agentleFS
SeanathonSeanathon/vault-starter/.claude/skills/consult/SKILL.md

consult

Skill0 starsChanged 49 days agoMITView on GitHub

Consult the vault's accumulated knowledge for a decision or topic — gather relevant claims and typed relations, deepen shallow-but-load-bearing sources on the spot, and return a balanced, cited consultation. Use when the user says /consult {topic}, "what does the vault say about X", or is making a work decision the vault has coverage on.

21 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use discover, init and 4 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • anombyte93Skill

    discover

    Phase 1 of the prd-taskmaster pipeline: brainstorm-driven discovery. Delegates to superpowers:brainstorming in Interactive Mode (one adaptive question at a time), or self-brainstorms in Autonomous Mode when no user is present. Intercepts before the brainstorming chain hands off to writing-plans — this skill owns the exit. Extracts constraints, calibrates scale (Solo / Team / Enterprise), and advances the pipeline to GENERATE.

    Also in 3 projects that use this

  • MicrosoftSkill

    init

    Also in 3 projects that use this

  • AnthropicSkill

    status

    Case status summary by audience — client-facing (plain language), internal (for the professor), or court-ready (formal caption format per local rules). Same facts, different framing and depth. Use when a student needs to update the client, brief the professor, or prepare a court status report.

    Also in 3 projects that use this

  • MicrosoftSkill

    council

    Convene the persona panel — six orthogonal review lenses — on a target. Cold independent fan-out, debate-to-consensus, synthesized verdict with recorded dissent and a roster manifest.

    Also in 3 projects that use this

  • AWSSkill

    ship

    Finalize dev work on an issue branch — update docs, push, and open a PR.

    Also in 3 projects that use this

  • Hugging FaceSkill

    implement

    Make tests pass. Invoke after /write-tests produces failing tests.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 17 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Vault first, labeled honestly: Search before answering. Anything from general knowledge rather than the vault is clearly marked as such.

    line 12
  2. Distinctive:

    Search broadly across the graph — not just filenames:

    line 22
  3. Distinctive:

    Do any claims conflict? That's a finding, not noise.

    line 42
  4. Distinctive:

    If the decision is one the user owns deeply, name the one or two sources worth their own pass-3 reading.

    line 51

“Common” is measured: kept by at least 6 in 10 of the 17 edited versions of this file we could read. “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

/consult — bring the vault's knowledge to a decision

Input: a topic, question, or decision ("stopping conditions for my harness", "should I build a custom harness or adopt one"). Output: a consultation built from the vault's claims and relations, with citations, tensions, and gaps — not a summary of one note, and not general knowledge dressed up as vault knowledge.

Contract (binding)

  1. Vault first, labeled honestly. Search before answering. Anything from general knowledge rather than the vault is clearly marked as such. If the vault has nothing relevant, say so plainly — never pad.
  2. Close coverage gaps before consulting on them. A relevant source still at processing: captured (or an unpromoted bookmark) is a skim of a skim — not consultable evidence. Deepen load-bearing shallow sources first (per the /deepen contract), capped at 2 per consult; list any further shallow-but-relevant sources as "shallow — not deeply consulted" rather than silently treating them as read.
  3. No recency bias. Material deepened during this consult gets no extra weight for being fresh or effortful. Enforce structurally: finish all deepening before forming any conclusion; then build one claim table across ALL relevant sources as equals; weigh by evidence quality (data > worked example > assertion), source incentive (vendor pitch vs. practitioner report vs. research), and corroboration (claims supported across sources beat any single source). Tag just-deepened sources in the output so the user sees their provenance.
  4. Tensions are the product. contradicts:: edges and disagreeing claims are the most valuable output — surface them explicitly instead of averaging them away.
  5. This is an advisor, not the user's understanding. The consultation retrieves; it does not mean the user has internalized anything. For decisions the user owns deeply, recommend which one or two sources deserve their own pass-3 read.

Procedure

1. Gather

Search broadly across the graph — not just filenames:

  • 50-resources/sources/ records: claims (^claim- blocks), Connections sections, executive takeaways.
  • 40-areas/ synthesis and topic notes; 50-resources/concepts/.
  • Typed relations (supports::, extends::, contradicts::, special-case-of::, supersedes::) touching the topic — follow edges one hop.
  • 60-bookmarks/ for relevant unpromoted captures.
  • Evidence packages only to verify a quotation or read a slide frame, not as primary text.

2. Assess coverage

Sort what surfaced: deep (processed/connected/synthesized, or evergreen notes), shallow (captured records, bare bookmarks), absent (aspects of the question the vault simply doesn't cover).

If a shallow source is load-bearing — its claims would plausibly change the answer — run /deepen pass 2 on it now (full read, claims with block IDs, evidence quality, limitations; connect pass if relations are obvious). Max 2 per consult; report any that remain shallow. These deepenings persist and are logged normally — a consult should leave the vault better than it found it.

3. Weigh — after all reading is done

Build a single claim table across every relevant source: claim, source, evidence type, incentive, corroboration. Only then form conclusions. Checks:

  • Does any conclusion rest solely on a just-deepened source? Flag it as single-source, newly read rather than presenting it as settled.
  • Is an older, thinner note being ignored because a richer note is fresher? Corroboration and evidence quality decide, not richness.
  • Do any claims conflict? That's a finding, not noise.

4. Deliver the consultation

  • Answer first: what the vault's knowledge implies for this specific decision, in a few sentences.
  • Grounds: the key claims with citations ([[Note#^claim-slug]], timestamps for video claims), grouped by where they agree.
  • Tensions: where sources disagree or a relation marks contradiction/supersession — stated as open choices for the user, with what each side's evidence is.
  • Gaps: what the question needs that the vault doesn't hold (candidate /know captures), and shallow sources not deeply consulted.
  • Provenance flags: which sources were deepened during this consult; which claims are single-source; which come from parties with an incentive.
  • If the decision is one the user owns deeply, name the one or two sources worth their own pass-3 reading.

5. Log

Append one row: | YYYY-MM-DD | query | /consult: <topic> — <n> sources (<k> deepened), <m> claims, <t> tensions |. If sources were deepened, their /deepen log rows are appended too. Do not update the index.

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.