fix(execute-phase): add one-subagent-per-plan constraint

Context isolation is the primary goal, not parallelization.
Even sequential plans spawn separate subagents for fresh 200k context.

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

.planning/ROADMAP.md

@@ -19,6 +19,7 @@ None - this is internal GSD development following existing command/workflow/temp
 - [ ] **Phase 3: Integration** - Wire brownfield support into existing GSD workflows
 - [x] **Phase 10: Parallel Phase Execution** - Separate single-plan vs multi-plan execution with intelligent parallelization
 - [x] **Phase 11: Parallel-Aware Planning** - Update plan-phase.md to create parallelizable plans when config enables it
+- [x] **Phase 99: Test Parallel (THROWAWAY)** - Create 3 silly independent files to test parallel execution
 
 ## Phase Details
 
@@ -190,6 +191,22 @@ Current plan-phase.md has sequential execution bias - later plans reference earl
 
 This enables execute-phase to produce more Wave 1 plans (true independence) instead of sequential chains.
 
+### Phase 99: Test Parallel (THROWAWAY)
+
+**Goal:** Create 3 independent silly files to test parallel execution - DELETE AFTER TESTING
+**Depends on:** Nothing (independent test)
+**Research:** No
+**Plans:** 3 plans (all parallelizable)
+
+Plans:
+- [x] 99-01: Create animal-facts.md - A file with 5 animal facts
+- [x] 99-02: Create dad-jokes.md - A file with 5 dad jokes
+- [x] 99-03: Create random-numbers.md - A file with 5 random numbers
+
+**Details:**
+Each plan touches completely different files, no dependencies, perfect for testing Wave 1 parallel execution.
+Files go in: `test-output/` (gitignored throwaway directory)
+
 ## Progress
 
 **Execution Order:**
@@ -208,3 +225,4 @@ Phases execute in numeric order: 1 → 2 → 3 → 4 → 5 → 6
 | 9. Integrate Verify-Work | 1/1 | Complete | 2026-01-08 |
 | 10. Parallel Phase Execution | 4/4 | Complete | 2026-01-12 |
 | 11. Parallel-Aware Planning | 4/4 | Complete | 2026-01-12 |
+| 99. Test Parallel (THROWAWAY) | 3/3 | Complete | 2026-01-12 |

.planning/STATE.md

@@ -19,12 +19,12 @@
 
 ## Current Position
 
-Phase: 11 of 11 (Parallel-Aware Planning)
-Plan: 4 of 4 in current phase
-Status: Phase complete
-Last activity: 2026-01-12 - Completed 11-04-PLAN.md
+Phase: 99 of 99 (Test Parallel - THROWAWAY)
+Plan: 3 of 3 in current phase
+Status: Phase complete (via parallel execution)
+Last activity: 2026-01-12 - Completed Phase 99 via /gsd:execute-phase
 
-Progress: ██████████████ 100% (23 of 23 plans)
+Progress: Test phase complete (throwaway - not counted in main milestone)
 
 ## Performance Metrics
 
@@ -48,10 +48,11 @@ Progress: ██████████████ 100% (23 of 23 plans)
 | 9 | 1 | 3 min | 3 min |
 | 10 | 4 | 33 min | 8.3 min |
 | 11 | 4 | 12 min | 3 min |
+| 99 | 3 | 1 min | <1 min (parallel) |
 
 **Recent Trend:**
-- Last 5 plans: 11-01 (4m), 11-02 (2m), 11-03 (3m), 11-04 (3m)
-- Trend: Workflow updates fast (avg 3m)
+- Last 5 plans: 11-02 (2m), 11-03 (3m), 11-04 (3m), 99-01/02/03 (<1m parallel)
+- Trend: Parallel execution dramatically faster
 
 *Updated after each plan completion*
 
@@ -105,5 +106,5 @@ Drift notes: None
 ## Session Continuity
 
 Last session: 2026-01-12
-Stopped at: Completed 11-04-PLAN.md - Phase 11 complete, milestone complete
+Stopped at: Completed Phase 99 via /gsd:execute-phase (parallel test)
 Resume file: None

commands/gsd/execute-phase.md

@@ -21,6 +21,8 @@ Execute all unexecuted plans in a phase with parallel agent spawning.
 Analyzes plan dependencies to identify independent plans that can run concurrently.
 Spawns background agents for parallel execution, each agent commits its own tasks atomically.
 
+**Critical constraint:** One subagent per plan, always. This is for context isolation, not parallelization. Even strictly sequential plans spawn separate subagents so each starts with fresh 200k context at 0%.
+
 Use this command when:
 - Phase has 2+ unexecuted plans
 - Want "walk away, come back to completed work" execution

get-shit-done/workflows/execute-phase.md

@@ -1,6 +1,8 @@
 <purpose>
 Execute all plans in a phase with intelligent parallelization.
 Analyzes plan dependencies to identify independent plans that can run in parallel.
+
+**Critical constraint:** One subagent per plan, always. This is for context isolation, not parallelization. Even strictly sequential plans spawn separate subagents so each starts with fresh 200k context at 0%. Quality degrades above 50% context - executing multiple plans in one subagent defeats the entire segmentation model.
 </purpose>
 
 <when_to_use>