legends-empire

agent
Security Audit
Warn
Health Warn
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 5 GitHub stars
Code Pass
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested

No AI report is available for this listing yet.

SUMMARY

Turn your Obsidian vault into AI agent memory and a source-cited markdown knowledge base with instant search, the Seven Provinces map, and hash-locked writes.

README.md

legends-empire: turn your Obsidian vault into AI agent memory and a source-cited markdown knowledge base

legends-empire: turn your Obsidian vault into AI agent memory and a markdown knowledge base

release
checks
Python 3.11+
License: MIT

Every coding agent session begins with amnesia unless your agent has a durable memory it cannot corrupt. Point a raw assistant directly at your folder of notes and it will edit files in place, break wikilinks, bloat startup instructions, and blend invented summaries straight into primary research.

Your vault is the memory. The agents do the operating.

legends-empire turns any local Obsidian vault or Markdown folder into source-cited AI agent memory. You speak to your agent in plain english, and legends-empire keeps raw evidence quarantined from synthesized wiki notes, organizes your digital reality across seven provinces you actually own, and protects every file write behind a previewed, hash-locked transaction.

Agent handoff: paste this into Claude Code, Codex, or any coding agent and it does the setup

Set up legends-empire through cto-legends and initialize or adopt my Obsidian vault,
showing me the plan before writing any files.

Three pillars of legends-empire: 01 The seven provinces, 02 Source-cited memory, and 03 Hash-locked writes

Plain english · Seven provinces · Source-cited memory · Stewardship & ontology · Reference


AI agent memory in plain english

You should never have to babysit file paths, frontmatter schemas, or SHA-256 digests by hand just to keep an Obsidian vault clean. You tell your coding agent what you want remembered, searched, or audited in normal words, and cto-legends routes the job straight to legends-empire.

Say the job in plain words: four spoken sentences routing directly into legends-empire vault operations

Here are four everyday sentences legends-empire was built to answer.

"Set up my Obsidian vault."

Whether you are starting a brand-new folder or adopting a vault you have lived in for years, legends-empire previews every file before touching your disk. New vaults receive the complete Seven Provinces starter map, the wiki/hot.md doorway for fast agent orientation, an inbox/ drop zone, and an immutable .raw/ evidence store. Existing vaults are adopted non-destructively without overwriting a single note or Obsidian setting you customized by hand.

"Save this research in my vault."

When legends-dataforseo pulls live Google search demand or legends-firecrawl scrapes a competitor's site and PDF brochures, they export portable evidence packages bound to a cryptographic manifest. legends-empire verifies every byte offline, locks the raw files inside .raw/ so primary data is never altered, and synthesizes source-cited wiki notes that link back to the exact origin.

"Search my vault for our pricing notes."

Instead of grepping thousands of random lines or re-reading your entire folder into the context window on every turn, your agent queries the local search index (empire search) and the ontology index in milliseconds. It gets back the exact notes, frontmatter facts, and source citations relevant to your question.

"What needs attention in my vault?"

Vault stewardship scans your Markdown knowledge base read-only for broken wikilinks, orphan notes, expired freshness windows, and stale project pages. It also audits your coding agent startup files (CLAUDE.md, AGENTS.md, GEMINI.md) so your agent instructions stay lean, consistent across hosts, and free of duplicated rules.


The seven provinces: Map and Blueprint

Your digital reality is already scattered across folders, repositories, servers, financial records, and half-built ideas. The challenge is keeping track of what actually exists today versus what you intend to build tomorrow.

legends-empire ships a whole-life starter framework organized into seven provinces. Inside every province live two distinct sections: The Map and The Blueprint.

The seven provinces: Map and Blueprint across Work, Money, Craft, Home, People, Self, and Ambition

The Map records verified reality: what is live right now, where its canonical files or repositories sit, and when it was last checked. The Blueprint records future intention. Keeping them separate enforces one crucial discipline for both humans and AI agents: a planned system is never confused with a working system, and an unwritten ambition stays unwritten until you decide what it is.

Province Starter path What lives here
Work wiki/work/Work.md Active operations, ventures, client deliveries, and commitments consuming your attention
Money wiki/money/Money.md Resources, obligations, pricing, runway, ownership, and the room you have to maneuver
Craft wiki/craft/Craft.md The things you make and the disciplines you sharpen because mastery and taste matter
Home wiki/home/Home.md The physical world you inhabit, your property and hardware, and the household machinery
People wiki/people/People.md Relationships, promises, collaborators, and the allies who matter beyond a transaction
Self wiki/self/Self.md Health, energy, capacity, operating rhythm, and the human running all of this machinery
Ambition wiki/ambition/Ambition.md What all of the machinery is for: long-range targets before they become active Work projects

The provinces organize attention without forcing you to move your existing project folders or git repositories. Your operational folders stay right where they work best, while the seven provinces act as your navigable map over the whole terrain.


Why source-cited memory beats raw agent file edits

Giving a coding agent raw filesystem write access to your personal or company notes feels convenient until the first time an assistant silently overwrites a decision log, invents a citation, or deletes an unlinked receipt during "cleanup."

