internationalcolorconsortium/iccdev/specsep-qa

Claude Code and agent skill by internationalcolorconsortium/iccdev

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Build, run, or diagnose the repository-owned iccSpecSepToTiff QA suites and their checked-in spectral TIFF fixtures.

Install

npx vskill@latest install internationalcolorconsortium/iccdev/specsep-qa

SKILL.md

iccSpecSepToTiff QA

Use this skill for iccSpecSepToTiff CLI contract, TIFF geometry, profile alignment, output cleanup, or fixture work.

Workflow

  1. Read ../../../Tools/CmdLine/IccSpecSepToTiff/Readme.md for the command

contract and limits.

  1. Build iccSpecSepToTiff, iccTiffDump, and iccFromXml in the same tree.

For tile coverage, also build iccApplyProfiles and iccTiffTilesTest.

  1. Run the repository wrapper from any directory:
bash
ICCDEV_TOOLS_DIR=/path/to/iccDEV/Build/Tools \
/path/to/iccDEV/.github/scripts/iccdev-specsep-qa.sh
  1. For a focused failure, rerun the named child script shown by the wrapper.
  2. Keep durable fixtures under .github/ci/test-data/; keep generated TIFFs

and logs under ICCDEV_TEST_OUTDIR.

For a bounded option/output campaign with exact pixel and TIFF-tag validation, run:

bash
ICCDEV_TOOLS_DIR=/path/to/iccDEV/Build/Tools \
/path/to/iccDEV/.github/scripts/iccdev-specsep-campaign.sh --seconds 300

The numbered-fixture child suite runs every checked-in valid TIFF as a single-channel input with all four compression and planar combinations.

The profile sweep runs twice, because a profile is accepted only when its sample count matches the generated TIFF: once at the default eight channels across Testing/CalcTest, which can only produce rejections, and once at --channels 3 across Testing/, whose single depth-1 profile is the 3-channel sRGB v4 preference. A run that accepts nothing says so, so a rejection-only corpus cannot look like full coverage. For a local registry download, rerun with --profile-dir /path/to/profiles and a --channels value the corpus can match; accepted profiles must be embedded, while profile-specific clean rejections are expected.

The extended campaign requires Python numpy, tifffile, and imagecodecs -- tifffile defers LZW decoding to imagecodecs and half the matrix writes LZW. It preserves a machine-readable text summary, representative iccTiffDump/tiffinfo inspections, generated inputs, and output TIFFs under ICCDEV_TEST_OUTDIR. It exits nonzero for either a functional failure or a specification-audit finding; review both categories in matrix-summary.txt.

Fixture Rules

  • spectral/spec_1 through spec_10 are the tiny normal conversion sequence.
  • specsep-harvest/gray300/spec_1 through spec_8 cover larger grayscale

input.

  • specsep-truncated/spec_1 is the late-read-failure cleanup fixture.
  • Do not replace these with local scratch paths or generated build artifacts.

Validation

Run the wrapper, then build the regression helpers before a direct filtered CTest run. Helpers such as iccTiffTilesTest are intentionally excluded from the default product build; check builds them automatically, but ctest does not:

bash
cmake --build Build --target build-test-binaries
ctest --test-dir Build -R '^iccdev\.specsep-' --output-on-failure --no-tests=error

For tiled input/output, BigTIFF, IFD inspection/extraction, spectral sample preservation, and scalar/threaded image parity, run on Unix or Windows:

bash
ctest --test-dir Build -C Release -R '^iccdev\.tiff-(tiles|tile-budget-cli|directory-listing-cli)' --output-on-failure --no-tests=error

The iccdev.tiff-tile-budget-cli test checks 16 mixed tiled/stripped channels for pixel parity and verifies that the shared 256 MiB input tile-band limit rejects before creating output. The output tile band has its own cap; do not describe either as a process-wide memory limit.

The iccdev.tiff-directory-listing-cli test checks plain and JSON inventory, including raw photometric metadata, of 4096 main IFDs and clean rejection of 4097 main IFDs. Compare elapsed times separately when investigating listing performance; avoid a timing gate.

For multi-configuration generators such as Visual Studio, set -C to the configuration that was built (Release above, or Debug).

The native helper generates fixtures from libtiff independently of CTiffImg; the CLI runner needs only Python's standard library. Do not claim SVS decoding, ROI or pyramid rewriting from these generic TIFF tests.

Verify changed text files are ASCII and run git diff --check.

References

  • ../../prompts/specsep-qa.prompt.md
  • ../../../docs/ctest.md
  • ../../instructions/testing.instructions.md

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
5

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/internationalcolorconsortium/iccdev/specsep-qa/badge)](https://verified-skill.com/skills/internationalcolorconsortium/iccdev/specsep-qa)

Something wrong with this skill or its page? Tell us.