refactor: slim principles.md and load in core commands

Reduce principles.md from 158 to 73 lines:
- Remove duplicates (atomic_commits, tdd, deviation_rules)
- Remove version drag from claude_automates
- Keep core orientation: solo dev model, plans are prompts,
  scope control, ship fast, anti-enterprise

Add principles.md to 9 core commands so Claude always
understands what GSD is.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

commands/gsd/create-roadmap.md

@@ -18,6 +18,7 @@ Run after `/gsd:define-requirements`.
 </objective>
 
 <execution_context>
+@~/.claude/get-shit-done/references/principles.md
 @~/.claude/get-shit-done/workflows/create-roadmap.md
 @~/.claude/get-shit-done/templates/roadmap.md
 @~/.claude/get-shit-done/templates/state.md

commands/gsd/define-requirements.md

@@ -23,6 +23,7 @@ Output: `.planning/REQUIREMENTS.md`
 </objective>
 
 <execution_context>
+@~/.claude/get-shit-done/references/principles.md
 @~/.claude/get-shit-done/workflows/define-requirements.md
 @~/.claude/get-shit-done/templates/requirements.md
 </execution_context>

commands/gsd/discuss-milestone.md

@@ -11,6 +11,7 @@ Output: Context gathered, then routes to /gsd:new-milestone
 </objective>
 
 <execution_context>
+@~/.claude/get-shit-done/references/principles.md
 @~/.claude/get-shit-done/workflows/discuss-milestone.md
 </execution_context>
 

commands/gsd/discuss-phase.md

@@ -13,6 +13,7 @@ Output: {phase}-CONTEXT.md capturing the user's vision for the phase
 </objective>
 
 <execution_context>
+@~/.claude/get-shit-done/references/principles.md
 @~/.claude/get-shit-done/workflows/discuss-phase.md
 @~/.claude/get-shit-done/templates/context.md
 </execution_context>

commands/gsd/execute-phase.md

@@ -23,6 +23,7 @@ Context budget: ~15% orchestrator, 100% fresh per subagent.
 </objective>
 
 <execution_context>
+@~/.claude/get-shit-done/references/principles.md
 @~/.claude/get-shit-done/workflows/execute-phase.md
 @~/.claude/get-shit-done/templates/subagent-task-prompt.md
 </execution_context>

commands/gsd/execute-plan.md

@@ -21,6 +21,7 @@ Context budget: ~15% orchestrator, 100% fresh for subagent.
 </objective>
 
 <execution_context>
+@~/.claude/get-shit-done/references/principles.md
 @~/.claude/get-shit-done/templates/subagent-task-prompt.md
 </execution_context>
 

commands/gsd/plan-phase.md

@@ -21,6 +21,7 @@ Output: One or more PLAN.md files in the phase directory (.planning/phases/XX-na
 </objective>
 
 <execution_context>
+@~/.claude/get-shit-done/references/principles.md
 @~/.claude/get-shit-done/workflows/plan-phase.md
 @~/.claude/get-shit-done/templates/phase-prompt.md
 @~/.claude/get-shit-done/references/plan-format.md

commands/gsd/research-phase.md

@@ -27,6 +27,7 @@ Output: RESEARCH.md with ecosystem knowledge that informs quality planning.
 </objective>
 
 <execution_context>
+@~/.claude/get-shit-done/references/principles.md
 @~/.claude/get-shit-done/workflows/research-phase.md
 @~/.claude/get-shit-done/templates/research.md
 @~/.claude/get-shit-done/references/research-pitfalls.md

commands/gsd/research-project.md

@@ -28,6 +28,7 @@ Output: `.planning/research/` folder with ecosystem knowledge.
 </objective>
 
 <execution_context>
+@~/.claude/get-shit-done/references/principles.md
 @~/.claude/get-shit-done/workflows/research-project.md
 @~/.claude/get-shit-done/templates/research-project/SUMMARY.md
 @~/.claude/get-shit-done/templates/research-project/STACK.md

get-shit-done/references/principles.md

@@ -1,5 +1,6 @@
 <principles>
-Core principles for the Gets Shit Done planning system.
+
+Core principles for the GSD planning system.
 
