diff --git a/agents/gsd-researcher.md b/agents/gsd-researcher.md
index 49bcac568..cd02480ad 100644
--- a/agents/gsd-researcher.md
+++ b/agents/gsd-researcher.md
@@ -24,47 +24,14 @@ Your job: Answer research questions with verified, actionable findings. Produce
-## Why You Research
+## Research Feeds Planning
-Your research output directly feeds into `/gsd:plan-phase`. The planning workflow loads your RESEARCH.md and uses specific sections to create better plans.
+Your output is consumed by downstream GSD workflows. The orchestrator's prompt tells you:
+- `` — Phase research vs project research
+- `` — What workflow uses your output and how
+- `` — Checklist before declaring complete
-**The contract:**
-
-| Your Section | How plan-phase Uses It |
-|--------------|------------------------|
-| `## Standard Stack` | Plans use these libraries, not alternatives |
-| `## Architecture Patterns` | Task structure follows these patterns |
-| `## Don't Hand-Roll` | Tasks NEVER build custom solutions for listed problems |
-| `## Common Pitfalls` | Verification steps check for these |
-| `## Code Examples` | Task actions reference these patterns |
-
-**What this means for your output:**
-
-1. **Section names matter** — plan-phase greps for these exact headings
-2. **Be prescriptive, not exploratory** — "Use X" not "Consider X or Y"
-3. **Actionable over comprehensive** — A focused stack beats a survey of options
-4. **Examples are instructions** — Code snippets become task templates
-
-**Bad research (for GSD):**
-```markdown
-## Options for State Management
-There are several approaches: Redux, Zustand, Jotai, or React Context...
-```
-
-**Good research (for GSD):**
-```markdown
-## Standard Stack
-| Library | Version | Purpose |
-|---------|---------|---------|
-| Zustand | 4.5 | State management |
-
-## Don't Hand-Roll
-| Problem | Use Instead |
-|---------|-------------|
-| Global state | Zustand (not React Context) |
-```
-
-The second version tells the planner exactly what to put in tasks.
+**Universal principle:** Be prescriptive, not exploratory. "Use X" beats "Consider X or Y." Your research becomes instructions.
diff --git a/commands/gsd/research-phase.md b/commands/gsd/research-phase.md
index e22dbe7a4..41e408609 100644
--- a/commands/gsd/research-phase.md
+++ b/commands/gsd/research-phase.md
@@ -61,6 +61,23 @@ Present summary with phase description, requirements, prior decisions.
Research modes: ecosystem (default), feasibility, implementation, comparison.
```markdown
+
+Phase Research — investigating HOW to implement a specific phase well.
+
+
+
+The question is NOT "which library should I use?"
+
+The question is: "What do I not know that I don't know?"
+
+For this phase, discover:
+- What's the established architecture pattern?
+- What libraries form the standard stack?
+- What problems do people commonly hit?
+- What's SOTA vs what Claude's training thinks is SOTA?
+- What should NOT be hand-rolled?
+
+
Research implementation approach for Phase {phase_number}: {phase_name}
Mode: ecosystem
@@ -73,6 +90,26 @@ Mode: ecosystem
**Phase context:** {context_md_content}
+
+Your RESEARCH.md will be loaded by `/gsd:plan-phase` which uses specific sections:
+- `## Standard Stack` → Plans use these libraries
+- `## Architecture Patterns` → Task structure follows these
+- `## Don't Hand-Roll` → Tasks NEVER build custom solutions for listed problems
+- `## Common Pitfalls` → Verification steps check for these
+- `## Code Examples` → Task actions reference these patterns
+
+Be prescriptive, not exploratory. "Use X" not "Consider X or Y."
+
+
+
+Before declaring complete, verify:
+- [ ] All domains investigated (not just some)
+- [ ] Negative claims verified with official docs
+- [ ] Multiple sources for critical claims
+- [ ] Confidence levels assigned honestly
+- [ ] Section names match what plan-phase expects
+
+
diff --git a/commands/gsd/research-project.md b/commands/gsd/research-project.md
index 9ca26b519..2766f7101 100644
--- a/commands/gsd/research-project.md
+++ b/commands/gsd/research-project.md
@@ -64,24 +64,168 @@ Present for approval.
mkdir -p .planning/research
```
-Spawn all 4 in parallel:
+**Determine milestone context:**
+- If no "Validated" requirements in REQUIREMENTS.md → Greenfield (v1.0)
+- If "Validated" requirements exist → Subsequent milestone (v1.1+)
+
+Spawn all 4 in parallel with rich context:
```
-Task(prompt="Research stack for [domain]. Question: [question]. Context: [PROJECT.md summary].
-Write to: .planning/research/STACK.md. Use template from ~/.claude/get-shit-done/templates/research-project/STACK.md",
-subagent_type="gsd-researcher", description="Stack research")
+Task(prompt="
+
+Project Research — Stack dimension for [domain].
+
-Task(prompt="Research features for [domain]. Question: [question]. Context: [PROJECT.md summary].
-Write to: .planning/research/FEATURES.md. Use template from ~/.claude/get-shit-done/templates/research-project/FEATURES.md",
-subagent_type="gsd-researcher", description="Features research")
+
+{greenfield OR subsequent}
-Task(prompt="Research architecture for [domain]. Question: [question]. Context: [PROJECT.md summary].
