@brainjar/cli 0.2.2 → 0.2.3

package/README.md

@@ -9,6 +9,8 @@ Shape how your AI thinks — identity, soul, persona, rules.
 
 brainjar manages AI agent behavior through composable layers. Instead of one monolithic config file, you separate **what the agent sounds like** (soul), **how it works** (persona), and **what constraints it follows** (rules). Each layer is a markdown file. Mix and match them per project, per task, or per session.
 
+**[Documentation](https://brainjar.sh)** · **[Getting Started](https://brainjar.sh/getting-started/)**
+
 ## Quick start
 
 ```bash
@@ -22,272 +24,14 @@ brainjar init --default
 brainjar status
 ```
 
-That gives you a soul (craftsman), a persona (engineer), and rules (boundaries, git discipline, security) — all wired into your `CLAUDE.md` and ready to go.
-
 ## Concepts
 
-brainjar has four core concepts. Understanding these makes everything else click.
-
-### Soul — who the agent is
-
-The soul defines personality: tone, character, standards. It's the constant across all tasks. You probably only have one or two. Think of it as the agent's voice.
-
-### Persona — how the agent works
-
-Personas define role and workflow. An engineer persona works differently than a reviewer or an architect. Switch personas based on what you're doing — they're the agent's job description.
-
-Personas bundle their own rules via frontmatter:
-
-```yaml
----
-rules:
-  - default
-  - security
----
-```
-
-### Rules — what the agent must follow
-
-Rules are behavioral constraints — guardrails that apply regardless of persona. Single files or multi-file packs (directories). Rules from a persona's frontmatter activate automatically when that persona is active.
-
-### Brain — a saved configuration
-
-A brain is a named snapshot of soul + persona + rules. Instead of switching three things separately, save your setup once and activate it in one shot. Useful for repeatable workflows like "code review" or "design session."
-
-### How they compose
-
-```
-soul + persona + rules = full agent behavior
-      │                    │
-      └── bundled rules ───┘  (from persona frontmatter)
-```
-
-A brain captures all three layers. `compose` assembles them into a single prompt for subagent dispatch.
-
-## State cascade
-
-State merges in three tiers. Each tier overrides the previous:
-
-```
-global  →  local  →  env
-```
-
-| Tier | Storage | When to use |
-|------|---------|-------------|
-| **Global** | `~/.brainjar/state.yaml` | Default behavior across all projects |
-| **Local** | `.brainjar/state.yaml` (in project) | Per-project overrides |
-| **Env** | `BRAINJAR_*` environment variables | Per-session or CI overrides |
-
-```bash
-# Global (default for all projects)
-brainjar persona use engineer
-
-# Local (this project only)
-brainjar persona use planner --local
-
-# Env (this session only)
-BRAINJAR_PERSONA=reviewer claude
-
-# Or use a subshell for scoped sessions
-brainjar shell --persona reviewer --rules-add security
-```
-
-`brainjar status` shows where each setting comes from:
-
-```
-soul     craftsman (global)
-persona  planner (local)
-rules    default (global), security (+local)
-```
-
-### Scope annotations
-
-When you see scope labels in `status` and `rules list` output:
-
-| Label | Meaning |
+| Layer | Purpose |
 |-------|---------|
-| `(global)` | Set in `~/.brainjar/state.yaml` |
-| `(local)` | Overridden in `.brainjar/state.yaml` |
-| `(+local)` | Added by local override (not in global) |
-| `(-local)` | Removed by local override (active globally, suppressed here) |
-| `(env)` | Overridden by `BRAINJAR_*` env var |
-| `(+env)` | Added by env var |
-| `(-env)` | Removed by env var |
-
-### Environment variables
-
-These env vars override all other state when set:
-
-| Variable | Effect |
-|----------|--------|
-| `BRAINJAR_HOME` | Override `~/.brainjar/` location |
-| `BRAINJAR_SOUL` | Override active soul |
-| `BRAINJAR_PERSONA` | Override active persona |
-| `BRAINJAR_IDENTITY` | Override active identity |
-| `BRAINJAR_RULES_ADD` | Comma-separated rules to add |
-| `BRAINJAR_RULES_REMOVE` | Comma-separated rules to remove |
-
-Set to empty string to explicitly unset (e.g., `BRAINJAR_SOUL=""` removes the soul for that session).
-
-## What it does
-
-```
-~/.brainjar/
-  souls/            # Voice and character — who the agent is
-    craftsman.md
-  personas/         # Role and workflow — how the agent works
-    engineer.md
-    planner.md
-    reviewer.md
-  rules/            # Constraints — what the agent must follow
-    default/        # Boundaries, context recovery, task completion
-    git-discipline.md
-    security.md
-  brains/           # Full-stack snapshots — soul + persona + rules
-    review.yaml
-```
-
-brainjar reads these markdown files, merges the active layers, and inlines them into your agent's config (`~/.claude/CLAUDE.md` or `.codex/AGENTS.md`) between `<!-- brainjar:start -->` / `<!-- brainjar:end -->` markers. Everything outside the markers is yours. Change a layer, and the agent's behavior changes on next sync.