 <solo_developer_claude>
 
@@ -22,15 +23,6 @@ PLAN.md IS the prompt. It contains:
 When planning a phase, you are writing the prompt that will execute it.
 </plans_are_prompts>
 
-<initialization_leverage>
-
-The most leveraged moment is project initialization.
-- Deep questioning here = better everything downstream
-- Garbage in = garbage out
-- Spend the tokens on context gathering
-- Don't rush to "the work"
-</initialization_leverage>
-
 <scope_control>
 
 Plans must complete within reasonable context usage.
@@ -54,62 +46,8 @@ If Claude CAN do it via CLI/API/tool, Claude MUST do it.
 Checkpoints are for:
 - **Verification** - Human confirms Claude's work (visual, UX)
 - **Decision** - Human makes implementation choice
-
-Not for:
-- Deploying (use CLI)
-- Creating resources (use CLI/API)
-- Running builds/tests (use Bash)
-- Writing files (use Write tool)
 </claude_automates>
 
-<deviation_rules>
-
-Plans are guides, not straitjackets. During execution:
-
-1. **Auto-fix bugs** - Fix immediately, document
-2. **Auto-add critical** - Security/correctness gaps, add immediately
-3. **Auto-fix blockers** - Can't proceed, fix immediately
-4. **Ask about architectural** - Major changes, stop and ask
-5. **Log enhancements** - Nice-to-haves, log to Issues, continue
-</deviation_rules>
-
-<test_driven_when_beneficial>
-
-Use TDD when the work WOULD benefit from it. Not dogma—pragmatism.
-
-**TDD candidates (create dedicated TDD plan):**
-- Business logic with defined inputs/outputs
-- API endpoints and handlers
-- Data transformations and parsing
-- Validation rules
-- State machines and workflows
-- Anything where you can describe expected behavior before implementing
-
-**Skip TDD (use standard plan):**
-- UI layout and styling
-- Exploratory prototyping
-- One-off scripts and migrations
-- Configuration changes
-- Glue code with no logic
-
-**Decision heuristic:**
-Can you write `expect(fn(input)).toBe(output)` before writing `fn`?
-→ Yes: Create a TDD plan (one feature per plan)
-→ No: Standard plan, add tests after if needed
-
-**Why TDD gets its own plan:**
-TDD requires 2-3 execution cycles (RED → GREEN → REFACTOR), each with file reads, test runs, and potential debugging. This consumes 40-50% of context for a single feature. Dedicated TDD plans ensure full quality throughout the cycle.
-
-**TDD plan structure:**
-1. Write failing test (RED) → commit
-2. Implement to pass (GREEN) → commit
-3. Refactor if needed → commit
-
-This is about design quality, not test coverage metrics.
-
-See `~/.claude/get-shit-done/references/tdd.md` for TDD plan structure.
-</test_driven_when_beneficial>
-
 <ship_fast>
 
 No enterprise process. No approval gates.
@@ -119,29 +57,6 @@ Plan → Execute → Ship → Learn → Repeat
 Milestones mark shipped versions (v1.0 → v1.1 → v2.0).
 </ship_fast>
 
-<atomic_commits>
-
-**Git commits = context engineering for Claude.**
-
-Each task gets its own commit immediately after completion:
-- Format: `{type}({phase}-{plan}): {task-description}`
-- Types: feat, fix, test, refactor, perf, chore, docs
-- One final metadata commit per plan: `docs({phase}-{plan}): complete [plan-name]`
-
-**Why per-task commits:**
-- Git history becomes primary context source for future Claude sessions
-- `git bisect` finds exact failing task, not just failing plan
-- Each task independently revertable
-- Better failure recovery (task 1 committed ✅, retry task 2)
-- Observability optimized for AI workflow, not human browsing
-
-**Plans produce 3-4 commits total:**
-- 2-3 task commits (working code)
-- 1 metadata commit (SUMMARY + STATE + ROADMAP)
-
-See `~/.claude/get-shit-done/references/git-integration.md` for complete strategy.
-</atomic_commits>
-
 <anti_enterprise>
 
 NEVER include:
@@ -154,4 +69,5 @@ NEVER include:
 
 If it sounds like corporate PM theater, delete it.
 </anti_enterprise>
+
 </principles>