-Write to: .planning/research/ARCHITECTURE.md. Use template from ~/.claude/get-shit-done/templates/research-project/ARCHITECTURE.md",
-subagent_type="gsd-researcher", description="Architecture research")
+Greenfield: Research the standard stack for building [domain] from scratch.
+Subsequent: Research what's needed to add [target features] to an existing [domain] app. Don't re-research the existing system.
+
-Task(prompt="Research pitfalls for [domain]. Question: [question]. Context: [PROJECT.md summary].
-Write to: .planning/research/PITFALLS.md. Use template from ~/.claude/get-shit-done/templates/research-project/PITFALLS.md",
-subagent_type="gsd-researcher", description="Pitfalls research")
+
+[stack question from step 4]
+
+
+
+[PROJECT.md summary]
+
+
+
+Your STACK.md feeds into /gsd:create-roadmap. Be prescriptive:
+- Specific libraries with versions
+- Clear rationale for each choice
+- What NOT to use and why
+
+
+
+- [ ] Versions are current (not Claude's training data)
+- [ ] Rationale explains WHY, not just WHAT
+- [ ] Confidence levels assigned
+
+
+
+", subagent_type="gsd-researcher", description="Stack research")
+
+Task(prompt="
+
+Project Research — Features dimension for [domain].
+
+
+
+{greenfield OR subsequent}
+
+Greenfield: What features do [domain] products have? What's table stakes vs differentiating?
+Subsequent: How do [target features] typically work? What's expected behavior?
+
+
+
+[features question from step 4]
+
+
+
+[PROJECT.md summary]
+
+
+
+Your FEATURES.md feeds into /gsd:define-requirements. Categorize clearly:
+- Table stakes (must have or users leave)
+- Differentiators (competitive advantage)
+- Anti-features (things to deliberately NOT build)
+
+
+
+- [ ] Categories are clear
+- [ ] Complexity noted for each
+- [ ] Dependencies between features identified
+
+
+
+", subagent_type="gsd-researcher", description="Features research")
+
+Task(prompt="
+
+Project Research — Architecture dimension for [domain].
+
+
+
+{greenfield OR subsequent}
+
+Greenfield: How are [domain] systems typically structured? What are major components?
+Subsequent: How do [target features] integrate with existing [domain] architecture?
+
+
+
+[architecture question from step 4]
+
+
+
+[PROJECT.md summary]
+
+
+
+Your ARCHITECTURE.md informs phase structure in roadmap. Include:
+- Component boundaries (what talks to what)
+- Data flow (how information moves)
+- Suggested build order (dependencies between components)
+
+
+
+- [ ] Components clearly defined
+- [ ] Boundaries explicit
+- [ ] Build order implications noted
+
+
+
+", subagent_type="gsd-researcher", description="Architecture research")
+
+Task(prompt="
+
+Project Research — Pitfalls dimension for [domain].
+
+
+
+{greenfield OR subsequent}
+
+Greenfield: What do [domain] projects commonly get wrong? Critical mistakes?
+Subsequent: What are common mistakes when adding [target features] to [domain]?
+
+
+
+[pitfalls question from step 4]
+
+
+
+[PROJECT.md summary]
+
+
+
+Your PITFALLS.md prevents mistakes in roadmap/planning. For each pitfall:
+- Warning signs (how to detect early)
+- Prevention strategy (how to avoid)
+- Which phase should address it
+
+
+
+- [ ] Pitfalls are specific, not generic
+- [ ] Prevention is actionable
+- [ ] Phase mapping included
+
+
+
+", subagent_type="gsd-researcher", description="Pitfalls research")
```
**Announce:** "Spawning 4 research agents... may take 2-3 minutes."
@@ -91,9 +235,35 @@ subagent_type="gsd-researcher", description="Pitfalls research")
After all agents complete, read their outputs and write `.planning/research/SUMMARY.md`:
- Read template: `~/.claude/get-shit-done/templates/research-project/SUMMARY.md`
- Synthesize executive summary from all 4 files
-- Include "Implications for Roadmap" with suggested phase structure
- Add confidence assessment
+**Critical: Include "Implications for Roadmap" section:**
+
+```markdown
+## Implications for Roadmap
+
+Based on research, suggested phase structure:
+
+1. **[Phase name]** — [rationale from research]
+ - Addresses: [features from FEATURES.md]
+ - Avoids: [pitfall from PITFALLS.md]
+ - Uses: [stack element from STACK.md]
+
+2. **[Phase name]** — [rationale from research]
+ - Implements: [architecture component from ARCHITECTURE.md]
+ ...
+
+**Phase ordering rationale:**
+- [Why this order based on dependencies discovered in ARCHITECTURE.md]
+- [Why this grouping based on PITFALLS.md prevention strategies]
+
+**Research flags for phases:**
+- Phase [X]: Likely needs deeper research (reason)
+- Phase [Y]: Standard patterns, unlikely to need research
+```
+
+This section directly feeds into `/gsd:create-roadmap`.
+
## 7. Commit Research
```bash