How do I install the Jason Corpus Search skill?

Run npx skillmds@latest add asmartbear/jason-corpus-search in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Jason Corpus Search safe to use?

This skill has not completed SkillMD's automated safety review yet. Capability flags: executes scripts. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Jason Corpus Search?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Jason Corpus Search free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Jason Corpus Search?

asmartbear (@asmartbear) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Jason Corpus Search by asmartbear — Agent Skill

Name: Jason Corpus Search
Rating: 4.2 (5 reviews)
Author: asmartbear

Jason Corpus Search

Semantically search Jason Cohen's A Smart Bear articles and Hidden Multipliers book corpus, returning on-disk file paths that can be read or grep'd directly with normal filesystem tools. Load whenever you need to find what Jason has written about a topic, ground an asb-* skill in source material, hunt for quotes or passages, or check whether an idea already has a canonical write-up before drafting one.

asmartbear Updated Sep 10, 2026

File contents

Searching Jason's source corpus

Jason's writing — A Smart Bear articles and the Hidden Multipliers book — is indexed by a semantic-search CLI in a separate repo. This skill is the thin shim that lets you query it and then work with the hits as ordinary files on disk.

The corpus lives in exactly two directories:

/Users/jcohen/Obsidian/Longform/Projects/Articles/Content/ — A Smart Bear articles
/Users/jcohen/Obsidian/Longform/Projects/Hidden Multipliers/Content/ — the book

You can Grep or Glob either directory directly when an exact-string sweep is what you want (e.g. "find every place Jason uses the phrase 'profit whale'"). Use the semantic wrapper below when you want meaning-based recall.

Filenames are slugs (e.g. pricing-determines-your-business-model.md), and internal markdown links within these files reference siblings by that slug. If you're following a link like […](more-value-not-save-money) from inside one of these files, the target is <same-directory>/<slug>.md — just Read it directly, no search needed.

When to use

Drafting or revising an asb-* skill and you need to ground the framework in Jason's own prose.
Answering "what has Jason written about X?" or "does Jason have a take on Y?"
Pulling a specific phrasing or example before paraphrasing it into a skill.

The one move

Run the wrapper. It returns a minimal JSON array of {title, path} where path is the on-disk file. Use Read / Grep / Glob from there.

.claude/skills/jason-corpus-search/search.sh --semantic "pricing power" --limit 10

[
  {"title": "pricing-determines-your-business-model",
   "path": "/Users/jcohen/Obsidian/Longform/Projects/Articles/Content/pricing-determines-your-business-model.md"},
  ...
]

From there: Read a specific hit, Grep across the list for a phrase, Glob for nearby siblings in the same folder. The corpus is small; the source filesystem is fast.

The wrapper (`search.sh`)

Bakes in two unbypassable filters so results never leak into daily notes, KB, AMA, blurbs, errata, or backups:

--domain obsidian
--path '^/Longform/Projects/(Articles/Content|Hidden Multipliers/Content)/'

Then trims each hit to {title, path} (on-disk path). All other args are pass-through to cli storage search. In practice you only need two:

Flag	Use
`--semantic "<phrase>"`	The actual query — phrase it like a sentence, not keywords. Searches two indexes in parallel: embeddings of the literal content AND embeddings of each doc's topics + the common questions it answers. So both content-level queries ("companies that raised prices and grew") and topical queries ("pricing strategy", "how should I price my SaaS?") work well.
`--limit <n>`	Cap results. Set this; the CLI default may not match your need.

Do not bypass the wrapper

Calling cli storage search directly will silently include daily notes, the AMA inbox, knowledge-base scraps, book blurbs, errata, and other vault content that is not Jason's published material. Always go through search.sh.

---name: jason-corpus-searchdescription: Semantically search Jason Cohen's A Smart Bear articles and Hidden Multipliers book corpus, returning on-disk file paths that can be read or grep'd directly with normal filesystem tools. Load whenever you need to find what Jason has written about a topic, ground an asb-* skill in source material, hunt for quotes or passages, or check whether an idea already has a canonical write-up before drafting one.---# Searching Jason's source corpusJason's writing — A Smart Bear articles and the Hidden Multipliers book — isindexed by a semantic-search CLI in a separate repo. This skill is the thinshim that lets you query it and then work with the hits as ordinary files ondisk.The corpus lives in exactly two directories:- `/Users/jcohen/Obsidian/Longform/Projects/Articles/Content/` — A Smart Bear articles- `/Users/jcohen/Obsidian/Longform/Projects/Hidden Multipliers/Content/` — the bookYou can `Grep` or `Glob` either directory directly when an exact-string sweepis what you want (e.g. "find every place Jason uses the phrase 'profitwhale'"). Use the semantic wrapper below when you want meaning-based recall.Filenames are slugs (e.g. `pricing-determines-your-business-model.md`), andinternal markdown links within these files reference siblings by that slug.If you're following a link like `[…](more-value-not-save-money)` from insideone of these files, the target is `<same-directory>/<slug>.md` — just `Read`it directly, no search needed.## When to use- Drafting or revising an `asb-*` skill and you need to ground the framework in Jason's own prose.- Answering "what has Jason written about X?" or "does Jason have a take on Y?"- Pulling a specific phrasing or example before paraphrasing it into a skill.## The one moveRun the wrapper. It returns a minimal JSON array of `{title, path}` where`path` is the on-disk file. Use Read / Grep / Glob from there.```sh.claude/skills/jason-corpus-search/search.sh --semantic "pricing power" --limit 10``````json[ {"title": "pricing-determines-your-business-model", "path": "/Users/jcohen/Obsidian/Longform/Projects/Articles/Content/pricing-determines-your-business-model.md"}, ...]```From there: `Read` a specific hit, `Grep` across the list for a phrase, `Glob`for nearby siblings in the same folder. The corpus is small; the sourcefilesystem is fast.## The wrapper (`search.sh`)Bakes in two unbypassable filters so results never leak into daily notes, KB,AMA, blurbs, errata, or backups:- `--domain obsidian`- `--path '^/Longform/Projects/(Articles/Content|Hidden Multipliers/Content)/'`Then trims each hit to `{title, path}` (on-disk path). All other args arepass-through to `cli storage search`. In practice you only need two:| Flag | Use || :--- | :-- || `--semantic "<phrase>"` | The actual query — phrase it like a sentence, not keywords. Searches two indexes in parallel: embeddings of the literal content AND embeddings of each doc's topics + the common questions it answers. So both content-level queries ("companies that raised prices and grew") and topical queries ("pricing strategy", "how should I price my SaaS?") work well. || `--limit <n>` | Cap results. Set this; the CLI default may not match your need. |## Do not bypass the wrapperCalling `cli storage search` directly will silently include daily notes, theAMA inbox, knowledge-base scraps, book blurbs, errata, and other vaultcontent that is not Jason's published material. Always go through`search.sh`.

Frequently asked questions