Ejecuta cualquier Skill en Manus
con un clic

Ejecuta cualquier Skill en Manus con un clic

convert

Estrellas0

Forks0

Actualizado18 de junio de 2026 a las 16:49

Convert a non-markdown file into an OKF concept — turn a .txt/.html/.csv/.json file (or, via your own extraction, a PDF/docx) into one conformant markdown concept and wire it into the bundle. Use when the user says "convert this file", "turn this CSV/JSON/HTML into a note", "import this PDF into my knowledge base", or points at a non-markdown source. For loose markdown, use the `add` skill.

Instalación

Instalar con Codex o Claude Copia este prompt, pégalo en Codex, Claude u otro asistente, y deja que revise la página de la skill y la instale por ti.

Ejecutar en Manus

Fuente

RadOrigin-LLC

RadOrigin-LLC/okf-manager

Abrir repositorio de GitHub Ver repositorios del creador

Descarga

Ejecutar en Manus

Ocupaciones relacionadasSOC

Basado en la clasificación ocupacional SOC

Desarrolladores de softwareOcupaciones informáticas y matemáticas·SOC 15-1252

SKILL.md

readonly

name	convert
description	Convert a non-markdown file into an OKF concept — turn a .txt/.html/.csv/.json file (or, via your own extraction, a PDF/docx) into one conformant markdown concept and wire it into the bundle. Use when the user says "convert this file", "turn this CSV/JSON/HTML into a note", "import this PDF into my knowledge base", or points at a non-markdown source. For loose markdown, use the `add` skill.
argument-hint	<source-file> [into <bundle path>]
user-invocable	true
allowed-tools	Bash Read

okfm: convert

Bring a non-markdown source into the bundle as one concept.

Identify the source format:
- .txt / .html / .csv / .json — handled natively by the engine.
- .md — already a concept; use the add skill instead.
- PDF / docx / other binary — read it yourself (your own file reading), then create the concept with new --body "<extracted text>". The engine stays dependency-free.
Apply the capture filter — only convert what is Relevant, Actionable, has Depth, and is Authoritative.
Decide the dest path (<folder>/<slug>.md) and the frontmatter: type, title, description, tags. Read the source to ground these.
One file → one concept. If a source is too large or covers several distinct concepts, split it: run convert (or new --body) once per concept rather than producing one giant note.
Preview: run with --dry-run --json and summarize the plan (source, format, dest, what gets indexed): python "${CLAUDE_PLUGIN_ROOT}/scripts" convert <src> <dest> --bundle <root> --type "<type>" --title "<title>" --description "<desc>" --dry-run --json
On the user's OK, run without --dry-run. The engine converts the body (CSV → table, JSON → fenced block, HTML → headings/lists/links, TXT → text), writes the concept, regenerates the index, and logs a Convert entry.
Run check and report. Collisions are refused, not clobbered — rename or pass --force only on explicit confirmation.

HTML/PDF conversion is best-effort; skim the result and tidy it with the okf conventions (source-preserving edits) if needed.

Más de este repositorio

mismo repositorio

enrich

RadOrigin-LLC/okf-manager

Enrich an OKF bundle from the web — fetch a bounded set of seed URLs, and for each page decide whether to augment an existing concept, add a new reference under references/, or skip. Use when the user says "enrich my bundle from these URLs", "pull in docs from <site>", "add web references", or "research and add sources for <concept>". Requires a domain allowlist and never crawls openly. For seeding from a local structured source (database, OpenAPI, repo), use the `seed` skill.

2026-06-180

add

RadOrigin-LLC/okf-manager

Import existing markdown into an OKF knowledge bundle — apply the capture filter, propose placement, fill in missing frontmatter, and wire it into the index. Use when the user says "import this note", "add this file to my bundle", "bring these markdown files in", or points at a loose .md file or folder. For authoring a concept from scratch, use the `new` skill.

2026-06-180

check

RadOrigin-LLC/okf-manager

Validate an OKF knowledge bundle — conformance (frontmatter + required type), broken cross-links, orphaned concepts, and staleness. Use when the user says "check my OKF bundle", "validate my knowledge base", "what's broken in my notes", or after importing or editing concepts.

2026-06-180

find

RadOrigin-LLC/okf-manager

Search an OKF knowledge bundle by text, type, tag, or status — returns ranked concepts with their paths and the concepts they link to. Use when the user says "find notes about X", "search my knowledge base", "what do I have on Y", "list all <type> concepts", or "show concepts tagged Z".

2026-06-180

map

RadOrigin-LLC/okf-manager

Generate a self-contained HTML graph of an OKF knowledge bundle (no dependencies, opens in a browser). Use when the user says "visualize my knowledge base", "show me the map", "map my OKF bundle", or wants to see how concepts connect.

2026-06-180

move

RadOrigin-LLC/okf-manager

Rename or relocate a concept in an OKF bundle without breaking links — finds every backlink and surgically rewrites it to the new location, then updates the index. Use when the user says "rename this concept", "move this note to", "reorganize my bundle", or "relocate this file". Link-safe and non-destructive.

2026-06-180

name	convert
description	Convert a non-markdown file into an OKF concept — turn a .txt/.html/.csv/.json file (or, via your own extraction, a PDF/docx) into one conformant markdown concept and wire it into the bundle. Use when the user says "convert this file", "turn this CSV/JSON/HTML into a note", "import this PDF into my knowledge base", or points at a non-markdown source. For loose markdown, use the `add` skill.
argument-hint	<source-file> [into <bundle path>]
user-invocable	true
allowed-tools	Bash Read

okfm: convert

Bring a non-markdown source into the bundle as one concept.

Identify the source format:
- .txt / .html / .csv / .json — handled natively by the engine.
- .md — already a concept; use the add skill instead.
- PDF / docx / other binary — read it yourself (your own file reading), then create the concept with new --body "<extracted text>". The engine stays dependency-free.
Apply the capture filter — only convert what is Relevant, Actionable, has Depth, and is Authoritative.
Decide the dest path (<folder>/<slug>.md) and the frontmatter: type, title, description, tags. Read the source to ground these.
One file → one concept. If a source is too large or covers several distinct concepts, split it: run convert (or new --body) once per concept rather than producing one giant note.
Preview: run with --dry-run --json and summarize the plan (source, format, dest, what gets indexed): python "${CLAUDE_PLUGIN_ROOT}/scripts" convert <src> <dest> --bundle <root> --type "<type>" --title "<title>" --description "<desc>" --dry-run --json
On the user's OK, run without --dry-run. The engine converts the body (CSV → table, JSON → fenced block, HTML → headings/lists/links, TXT → text), writes the concept, regenerates the index, and logs a Convert entry.
Run check and report. Collisions are refused, not clobbered — rename or pass --force only on explicit confirmation.

HTML/PDF conversion is best-effort; skim the result and tidy it with the okf conventions (source-preserving edits) if needed.