obsidian-distill

obsidian-distill is a command for Claude Code from eugeniughelbur/obsidian-second-brain. It costs 24 tokens per session (1,581 once invoked), scanned A, original, MIT.

An Obsidian command that turns a long note, folder, file, or web source into key claims linked to the exact original passages. Obsidian is a note-taking app that uses linked Markdown files.

In plain words
What is it for?
Distilling research notes and source material into reference notes while preserving links to the evidence.
Why use it?
It keeps condensed information traceable, so someone can check each claim instead of relying on an unsupported summary.

Command for Claude Code

Written for Claude Code: $ARGUMENTS substitution.

Part of the obsidian-second-brain plugin — 1 skill, 47 commands, 3 hooks shipped together

Good fit Distilling research notes and source material into reference notes while preserving links to the evidence.

Compare 6 commands from other repositories ↓
Install with agentmods
npx agentmods add commands/eugeniughelbur/obsidian-second-brain/obsidian-distill
About the project

obsidian-second-brain turns an Obsidian vault into persistent, searchable memory for Claude Code and other command-line coding agents, storing knowledge as linked Markdown notes. It is for developers, founders, writers, and researchers who want agents to retain project context across sessions. Its catalogue entries provide commands, hooks, a plugin, a skill, and instructions for capturing, finding, and maintaining that memory.

eugeniughelbur/obsidian-second-brain · 4,392 stars · on GitHub · eugeniughelbur.github.io

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Clone the repo
git clone --depth 1 https://github.com/eugeniughelbur/obsidian-second-brain

Made for: Claude Code.

Or install obsidian-second-brain, the plugin that ships this one along with the rest of its 1 skill, 47 commands, 3 hooks.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for obsidian-distill

README.md
[![agentmods](https://agentmods.dev/badge/commands/eugeniughelbur/obsidian-second-brain/obsidian-distill/github.svg)](https://agentmods.dev/commands/eugeniughelbur/obsidian-second-brain/obsidian-distill)
Your own site
<a href="https://agentmods.dev/commands/eugeniughelbur/obsidian-second-brain/obsidian-distill"><img src="https://agentmods.dev/badge/commands/eugeniughelbur/obsidian-second-brain/obsidian-distill/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for obsidian-distill

Your own site · 80×15
<a href="https://agentmods.dev/commands/eugeniughelbur/obsidian-second-brain/obsidian-distill"><img src="https://agentmods.dev/badge/commands/eugeniughelbur/obsidian-second-brain/obsidian-distill.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 24 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,581 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00024 $0.01581
Opus 5 $0.00012 $0.00790
Sonnet 5 $0.00005 $0.00316
Haiku 4.5 $0.00002 $0.00158

Measured 11d ago against content hash 2e38bb7393eb, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade A, and why

obsidian-distill scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

commands/obsidian-distill.md · 49 lines

How it starts

The opening of the file, as written. The whole thing — 49 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Use the obsidian-second-brain skill. Execute /obsidian-distill $ARGUMENTS:

The optional argument is a note path, a [[wikilink]], a folder, or a source URL/file. If none is given, ask what to distill (offer the longest recently-touched notes in raw/ and the research folders as candidates).

A distillation is not a summary. A summary throws the source away; a distillation keeps a verifiable trail, so a teammate (or a future agent) can check every claim against the exact place it came from. This is the trust primitive: condensed but never unmoored from evidence.

  1. Read _CLAUDE.md first if it exists in the vault root. Resolve the destination folder per references/folder-map.md (a distillation is a concept/reference note: wiki-style wiki/concepts/, Obsidian-style Knowledge/).

  2. Resolve and read the source:

    • A vault note or [[wikilink]]: read it in full.
    • A folder: read every note in it and distill them as one corpus (note which file each claim comes from).
    • A URL: fetch it (read the page). A local file: read it. Save the untouched original to raw/ first if it is not already there - the distillation must point at a stable source.
    • Never distill from memory or a partial read. If the source is too large to read in one pass, read it in ordered chunks and keep going; do not sample (see the anti-fabrication rule).
  3. Segment the source into stable, citable source blocks and number them. A block is a heading section, a paragraph, a list, or a transcript turn - whatever the source's natural unit is. Record for each block a locator the reader can find again: a heading path, a paragraph index, or a timestamp/line range. These locators are the provenance anchors.

  4. Extract the key claims. For each claim, write one tight sentence and attach its provenance: which block(s) it came from, as (src: B3) or (src: B3, B7). Every claim MUST carry at least one block reference - a claim with no source does not go in the distillation (if you believe it but the source does not say it, it is your inference, not a distilled claim; see step 6). Mark each claim's confidence (stated | high | medium | speculation) and add a recency marker on any time-sensitive external fact ((as of YYYY-MM, domain)).

  5. Group the claims under short thematic headings (the distilled structure), preserving the provenance tags inline. Keep the source's own terms; do not smooth them into generic phrasing.

  6. Keep inference separate from evidence. If the distillation surfaces a conclusion the source implies but never states, put it under a clearly labelled ## Inferences (not in the source) section, each marked confidence: speculation, so distilled fact and your reasoning never blur.

  7. Write the distillation note. Path: <resolved-folder>/Distill - <source-title> (YYYY-MM-DD).md. It MUST follow references/ai-first-rules.md. If that path does not resolve from your working directory, search upward for it; if you still cannot read it, say so before writing rather than producing a note that silently skips the rule. Frontmatter includes type: distillation, ai-first: true, source (the verbatim path/URL of what was distilled), source-blocks (the count), date, tags: [distillation, thinking], related-people and related-projects (wikilinks, per the type: distillation schema). Body order:

    • ## For future agent - 2-3 sentences: what was distilled, why, and that every claim carries a (src: Bn) pointer back to the numbered source blocks at the bottom for verification.
    • ## Distilled claims - the grouped claims with inline provenance tags.
    • ## Inferences (not in the source) - only if step 6 produced any.
    • ## Source blocks - the numbered list (B1, B2, ...) with each block's locator (heading/paragraph/timestamp) and a short quote or first line, so the anchors are resolvable from inside the vault. This is the recency/verification anchor.
  8. Link the distillation from the original source note (add a Distilled: [[...]] line) and from today's daily note. Append a one-line entry to the operation log (Logs/YYYY-MM-DD.md if it exists, else log.md).

Read the full file on GitHub · 49 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 11d ago First seen · 49 lines · 24 tokens per session scan A 2e38bb7393eb

Subscribe to this mod's changes

obsidian-distill is a command published in the GitHub repository eugeniughelbur/obsidian-second-brain (4,392 stars, last pushed 4d ago), licensed MIT. It adds 24 tokens to every session and 1,581 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.