vault-challenge
Skill0 starsChanged 7 months agoMITView on GitHub
Vault argues back. Pulls your own history to challenge current thinking. Use when making decisions, proposing architecture, or before committing to a direction. The vault is not a yes-man -- it remembers what you've learned the hard way.
In this repository9 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- llm-wikiSkill
- vault-connectSkill
- vault-emergeSkill
- vault-graduateSkill
- vault-ingestSkill
- vault-saveSkill
- vault-worldSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 4 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 18
If no explicit topic, challenge the most recent significant decision in conversation.
- Very specific: line 22
History agent: Search daily notes (last 30 days) + log.md for past instances where the user tried something similar.
- Distinctive: line 23
Decision agent: Search project + research folders for past decisions on the same topic.
- Distinctive: line 24
Pattern agent: Search other domain folders for structural analogues.
- Distinctive: line 66
Before committing significant time/resources
≈ “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.
Files
/vault-challenge [topic or decision]
The vault challenges your current thinking using your own history.
Steps
- Read
_CLAUDE.mdat vault root for folder map. - Identify the claim -- what is the user proposing, deciding, or assuming? If no explicit topic, challenge the most recent significant decision in conversation.
Deep vault search -- spawn 3 parallel search agents:
- History agent: Search daily notes (last 30 days) +
log.mdfor past instances where the user tried something similar. What happened? What was the lesson? - Decision agent: Search project + research folders for past decisions on the same topic. Read "Key Decisions" sections. Were there reversals?
- Pattern agent: Search other domain folders for structural analogues. Has this pattern failed in a different domain?
- History agent: Search daily notes (last 30 days) +
Build the challenge:
## Challenge: [one-line summary of what's being questioned] ### Your Own Words > [Direct quotes from vault notes that contradict or complicate the current plan] ### What Happened Last Time [Concrete outcome from past attempt, with date and source note] ### The Steel-Man Counter [Strongest possible argument against the current direction, built from vault evidence] ### What This Doesn't Mean [Acknowledge where the current plan IS strong -- fair challenge, not nihilism] ### Questions You Should Answer Before Proceeding 1. [Specific question derived from evidence] 2. [Another] 3. [Another]- If no counter-evidence found -- say so honestly. "Vault has nothing against this -- either it's genuinely new territory or your notes don't cover this domain yet."
- Append to log.md:
## [YYYY-MM-DD] challenge | [topic] -- [outcome: challenged/clear]
Rules
- All output in the user's language (from
_CLAUDE.mdLanguage section). The templates below are structural examples in English -- translate ALL headings, labels, and descriptions. E.g., Chinese user sees "## 挑战:", "### 你自己说过的话", "### 上次发生了什么", not English headings. Only direct quotes from vault notes stay in their original language. - Quote the user's own words -- this is what makes it hit different from generic critique
- Never fabricate evidence -- if the vault doesn't have it, say so
- Be specific -- "you tried X on 2026-03-15 and it failed because Y" not "you've had mixed results"
- Include the source note path so they can go read the full context
- This is not a blocker -- present evidence, let the user decide
When to Suggest
- User says "I'm going to..." or "let's do..." on something non-trivial
- Architecture decisions that affect multiple projects
- Before committing significant time/resources
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
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.

