Compatibility and Releases

Compatibility and Releases

The skills release and the WordPress plugin versions are separate. Updating a skill does not upgrade GenerateBlocks or install Pro.

Current documented release

GenerateBlocks Skills v2026.09.20 contains all four skills, standalone converter dependencies and matching .skill/.zip archives. Its eight downloads have a SHA256SUMS.txt manifest.

Use the latest release page for available downloads and the changelog for subsequent updates. The handbook originated with v2026.09.18 and its icon/pattern guidance was updated for v2026.09.20 on September 20, 2026.

Version evidence

Reference	What it establishes
Free 2.5.0-beta.1 + Pro 2.8.0-beta.1	Tested workflows on WordPress 7.1.1/PHP 8.4, including local examples and the About-page test
Free 2.4.1 + Pro 2.7.1	Earlier documented guidance and a separate historical live-site observation
Local older Pro folder reporting 2.7.0-rc.1	A package-header observation; it must not be relabeled 2.7.1
Tracked generateblocks/ source	Free 2.4.1 reference code; it is not the beta installed by the skills

Read the target site's actual versions and block registry. A folder name, unchanged schema or successful test on another site is not a compatibility guarantee. The skill packages do not include vendor beta/Pro plugins.

Icon and pattern fix release

Release v2026.09.20 corrects an invalid Text-with-icon example and adds native icon fixtures, pattern-source guidance and preflight checks. The suite has 22 tests. Updated downloads are attached to v2026.09.20; the older v2026.09.18 assets remain unchanged. See the fix report.

AI assistant compatibility

The skills are instructions and helpers, not model-specific fine-tunes. Results depend on whether an assistant can read the necessary files, execute approved helpers, inspect the target when needed and follow the task's scope. File upload support is also not equivalent to persistent skill installation.

This project does not publish a controlled success-rate matrix covering every named model. Choose a model and tool environment that can perform the actual task, then verify the output. See Installation for the released installer behavior and the Codex legacy-path caveat.

Efficiency claims

For a familiar responsive layout based on a supplied design, the measured reading path decreased from 11,446 to 2,019 words, or 82.4%. These are instruction-word counts, excluding conversation, tool schemas, source inspection and generated output. They are not measurements of total token bills, model latency or a guaranteed quality improvement.

The instruction report explains the test conditions. The About correction also records a failed initial approach: source appearance and valid markup were insufficient when styling and behavior lived outside native blocks.

Verification boundaries

The release's 15 tooling tests and archive checks passed. Recorded runtime checks covered selected native layouts, responsive behavior, dark mode, queries and interactive states. They do not constitute an exhaustive theme/plugin/role matrix or a complete accessibility audit. The later FAQ padding fix passed frontend and parser checks, but the editor canvas was unavailable for visual inspection during that follow-up.

---

Back to wiki home · Reference library