narrows the claim, the boundary is now stated explicitly in the README and How it works rather than left implied — it identifies a package by name and version, so provenance, signatures and typosquats are outside what a report can speak to.

Tone

Dropped the "Find. Resolve. Check. Report." slogan. Retitled sections to say what they contain:

Before After
The one rule What an empty findings list means
Built to be scanned as well as to scan The tool's own dependencies
Real dependency graphs Transitive dependency graphs
Findings never fail your build Findings never change the exit code
The part that matters Before reporting a project clean
Caveats worth knowing Caveats

Trimmed flourishes like "matters more than it looks", "gets || true appended within a week" and "and — most importantly —". The CI workflow examples are renamed from name: Supply chain to name: Dependency scan. No technical content was removed.

Verification

ruff check, ruff format --check, mypy and pytest (384 passed) all pass. The Zensical site builds clean (No issues found) and the rendered index.html carries the new wording. Checked that nothing links to the renamed heading anchors.

Two caveats worth flagging: the dogfood icebergsca scan . could not reach OSV from the sandbox (api.osv.dev returns 403 through the agent proxy), so that path is unverified — though the tool responded correctly by reporting "No packages could be checked — this is not a clean result" rather than a false clean. And there is no CHANGELOG entry, since everything under [Unreleased] describes the unshipped 0.1.0 and wording changes to it would be noise; happy to add one if you'd prefer.


Generated by Claude Code

…analysis

"Supply chain analysis" oversold what this does. It inventories declared
third-party components and checks them against OSV — software composition
analysis. It does not look at build provenance, signatures, or what a
downloaded artefact actually contains, and the docs should not suggest
otherwise.

Renamed the category everywhere it describes the tool: README, website hero
and site description, package description and CLI help, the bundled agent
skill, and the horizontal wordmark (letter-spacing dropped to 1.8 so the
longer subtitle sits flush with the wordmark above it). Left "supply chain"
where it names a genuine risk — shelling out to `mvn`, taking on
dependencies of our own, our own CI — since those statements are still true.

Added an explicit scope note to the README and How it works so the boundary
is stated rather than implied.

Also toned down the docs: dropped the "Find. Resolve. Check. Report."
slogan, retitled "The one rule" to "What an empty findings list means",
"Built to be scanned as well as to scan" to "The tool's own dependencies",
"Real dependency graphs" to "Transitive dependency graphs", and trimmed the
rhetorical flourishes ("matters more than it looks", "within a week", "worth
knowing") without touching any technical content.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CgjzvMcy8B8yuir8AZBgAW
@richardmhope
richardmhope merged commit 4b88ceb into main Jul 28, 2026
8 checks passed
@richardmhope
richardmhope deleted the claude/supply-chain-docs-update-hp4ybv branch July 28, 2026 10:35
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.

2 participants