Markdown parser, done right. Commonmark support, extensions, syntax plugins, high speed - all in one. Gulp and metalsmith plugins available. Used by Facebook, Docusaurus and many others! Use https://github.com/breakdance/breakdance for HTML-to-markdown conversion. Use https://github.com/jonschlinkert/markdown-toc to generate a table of contents.
| Rating | Summary | |
|---|---|---|
| Quality | solid (18/24) | Exceptionally documented and widely adopted library, held back only by a 2019 final release and 116 open issues. |
| PAI Relevance | integrate (0.63) | PAI uses markdown+YAML pervasively but carries no dedicated programmatic Markdown parser; remarkable fills that gap at bun-add cost. |
Formula verdict (INTEGRATE) holds at the composite threshold boundary. Practical note: markdown-it is the actively maintained successor and a stronger choice for net-new integration; remarkable's INTEGRATE rating reflects architectural and gap-fill value, not an unconditional endorsement of this specific library over its successor.
18/24 — stale-risk / well-documented / high-discipline
Failed:
Passed:
Failed:
Passed:
Failed:
Passed:
| Dimension | Score | Assessment |
|---|---|---|
| Harvest Value | 1 | The three-layer named-rule pipeline (block/inline/core) with per-rule enable/disable/replace is a studied architectural pattern applicable to PAI's skill pipeline and hook chain design; nothing novel enough to score 2. |
| Integration Readiness | 1 | Can be bun add remarkable and used in Bun/Node environments, but the package is JavaScript without native TypeScript types — a TS adapter or @types shim would be needed for clean PAI integration, placing this at moderate-glue territory. |
| Overlap Risk | 0 | No existing PAI skill, tool, or hook in the 79-tool/52-skill manifest covers programmatic Markdown parsing; the closest capabilities (Knowledge, Distillator, Editorial) are semantic, not syntactic. |
| Gap Fill | 1 | PAI's file-based memory (WORK/, LEARNING/, KNOWLEDGE/) is entirely markdown+YAML; programmatic parsing is a real but non-critical gap — useful for structured extraction from skill docs and knowledge files, but PAI currently manages without it. |
Composite: 0.63
Capture-to-Knowledge Pipeline ingest layer: Install remarkable (npm install remarkable), instantiate with { commonmark: true }, and pipe each captured note through md.render() before it enters the manual validation stage — every note exits ingest as spec-compliant HTML, eliminating ambiguity in the Haiku validation stage between rendering artifacts and genuine content errors.
Conservancy word-entry rendering: Write a md.core.ruler.push() plugin that intercepts a custom inline marker (e.g., {word:extinct}) and emits <span data-status="extinct" data-word="..."> nodes instead of raw text — SvelteKit templates get structured, CSS-hookable, semantically annotated markup without requiring post-processing of generic Markdown output across the editorial pipeline.
Fabric Recommender pattern indexing: At fab index time, parse each Fabric pattern .md file with remarkable and walk the token stream to tally heading, fence, list, and paragraph token counts, storing a structural fingerprint per pattern. When scoring query matches, weight queries with high code-fence density toward patterns with high fence-token ratios — structural pattern matching that requires zero additional LLM calls and runs entirely at index build time.
Category: Content & Media
In this category: heygen-com--hyperframes (excellent, integrate) — first formal entry for Markdown parsing specifically
Standing: remarkable is the most adopted parsing-focused library in the category with 5.8k stars and a decade of production use, but heygen-com--hyperframes occupies a complementary niche (HTML-to-video rendering); there is no functional overlap between the two.
Density: 9/10 — README (full, 8KB), dependency manifest (package.json with scripts, deps, devDeps), release data (v2.0.0), commit timestamps, star/fork counts, topic tags, license, and CI badge all present; missing: CI workflow file contents, actual test file structure, and contributor activity breakdown.
remarkable is the predecessor project to markdown-it; the author and key contributors effectively migrated efforts there after 2019. Any new PAI integration should evaluate markdown-it first — it ships with TypeScript types natively, has an active release cadence, and is API-compatible in most usage patterns. remarkable's INTEGRATE verdict reflects that the category gap is real and the pattern is sound, not that this specific package is the optimal fill.