Publish and review static technical documentation¶
The build problem¶
Publish searchable static technical documentation and review its prose without making an editorial model authoritative.
Reviewed answer
Recommended path: Use vale-cli/vale; Try zensical/zensical; Learn from addyosmani/agent-skills; Study shubhamsaboo/awesome-llm-apps.
You still own: Human review must still decide whether claims are true, concise, useful, and supported by repository evidence. Simulated readers do not replace measured reader behavior.
Option-by-option decision¶
Try
zensical/zensical
Static site generator for modern, searchable project documentation.
What it contributes: Builds responsive Markdown pages, tags, and private client-side search without a custom frontend.
Watch: The pinned pre-1.0 version remains removable and must keep proving deterministic, accessible, no-runtime-network output.
Revisit when: Strict build, search, accessibility, or no-third-party-request validation fails.
Checked 2026-09-17 · Repository · Evidence
Use
vale-cli/vale
Markup-aware prose linter with project-owned styles and offline execution.
What it contributes: Runs repeatable offline prose checks across hand-authored public Markdown.
Watch: It can flag deterministic patterns but cannot establish factual truth, reader usefulness, or editorial judgment.
Revisit when: The pinned version cannot lint selected sources deterministically or project rules create repeated false positives.
Checked 2026-09-17 · Repository · Evidence
Study
shubhamsaboo/awesome-llm-apps
100+ AI Agents, Agent Skills and RAG Apps - Free and Open Source.
What it contributes: Supplies the pinned First Reader method used for context-free review of first-contact public surfaces.
Watch: The parent repository is a learning map, not a production dependency or blanket endorsement. Unknown: Simulated reader responses are hypotheses rather than measured user behavior.
Revisit when: Reader reviews fail to find the primary path or require sending unpublished prose to a hosted service.
Checked 2026-09-17 · Repository · Evidence
Learn from
addyosmani/agent-skills
Open development guidance and reusable skills for agent-assisted engineering.
What it contributes: Provides pinned selective-context and source-driven-development guidance for the editorial workflow.
Watch: The skills are reviewed guidance only; they are neither runtime dependencies nor factual authorities.
Revisit when: A later pinned revision is fully reviewed and materially improves the workflow.
Checked 2026-09-17 · Repository · Evidence
What the existing tools already cover¶
- Static rendering, private search, deterministic prose linting, and bounded review methods already exist.
- Editorial aids can remain separate from factual source and approval authority.
What you still need to decide or build¶
- Human review must still decide whether claims are true, concise, useful, and supported by repository evidence.
- Simulated readers do not replace measured reader behavior.
That remaining work is the justified custom scope in this decision. The actions above apply only to the stated problem; they are not universal rankings or guarantees beyond the cited evidence.
Checked 2026-09-17 · See all decisions