feat: dogfood Bonsai on itself — generate station/ workspace with tech-lead agent

Ran `bonsai init` on the Bonsai repo itself. Migrated hand-crafted agent/
directory into Bonsai-generated station/ workspace with full scaffolding,
protocols, sensors, workflows, routines, and customized identity/memory.
Retired old agent/ and root Backlog.md.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: LastStep

.bonsai.yaml

@@ -0,0 +1,42 @@
+project_name: Bonsai
+description: CLI tool for scaffolding Claude Code agent workspaces
+docs_path: station/
+scaffolding:
+    - index
+    - playbook
+    - logs
+    - reports
+agents:
+    tech-lead:
+        agent_type: tech-lead
+        workspace: station/
+        skills:
+            - planning-template
+            - review-checklist
+        workflows:
+            - code-review
+            - planning
+            - pr-review
+            - security-audit
+            - session-logging
+            - test-plan
+        protocols:
+            - memory
+            - scope-boundaries
+            - security
+            - session-start
+        sensors:
+            - context-guard
+            - scope-guard-files
+            - session-context
+            - status-bar
+            - routine-check
+        routines:
+            - backlog-hygiene
+            - dependency-audit
+            - doc-freshness-check
+            - infra-drift-check
+            - memory-consolidation
+            - roadmap-accuracy
+            - status-hygiene
+            - vulnerability-scan

.claude/settings.json

@@ -0,0 +1,49 @@
+{
+  "hooks": {
+    "PreToolUse": [
+      {
+        "hooks": [
+          {
+            "type": "command",
+            "command": "bash station/agent/Sensors/scope-guard-files.sh"
+          }
+        ],
+        "matcher": "Edit|Write"
+      }
+    ],
+    "SessionStart": [
+      {
+        "hooks": [
+          {
+            "type": "command",
+            "command": "bash station/agent/Sensors/session-context.sh"
+          },
+          {
+            "type": "command",
+            "command": "bash station/agent/Sensors/routine-check.sh"
+          }
+        ]
+      }
+    ],
+    "Stop": [
+      {
+        "hooks": [
+          {
+            "type": "command",
+            "command": "bash station/agent/Sensors/status-bar.sh"
+          }
+        ]
+      }
+    ],
+    "UserPromptSubmit": [
+      {
+        "hooks": [
+          {
+            "type": "command",
+            "command": "bash station/agent/Sensors/context-guard.sh"
+          }
+        ]
+      }
+    ]
+  }
+}

.gitignore

@@ -3,14 +3,17 @@ bonsai
 bonsai.exe
 Bonsai
 
-# Python
-.venv/
-
 # Go
 *.test
 *.out
 vendor/
 
+# Bonsai
+.bonsai-lock.yaml
+*.bak
+
+# Legacy
+.venv/
 
 # Research / scratch
 RESEARCH*.md

Backlog.md

@@ -5,7 +5,7 @@
 **Stack:** Go 1.24+, Cobra, Huh (forms), LipGloss (styling), BubbleTea (TUI)
 
 > [!warning]
-> **FIRST:** Read `agent/Core/identity.md`, then `agent/Core/memory.md`.
+> **FIRST:** Read `station/agent/Core/identity.md`, then `station/agent/Core/memory.md`.
 