-
-## Layers
-
-### Soul
-
-```bash
-brainjar soul create mysoul --description "Direct and rigorous"
-brainjar soul use mysoul
-```
-
-### Persona
-
-```bash
-brainjar persona use planner    # Design session
-brainjar persona use engineer   # Build session
-brainjar persona use reviewer   # Review session
-```
-
-### Rules
-
-```bash
-brainjar rules create no-delete --description "Never delete files without asking"
-brainjar rules add no-delete
-brainjar rules remove no-delete
-```
-
-### Brain
-
-```bash
-brainjar brain save review           # Snapshot current state as a brain
-brainjar brain use review            # Activate soul + persona + rules in one shot
-brainjar brain list                  # See available brains
-brainjar brain show review           # Inspect a brain's config
-```
-
-When to use a brain vs. switching layers individually:
-- **Brain:** Repeatable workflow you do often (code review, design, debugging). Save once, activate in one command.
-- **Individual layers:** Exploratory work, one-off overrides, or when you only need to change one thing.
-
-## Subagent orchestration
-
-Personas can spawn other personas as subagents. For example, a tech-lead persona can:
-
-1. Spawn an **architect** subagent for design — produces a design doc
-2. Get user approval
-3. Implement the design itself
-4. Spawn a **reviewer** subagent to verify — compares code against the spec
-
-Each subagent gets the full brain context: soul + persona + rules. The `compose` command assembles the full prompt in a single call:
-
-```bash
-# Primary path — brain drives everything
-brainjar compose review --task "Review the changes in src/sync.ts"
-
-# Ad-hoc — no saved brain, specify persona directly
-brainjar compose --persona reviewer --task "Review the changes in src/sync.ts"
-```
-
-For more granular control, use `brainjar persona show <name>` and `brainjar rules show <name>` to retrieve individual layers.
-
-## Recipes
-
-### Code review session
-
-```bash
-# Save a review brain once
-brainjar soul use craftsman
-brainjar persona use reviewer
-brainjar rules add default
-brainjar rules add security
-brainjar brain save review
-
-# Then activate it anytime
-brainjar brain use review
-
-# Or scope it to a single session
-brainjar shell --brain review
-```
-
-### CI pipeline — enforce rules without a persona
-
-```bash
-# In CI, use env vars to override behavior
-BRAINJAR_PERSONA=auditor BRAINJAR_RULES_ADD=security,compliance brainjar status --sync
-```
-
-### Project-specific persona
-
-```bash
-# In your project directory
-brainjar persona use planner --local
-brainjar rules add no-delete --local
-
-# Global settings still apply — local just overrides what you specify
-brainjar status
-# soul     craftsman (global)
-# persona  planner (local)
-# rules    default (global), no-delete (+local)
-```
-
-## Pack
-
-Packs are self-contained, shareable bundles of a brain and all its layers — soul, persona, and rules. Export a brain as a pack directory, hand it to a teammate, and they import it in one command.
-
-A pack mirrors the `~/.brainjar/` structure with a `pack.yaml` manifest at the root. No tarballs, no magic — just files you can inspect with `ls` and `cat`.
-
-```bash
-# Export a brain as a pack
-brainjar pack export review                        # creates ./review/
-brainjar pack export review --out /tmp             # creates /tmp/review/
-brainjar pack export review --name my-review       # override pack name
-brainjar pack export review --version 1.0.0        # set version (default: 0.1.0)
-brainjar pack export review --author frank         # set author field
-
-# Import a pack
-brainjar pack import ./review                      # import into ~/.brainjar/
-brainjar pack import ./review --force              # overwrite conflicts
-brainjar pack import ./review --merge              # rename conflicts as <name>-from-<packname>
-brainjar pack import ./review --activate           # activate the brain after import
-```
-
-On conflict (a file already exists with different content), import fails by default and lists the conflicts. Use `--force` to overwrite or `--merge` to keep both versions. Identical files are silently skipped.
-
-## Hooks
-
-brainjar integrates with Claude Code's hook system for automatic context injection. When hooks are installed, brainjar syncs your config on every session start — no manual `brainjar sync` needed.
-
-```bash
-# Install hooks (writes to ~/.claude/settings.json)
-brainjar hooks install
-
-# Install for this project only
-brainjar hooks install --local
-
-# Check hook status
-brainjar hooks status
-
-# Remove hooks
-brainjar hooks remove
-```
+| **Soul** | Who the agent is — voice, character, standards |
+| **Persona** | How the agent works — role, workflow, bundled rules |
+| **Rules** | Behavioral constraints — guardrails that apply regardless of persona |
+| **Brain** | Saved snapshot of soul + persona + rules — activate in one shot |
 