Comparison: Direct agent file edits versus One vault engine with legends-empire

legends-empire treats your Markdown vault with the same rigor a database engineer gives a production ledger:

  • One canonical home per fact. Raw captures and API payloads live in .raw/ bound to a SHA-256 source ledger. Human-readable synthesis lives in wiki/. An agent summary never overwrites the primary source behind it.
  • Hash-locked write transactions. Every proposed mutation binds each target file to its expected prior SHA-256 digest. If you edited a note in Obsidian while the agent was thinking, the transaction halts cleanly instead of clobbering your work.
  • Crash-safe recovery across macOS, Linux, and Windows NTFS. Every transaction records a private recovery journal before modifying files, so an interrupted process can be rolled back to a verified state in one step.
  • Hot doorway orientation (wiki/hot.md). Instead of stuffing your entire vault into every prompt, wiki/hot.md gives your coding agent a short, current doorway into what matters right now and where the deeper notes live.

Vault stewardship, instant search and the ontology engine

As a Markdown knowledge base grows from fifty notes to five thousand, three capabilities keep it fast and trustworthy without turning your files into a proprietary database.

Instant local search and hybrid retrieval

Your notes remain plain .md files on your own drive, ready to open in Obsidian at any moment. Behind the scenes, legends-empire indexes headings, wikilinks, and note bodies locally so your agent can run instant BM25 and contextual retrieval searches offline without sending your private vault to a third-party vector cloud.

The ontology engine

Standard Markdown frontmatter becomes an answerable fact graph (docs/ontology.md). You and your agent can declare structured entities, aliases, and certified frontmatter facts while preserving every byte of your human prose below the frontmatter fence. Automated workflows can check whether a required fact is fresh and certified before acting, halting safely when key information is still a draft or out of date.

Vault stewardship and attachable knowledge packs

Running a stewardship check (legends-vault-stewardship, bundled inside the same installation) gives your agent a repeatable health report: broken wikilinks, unreadable files, expired freshness dates, and drift across your coding agent entry files.

Stewardship also lets you attach curated Markdown knowledge packs (docs/knowledge-packs.md) from sibling modules such as legends-hetzner, legends-coolify, and legends-cloudflare, bringing verified infrastructure runbooks straight into your vault without touching your private notes.


Reference

Everything you need to run legends-empire through your coding agent is in the essay above. Open these drawers whenever you want the CLI command reference, architecture layout, or release notes.

CLI reference and agent setup

Agent setup (via cto-legends)

Part of the cto-legends ecosystem. cto-legends is the only registered skill; this repository vendors a pinned copy at skills/cto-legends/SKILL.md. Install via cto-legends install legends-empire and let the router resolve docs/AGENT-RECIPE.md on demand. Do not register this module as a standalone skill folder.

python -m cto_legends --home <cto-legends-home> sync --apply
python -m cto_legends --home <cto-legends-home> install legends-empire --apply
python -m cto_legends --home <cto-legends-home> handoff legends-empire

Run vault memory, search, ontology, or stewardship commands through the router from any working directory:

python -m cto_legends --home <cto-legends-home> run legends-empire -- --help
python -m cto_legends --home <cto-legends-home> run legends-vault-stewardship -- steward --help
Surface Key subcommands What it does
Vault setup & memory init, adopt, doctor, capture, plan-transaction, apply-transaction, recover, checkpoint Non-destructive Seven Provinces initialization, source-cited capture, hash-locked transactions, and crash recovery
Search & retrieval search (retrieve.py, bm25-index.py, read-hot-cache.sh) Instant local search, BM25 indexing, and wiki/hot.md orientation
Research intake verify-research-evidence, plan-research-intake Offline SHA-256 verification and intake planning for legends-dataforseo and legends-firecrawl evidence packs
Ontology engine ontology resolve, ontology ask, ontology need, ontology propose, ontology certify, ontology bench Sub-millisecond entity resolution, frontmatter fact queries, prerequisite gating, and differential frontmatter updates
Vault stewardship steward scan, steward observe, steward review, steward entry-audit, steward digest, steward law Read-only link and freshness audits, multi-vendor agent startup audits, and attachable legends.knowledge-pack/v1 packs
Vault layout, platform support and attribution
Obsidian Markdown Vault (user-owned files)
  +-- wiki/
  |     +-- hot.md, index.md, overview.md, log.md
  |     +-- work/Work.md, money/Money.md, craft/Craft.md, home/Home.md
  |     +-- people/People.md, self/Self.md, ambition/Ambition.md
  |     +-- meta/ (ledgers and ontology starter)
  +-- inbox/ (drop zone for captures and research intake)
  +-- .raw/ (immutable SHA-256 primary evidence archive)
  +-- .obsidian/ (graph view colors and workspace config)

Start here

Paste the agent handoff block at the top of this page into your coding agent, let it wire legends-empire through cto-legends, and give your agent a vault memory that compounds across every session.

GitHub Star List: The complete cto-legends collection on one page. Bookmark or star the router and every module from a single GitHub list.

Running a company and want this vault architecture and agent stack applied directly to your business? Message Avalon Reset at cto-legends.com/dm. The desk is async, so the conversation starts the moment you write and nobody forces you onto a sales call.

Want this way of working applied to your own stack? Message Avalon Reset at cto-legends.com/dm


Specification & benchmarks on cto-legends.com:

Reviews (0)

No results found