examples/reference-agent/skills/knowledge-harvest/SKILL.md
Use to EXTRACT multiple patterns from a source — code, docs, PRs, articles. Triggers on "learn from this", "harvest knowledge", "ingest this document", "extract patterns from". The agent reads the source and identifies what to capture. For saving a single known item, use vault-capture instead.
npx skillsauth add adrozdenko/soleri knowledge-harvestInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Point at code, docs, PRs, architecture decisions, or postmortems — the agent extracts every pattern, anti-pattern, decision, and principle, then captures them to the vault.
Read target content and classify: salvador_core op:route_intent params: { prompt: "Extract knowledge from: <source>" }
salvador_core op:search_intelligent
params: { query: "<topic of source material>" }
salvador_core op:vault_tags
salvador_core op:vault_domains
Focus extraction on gaps — skip what vault already covers.
| Type | What to Look For | | ---------------- | ------------------------------------ | | pattern | Repeatable approaches that work | | anti-pattern | Known mistakes to avoid | | decision | Architectural choices with rationale | | principle | Guiding rules or heuristics | | workflow | Step-by-step procedures |
For each: determine category, severity, and tags.
salvador_core op:capture_knowledge
params: {
title: "<clear, searchable name>",
description: "<what it is, when to apply, why it matters>",
type: "<pattern|anti-pattern|decision|principle|workflow>",
category: "<domain>",
tags: ["<tag1>", "<tag2>"],
example: "<code snippet or quote>",
why: "<reasoning>"
}
Present each capture as you go: Captured: api-auth-jwt (pattern, critical)
op:curator_detect_duplicates — find duplicates created during harvestop:curator_groom_all — normalize tags, fix metadataop:curator_contradictions — check for conflicts## Harvest Complete
Source: [name]
Extracted: X entries (Y patterns, Z anti-patterns, W decisions)
Duplicates: N (merged/skipped)
Contradictions: N (flagged)
Vault health: OK
Optionally promote universal patterns: op:memory_promote_to_global.
| Op | When to Use |
| ------------------------------ | ------------------------ |
| search_intelligent | Check existing knowledge |
| vault_tags / vault_domains | See what's covered |
| capture_knowledge | Capture each item |
| curator_detect_duplicates | Post-harvest dedup |
| curator_groom_all | Normalize entries |
| curator_contradictions | Find conflicts |
| memory_promote_to_global | Share cross-project |
| admin_vault_analytics | Knowledge quality |
testing
Triggers: "terse mode", "be brief", "less tokens", "fewer tokens", "compress output", "caveman", or invokes /terse. Token-efficient responses with full technical accuracy.
tools
Triggers: "compress this file", "compress CLAUDE.md", "compress memory", "shrink this", "reduce tokens in file", or invokes /compress. Compresses natural language files to save input tokens.
testing
Triggers: "release", "bump version", "publish packages", "cut a release", "version bump", "npm publish". Bumps monorepo versions, commits, tags, pushes to trigger CI release. Use deliver-and-ship for quality gates.
development
Triggers: "implement X", "build Y", "fix Z", "add feature", or any work task needing planning + execution. Full orchestration loop: plan, execute, complete with vault context and brain recs.