agentleFS
willsu88willsu88/skills/obsidian-capture/SKILL.md

obsidian-capture

Skill2 starsChanged 4 months agoMITView on GitHub

Distill an insight or story from the current conversation into a note in the user's Obsidian vault - use when the user says "save this", "capture this", or "add this to my notes", or when another skill needs to record into the vault.

3 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

12 rules. About 0 are common, 10 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈10

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Ask the user once, and suggest recording the answer in their global CLAUDE.md so this question never repeats.

    line 16
  2. Distinctive:

    If the standard PARA folders are not present, list the vault root and map to the structure that is actually there.

    line 19
  3. Distinctive:

    If the user named a target note or section, that is the target.

    line 25
  4. Very specific:

    Search the vault (find/grep by keyword) for an existing note that should hold it.

    line 28
  5. Distinctive:

    Prefer folding into an existing note over creating a new one - a vault of many tiny orphan notes is worse than fewer living ones.

    line 29
  6. Very specific:

    If nothing fits, create a new note in the folder that matches: active work with an end date -> Projects/, ongoing responsibilities -> …

    line 30
  7. Distinctive:

    If two existing notes both plausibly fit, ask the user rather than guessing.

    line 31
  8. Distinctive:

    Never restructure a note while capturing into it.

    line 35
  9. Distinctive:

    Keep the user's voice: first person ("I asked", "I checked") when the story is about the user's actions - these notes are often rehearsal …

    line 54
  10. Distinctive:

    Not a journal: no timestamps-of-conversation, no "Claude said / I said" framing unless the human-AI exchange is itself the point of the …

    line 58
  11. Distinctive:

    Not a dump: if the capture exceeds ~25 lines, it is probably a transcript wearing a trench coat - distill harder.

    line 59
  12. Distinctive:

    Not a restructuring tool: it never renames, moves, or reorganizes existing notes.

    line 60

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Obsidian capture

Turn something that just happened or was learned in this conversation into a durable note in the user's Obsidian vault.

Step 0: Resolve the vault path

In order, use the first that resolves:

  1. The OBSIDIAN_VAULT environment variable.
  2. A vault path recorded in the user's global CLAUDE.md or memory.
  3. Ask the user once, and suggest recording the answer in their global CLAUDE.md so this question never repeats.

This skill assumes a PARA-organized vault. If the standard PARA folders are not present, list the vault root and map to the structure that is actually there.

Process

  1. Identify the capture. From the user's request and the conversation, pin down the specific example, exchange, decision, or insight to save - not a transcript, the distilled version. If the user named a target note or section, that is the target; skip to step 3.
  2. Find where it belongs. Search the vault (find/grep by keyword) for an existing note that should hold it. Prefer folding into an existing note over creating a new one - a vault of many tiny orphan notes is worse than fewer living ones. If nothing fits, create a new note in the folder that matches: active work with an end date -> Projects/, ongoing responsibilities -> Areas/, study or reference material -> Resources/; when genuinely unclear, Inbox/. If two existing notes both plausibly fit, ask the user rather than guessing.
  3. Read the target note first. Match its existing structure, heading style, and voice. Never restructure a note while capturing into it; append or insert in the section the user pointed at.
  4. Distill, don't transcribe. The standard shape for an example or story:
    • What happened, concretely - real file names, real claims, real commands, absolute dates (never "today").
    • The turning point - the question asked, the check performed, the thing that contradicted the assumption.
    • The outcome - what changed as a result.
    • The transferable principle - one or two sentences stating what generalizes, bolded or labeled so it can be found at a glance. Cut anything that only made sense in the conversation's context.
  5. Report. Tell the user the note path, where in it the capture landed, and the principle line as written.

Style rules (non-negotiable)

  • Each full sentence on its own physical line.
  • Plain dash - only; never an em dash.
  • Wiki-links [[Note Name]] for references to other vault notes; add them liberally where a related note exists.
  • Absolute dates (2026-07-27), never relative ones.
  • Keep the user's voice: first person ("I asked", "I checked") when the story is about the user's actions - these notes are often rehearsal material the user will speak from.

What this skill is not

  • Not a journal: no timestamps-of-conversation, no "Claude said / I said" framing unless the human-AI exchange is itself the point of the story.
  • Not a dump: if the capture exceeds ~25 lines, it is probably a transcript wearing a trench coat - distill harder.
  • Not a restructuring tool: it never renames, moves, or reorganizes existing notes.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.