 ---
 
@@ -14,7 +14,8 @@
 ```
 Bonsai/
 ├── CLAUDE.md               ← you are here
-├── Backlog.md               ← prioritized todo queue (bugs, features, debt, research)
+├── .bonsai.yaml             ← project config (dogfooding — Bonsai manages itself)
+├── .bonsai-lock.yaml        ← file tracking (hashes + sources)
 ├── main.go                  ← entry point, embeds catalog/ via embed.FS
 ├── go.mod / go.sum          ← module config
 ├── Makefile                 ← build, install, clean
@@ -29,7 +30,8 @@ Bonsai/
 │   ├── catalog/
 │   │   └── catalog.go       ← loads YAML metadata from embedded catalog/
 │   ├── config/
-│   │   └── config.go        ← ProjectConfig, InstalledAgent + YAML I/O
+│   │   ├── config.go        ← ProjectConfig, InstalledAgent + YAML I/O
+│   │   └── lockfile.go      ← LockFile, content hashing, conflict detection
 │   ├── generate/
 │   │   ├── generate.go      ← renders templates, writes files to target project
 │   │   └── generate_test.go ← tests for core file generation
@@ -39,45 +41,44 @@ Bonsai/
 ├── catalog/                  ← bundled catalog (embedded into binary)
 │   ├── core/                 ← shared core files (memory, self-awareness)
 │   ├── agents/               ← agent type definitions + identity templates
-│   │   ├── tech-lead/
-│   │   ├── backend/
-│   │   └── frontend/
 │   ├── skills/               ← à la carte skills (meta.yaml + content.md)
 │   ├── workflows/            ← à la carte workflows
 │   ├── protocols/            ← à la carte protocols
 │   ├── sensors/              ← auto-enforced hooks (meta.yaml + script.sh.tmpl)
 │   ├── routines/             ← periodic maintenance routines (meta.yaml + content.md.tmpl)
 │   └── scaffolding/          ← project management infrastructure templates
-│       ├── manifest.yaml     ← scaffolding item definitions (name, description, required, affects)
-│       ├── INDEX.md.tmpl
-│       ├── Playbook/         ← Status, Roadmap, Backlog, Plans, SecurityStandards
-│       ├── Logs/             ← FieldNotes, KeyDecisionLog, RoutineLog
-│       └── Reports/          ← report-template, Pending/
-└── agent/                    ← agent instructions (this agent)
-    ├── Core/                 ← identity, memory
-    └── Skills/               ← domain skills + references
+└── station/                  ← tech-lead workspace (generated by bonsai init)
+    ├── CLAUDE.md             ← workspace nav (generated, then customized)
+    ├── index.md              ← code index — quick-nav to Go source
+    ├── INDEX.md              ← project snapshot
+    ├── Playbook/             ← status, roadmap, backlog, plans, standards
+    ├── Logs/                 ← field notes, key decisions, routine log
+    ├── Reports/              ← report templates, pending reports
+    └── agent/                ← agent instructions
+        ├── Core/             ← identity, memory, self-awareness, routines dashboard
+        ├── Skills/           ← planning-template, review-checklist, bubbletea
+        ├── Workflows/        ← code-review, planning, pr-review, security-audit, etc.
+        ├── Protocols/        ← memory, scope-boundaries, security, session-start
+        ├── Sensors/          ← context-guard, scope-guard-files, session-context, status-bar, routine-check
+        └── Routines/         ← backlog-hygiene, dependency-audit, doc-freshness-check, etc.
 ```
 
 ---
 
-## Skills (load when doing specific work)
+## Agent Instructions
 
-| Need | Read this |
-|------|-----------|
-| BubbleTea TUI development | `agent/Skills/bubbletea.md` |
-
-References for each skill live in a subdirectory (e.g. `agent/Skills/bubbletea/`) — load progressively as needed.
+All agent instructions live in `station/`. Start with `station/CLAUDE.md` for the full navigation table.
 
 ---
 
 ## Memory
 
 > [!warning]
-> **Do NOT use Claude Code's auto-memory system** (`~/.claude/projects/*/memory/`). All persistent memory goes in `agent/Core/memory.md` — version-controlled, auditable, inside the project.
+> **Do NOT use Claude Code's auto-memory system** (`~/.claude/projects/*/memory/`). All persistent memory goes in `station/agent/Core/memory.md` — version-controlled, auditable, inside the project.
 
-When you would normally write to auto-memory (feedback, references, project context, flags), write to the appropriate section in `agent/Core/memory.md` instead.
+When you would normally write to auto-memory (feedback, references, project context, flags), write to the appropriate section in `station/agent/Core/memory.md` instead.
 
-When you discover bugs, improvement ideas, tech debt, or feature requests outside your current task scope, add them to `Backlog.md` instead of fixing them inline or noting them only in memory.
+When you discover bugs, improvement ideas, tech debt, or feature requests outside your current task scope, add them to `station/Playbook/Backlog.md` instead of fixing them inline or noting them only in memory.
 
 ---