parse-knowledge

Overview of parse-knowledge skill

What parse-knowledge does

The parse-knowledge skill turns a messy text dump into a small set of structured Markdown notes for an OrbitOS-style knowledge base. Its core job is not summarization alone: it splits source material into one main research note plus reusable atomic wiki notes, then links them together with a consistent folder layout and YAML frontmatter.

Who should use parse-knowledge skill

parse-knowledge is best for people already keeping notes in an Obsidian-like vault, especially if they use OrbitOS conventions such as 30_Research, 40_Wiki, Areas, Topics, and wikilinks. If you want an AI to transform rough research, copied docs, or brainstorming text into notes you can file immediately, this skill is a better fit than a generic “summarize this” prompt.

What makes parse-knowledge different

The main differentiator is structure enforcement. The skill pushes the model to:

• identify an Area
• create a topic slug
• extract atomic concepts worth separate notes
• rewrite the main note to reference those concepts with wikilinks
• emit vault-ready file contents, not just prose

That makes parse-knowledge for Knowledge Bases useful when your real goal is retrieval, linking, and long-term note maintenance.

When this skill is a poor fit

Skip parse-knowledge if you do not use the OrbitOS folder model, do not want multiple output files, or only need a one-off summary. It also will not validate your vault, create files automatically, or infer deep taxonomy rules beyond what you provide. With only SKILL.md present, adoption is simple, but you must supply the organizational context yourself.

How to Use parse-knowledge skill

Install parse-knowledge in your skill runner

If your environment supports GitHub skills, install from the OrbitOS repository:

npx skills add MarsWang42/OrbitOS --skill parse-knowledge

Then inspect EN/.agents/skills/parse-knowledge/SKILL.md first. There are no companion scripts or templates bundled in the skill folder, so almost all behavior comes from the prompt instructions inside that file.

What input parse-knowledge needs

For good parse-knowledge usage, give it three things:

1. the raw text blob
2. your target vault conventions
3. any category or naming preferences

A weak input:

• “Parse these notes into my vault.”

A strong input:

• “Convert the text below into OrbitOS format. Area should be SoftwareEngineering. Create one main note under 30_Research/SoftwareEngineering/<Topic>/<Topic>.md. Create atomic notes in 40_Wiki/<Category>/. Use concise definitions, strict YAML frontmatter, and aggressive wikilinking in the main note.”

This matters because the skill knows the default structure, but your prompt determines naming precision, scope boundaries, and whether concepts get split too aggressively.

Turn a rough goal into a good parse-knowledge prompt

A practical prompt pattern:

• State the source type: meeting notes, article excerpts, study notes, copied docs
• Name or constrain the Area
• Tell it whether to infer or preserve the topic slug
• Define how many atomic notes are acceptable
• Ask for exact file paths and full file contents
• Mention any forbidden output, such as commentary outside the files

Example workflow prompt:

• “Use parse-knowledge to ingest the text below. Infer the best Topic slug, but keep the Area as ProductManagement. Create one main reference note and up to 5 atomic wiki notes. Prefer durable concepts over project-specific trivia. Output each file with its path and Markdown content only.”

Suggested workflow and files to read first

Read SKILL.md first, then test on one medium-sized text sample before using it on a whole backlog. A good workflow is:

1. run parse-knowledge on a single source
2. review whether the chosen Area, Topic, and atomic concepts match your vault
3. tighten your prompt
4. rerun on larger inputs

Because the skill folder includes only SKILL.md, there are no hidden helper files to learn. The upside is low setup friction; the downside is that output quality depends heavily on your input discipline.

parse-knowledge skill FAQ

Is parse-knowledge better than an ordinary prompt?

Usually yes, if your problem is note structuring rather than simple summarization. A normal prompt may produce a nice summary, but parse-knowledge skill gives the model a clearer target: a main note, atomic concept notes, paths, frontmatter, and wikilink-heavy rewriting. That reduces formatting guesswork.

Is parse-knowledge beginner-friendly?

Yes, with one caution: beginners can install and try it quickly, but the skill assumes you understand your own knowledge-base layout. If you are new to Areas, topic slugs, or atomic notes, start with a small sample and explicitly tell the model what each folder means in your system.

Can I use parse-knowledge outside OrbitOS?

Yes, but only partially. The extraction logic is broadly useful, while the output conventions are OrbitOS-specific. If your vault uses different folders or metadata keys, say so directly in the prompt. Otherwise the skill will bias toward 30_Research, 40_Wiki, and OrbitOS naming patterns.

When should I not install parse-knowledge?

Do not choose parse-knowledge install if you need automatic file creation, schema validation, or robust repo-specific rules. The current skill is lightweight and text-instruction based. It is strongest as a reusable prompting scaffold, not as a full ingestion pipeline.

How to Improve parse-knowledge skill

Give parse-knowledge stronger source material

The biggest quality lever is cleaner input. Separate unrelated topics before running the skill. If one text blob mixes several domains, the model may pick the wrong Area or create vague atomic notes. You get better results when each run covers one coherent subject with enough context to define terms correctly.

Prevent the most common failure modes

Common issues include:

• atomic notes for terms that are too narrow or too obvious
• weak category placement in 40_Wiki
• topic slugs that reflect wording, not enduring concepts
• main notes that paraphrase but do not modularize

To prevent this, specify:

• desired category scheme
• max number of atomic notes
• whether to prefer timeless concepts over source-specific details
• whether examples belong in the main note or wiki note

Improve output quality with tighter review loops

After the first pass, do not just ask for “better.” Ask for a targeted revision:

• “Merge overlapping atomic notes.”
• “Rename the Topic slug to be more evergreen.”
• “Replace generic concepts with domain-specific ones.”
• “Reduce wikilinks to only concepts that deserve standalone notes.”

This makes parse-knowledge guide workflows much more reliable than rerunning from scratch.

Adapt parse-knowledge to your vault conventions

To improve parse-knowledge for Knowledge Bases, add your own house rules in the calling prompt: frontmatter keys, allowed categories, naming style, link style, and note granularity. The skill’s core structure is useful, but its real value appears when you combine it with explicit local conventions so outputs can be dropped into your vault with minimal cleanup.