Skip to content
SkillModel-led872 tokens in use

think

A ten-stage review loop for consequential or ambiguous decisions, from observation to a growth step, that stays read-only.

16,653 stars on the repository MIT LicenseAgriciDaniel/claude-obsidian · skills/think
claude plugin marketplace add AgriciDaniel/claude-obsidian claude plugin install claude-obsidian@agricidaniel-claude-obsidian
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing think costs

84tokens idle
872tokens in use

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Counts are approximate, made offline with cl100k_base.

Model dependence Model-led

Script-ledMixedModel-led

Mostly instructions, so how well it works depends on the model that follows them.

This describes the skill's structure, how much of it is scripts. It does not say how well it works.

Touches

Nothing detected: no shell, network, file writes, git or credentials in its files.

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Good
  • Frontmatter8/15
  • Structure16/25
  • Examples9/15
  • Output15/15
  • Size15/15
  • Resources10/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent

No findings.

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Good
  • Development activity84
  • Maintenance100
  • Community56
  • Maturity28
  • Momentum40

think applies a ten-stage loop to hard reasoning: observe externally and internally, listen, think, connect laterally and across the system, feel, accept, create and grow. Depth scales with novelty, stakes, uncertainty and how reversible the decision is.

It first separates what was observed, inferred, unknown and a matter of preference, then ends with a decision, evidence, counterargument, verification and a growth loop. It changes no files.

What You Get

  • An evidence frame that splits observed, inferred, unknown and preference
  • A recommendation or decision record with its assumptions marked
  • The strongest counterargument and a list of risks and unknowns
  • A verification section naming checks run and checks still needed
  • A growth loop with a feedback signal and a revisit condition

Common Use Cases

  • Reviewing an architecture choice before committing to it
  • Running a postmortem on a failed release
  • Testing a decision against its best counterargument

What Makes It Unique

It is a reasoning review, not a vault health check or a repair workflow, and it holds a firm read-only boundary: any change it recommends goes through a separate mutation skill after you authorize it. It asks for calibrated qualitative confidence with reasons and discourages decorative percentages.

Under The Hood

The skill is a single instruction file with no scripts and no references. It does not run commands, fetch web pages, call APIs or write files. It may read the evidence that is already available in the session, and saving a lesson is left to the separate save skill.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers