<< All versions

Skill v1.0.1

currentAutomated scan100/100
vincentkoc/dotskills/graincrawl

feat(technical-documentation): STE rules, large-docs audit playbook, Workflow support (#46)

──Details
PublishedSeptember 12, 2026 at 07:10 AM
Content Hashsha256:0bded3c03e998128...
Git SHA2997164896dd
Compare with v1.0.0
──Files
Files (1 file, 3.4 KB)
SKILL.md3.4 KBactive
SKILL.md · 89 lines · 3.4 KB

version: "1.0.1" name: graincrawl description: Maintain, verify, and release graincrawl, the local-first Granola archive CLI, including SQLite archive behavior, read-only Granola source boundaries, Homebrew tap packaging, and crawlkit-powered TUI/snapshot surfaces. license: MIT metadata: workflow-exemption: "Repository-owned contribution and release lifecycle; this skill supplies archive boundaries and validation commands." source: "https://github.com/vincentkoc/dotskills"


Graincrawl

Purpose

Maintain graincrawl, the Go CLI that archives Granola notes, transcripts, panels, people, workspaces, and sync metadata into a private local SQLite store. Keep it read-only against Granola, SQLite-first for browsing/export, and aligned with the other crawl apps.

When to use

  • Editing or reviewing a graincrawl checkout.
  • Investigating Granola source adapters: private-api, desktop-cache,

encrypted-json, opfs, public-api, or companion-cli.

  • Verifying note, transcript, panel, source object, search, Markdown export, or

TUI behavior.

  • Preparing a graincrawl release, Homebrew formula, or local binary refresh.
  • Checking crawlkit control metadata, snapshots, or TUI integration from the app

side.

Workflow

  1. Locate the graincrawl checkout from the current repository, an explicit

user path, or the configured workspace root. Read repo-local AGENTS.md, README.md, SPEC.md, and docs/security.md before release, source-adapter, or unlock work.

  1. Keep Granola boundaries explicit. Ordinary doctor, status, notes,

export, and tui commands must not prompt Keychain, unwrap safeStorage, or mutate Granola profile files.

  1. Use temp HOME, temp XDG dirs, temp config, and temp SQLite databases for

tests and smoke checks. Do not mutate live ~/.config/graincrawl archives or Granola app data unless the user explicitly asks.

  1. Prefer archived SQLite reads for UI/export behavior. If Markdown export has

the data, TUI detail should read that same archived source, not re-query Granola.

  1. Run the normal local gate with workspace masking disabled:

``bash GOWORK=off go mod tidy git diff --exit-code -- go.mod go.sum GOWORK=off go vet ./... GOWORK=off go test -count=1 ./... make smoke goreleaser check ``

  1. For release work, use semver tags and verify the tap path:

``bash git tag -a v0.1.0 -m "graincrawl v0.1.0" git push origin main v0.1.0 HOMEBREW_NO_AUTO_UPDATE=1 brew install --build-from-source vincentkoc/tap/graincrawl HOMEBREW_NO_AUTO_UPDATE=1 brew test vincentkoc/tap/graincrawl ``

  1. After app changes, refresh the local binary when requested or when handing

off a testable CLI:

``bash GOWORK=off go build -trimpath -o ~/.local/bin/graincrawl ./cmd/graincrawl graincrawl version --json ``

Inputs

  • Target task: app edit, source-adapter investigation, local archive/TUI fix,

release verification, tap update, or binary refresh.

  • Optional version tag such as v0.1.0.
  • Optional temp fixture paths or a copied Granola profile for encrypted/OPFS

investigation.

Outputs

  • A focused graincrawl change or readiness report.
  • Validation commands and results.
  • Explicit notes about any private API, Keychain, safeStorage, OPFS, or live

profile boundary crossed.

  • Tap/release status and local binary path when release or install work is in

scope.

← v1.0.0All versions