 ## Commands
 
@@ -296,7 +40,6 @@ brainjar init [--default] [--obsidian] [--backend claude|codex]
 brainjar status [--sync] [--global|--local] [--short]
 brainjar sync [--quiet]
 brainjar compose <brain> [--task <text>]
-brainjar compose --persona <name> [--task <text>]
 
 brainjar brain save|use|list|show|drop
 brainjar soul create|list|show|use|drop
@@ -310,38 +53,7 @@ brainjar shell [--brain|--soul|--persona|--identity|--rules-add|--rules-remove]
 brainjar reset [--backend claude|codex]
 ```
 
-`show` accepts an optional name to view any item, not just the active one. Use `--short` to get just the active name (useful in scripts and statuslines):
-
-```bash
-brainjar persona show reviewer    # View a specific persona
-brainjar soul show                # View the active soul
-brainjar rules show security      # View a rule's content
-brainjar status --short           # One-liner: soul | persona | identity
-brainjar soul show --short        # Just the active soul name
-```
-
-## Obsidian support
-
-`~/.brainjar/` is a folder of markdown files — it's already almost an Obsidian vault.
-
-```bash
-brainjar init --obsidian
-```
-
-Adds `.obsidian/` config that hides private files (state, identities) from the file explorer and includes templates for creating new souls, personas, and rules from within Obsidian.
-
-## Backends
-
-```bash
-brainjar init --backend codex      # writes ~/.codex/AGENTS.md
-brainjar reset --backend codex
-```
-
-Supported: `claude` (default), `codex`
-
-## Backup & restore
-
-On first sync, brainjar backs up any existing config file to `CLAUDE.md.pre-brainjar`. Running `brainjar reset` removes brainjar-managed config and restores the backup.
+See the [CLI reference](https://brainjar.sh/reference/cli/) for full details.
 
 ## Development
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@brainjar/cli",
-  "version": "0.2.2",
+  "version": "0.2.3",
   "description": "Shape how your AI thinks — composable identity, soul, persona, and rules for AI agents",
   "type": "module",
   "license": "MIT",
@@ -8,7 +8,7 @@
     "type": "git",
     "url": "https://github.com/brainjar-sh/brainjar-cli.git"
   },
-  "homepage": "https://github.com/brainjar-sh/brainjar-cli",
+  "homepage": "https://brainjar.sh",
   "bugs": "https://github.com/brainjar-sh/brainjar-cli/issues",
   "keywords": [
     "ai",