Files
msd-core/get-shit-done/workflows/plan-phase.md
Lex Christopherson 8d977328e4 fix(tools): write large JSON payloads to tmpfile to prevent truncation (closes #493)
When init commands include many file contents (state, roadmap, requirements,
etc.), the JSON output can exceed Claude Code's Bash tool buffer (~50KB),
causing parse errors. The output() function now auto-detects large payloads
and writes to a tmpfile, returning @file:/path instead. All workflows that
consume init output handle both formats.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-15 14:33:50 -06:00

15 KiB

Create executable phase prompts (PLAN.md files) for a roadmap phase with integrated research and verification. Default flow: Research (if needed) -> Plan -> Verify -> Done. Orchestrates gsd-phase-researcher, gsd-planner, and gsd-plan-checker agents with a revision loop (max 3 iterations).

<required_reading> Read all files referenced by the invoking prompt's execution_context before starting.

@~/.claude/get-shit-done/references/ui-brand.md </required_reading>

1. Initialize

Load all context in one call (include file contents to avoid redundant reads):

INIT_RAW=$(node ~/.claude/get-shit-done/bin/gsd-tools.cjs init plan-phase "$PHASE" --include state,roadmap,requirements,context,research,verification,uat)
# Large payloads are written to a tmpfile — output starts with @file:/path
if [[ "$INIT_RAW" == @file:* ]]; then
  INIT_FILE="${INIT_RAW#@file:}"
  INIT=$(cat "$INIT_FILE")
  rm -f "$INIT_FILE"
else
  INIT="$INIT_RAW"
fi

Parse JSON for: researcher_model, planner_model, checker_model, research_enabled, plan_checker_enabled, commit_docs, phase_found, phase_dir, phase_number, phase_name, phase_slug, padded_phase, has_research, has_context, has_plans, plan_count, planning_exists, roadmap_exists.

File contents (from --include): state_content, roadmap_content, requirements_content, context_content, research_content, verification_content, uat_content. These are null if files don't exist.

If planning_exists is false: Error — run /gsd:new-project first.

2. Parse and Normalize Arguments

Extract from $ARGUMENTS: phase number (integer or decimal like 2.1), flags (--research, --skip-research, --gaps, --skip-verify).

If no phase number: Detect next unplanned phase from roadmap.

If phase_found is false: Validate phase exists in ROADMAP.md. If valid, create the directory using phase_slug and padded_phase from init:

mkdir -p ".planning/phases/${padded_phase}-${phase_slug}"

Existing artifacts from init: has_research, has_plans, plan_count.

3. Validate Phase

PHASE_INFO=$(node ~/.claude/get-shit-done/bin/gsd-tools.cjs roadmap get-phase "${PHASE}")

If found is false: Error with available phases. If found is true: Extract phase_number, phase_name, goal from JSON.

4. Load CONTEXT.md

Use context_content from init JSON (already loaded via --include context).

CRITICAL: Use context_content from INIT — pass to researcher, planner, checker, and revision agents.

If context_content is not null, display: Using phase context from: ${PHASE_DIR}/*-CONTEXT.md

5. Handle Research

Skip if: --gaps flag, --skip-research flag, or research_enabled is false (from init) without --research override.

If has_research is true (from init) AND no --research flag: Use existing, skip to step 6.

If RESEARCH.md missing OR --research flag:

Display banner:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 GSD ► RESEARCHING PHASE {X}
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

◆ Spawning researcher...

Spawn gsd-phase-researcher

PHASE_DESC=$(node ~/.claude/get-shit-done/bin/gsd-tools.cjs roadmap get-phase "${PHASE}" | jq -r '.section')
# Use requirements_content from INIT (already loaded via --include requirements)
REQUIREMENTS=$(echo "$INIT" | jq -r '.requirements_content // empty' | grep -A100 "## Requirements" | head -50)
STATE_SNAP=$(node ~/.claude/get-shit-done/bin/gsd-tools.cjs state-snapshot)
# Extract decisions from state-snapshot JSON: jq '.decisions[] | "\(.phase): \(.summary) - \(.rationale)"'

Research prompt:

<objective>
Research how to implement Phase {phase_number}: {phase_name}
Answer: "What do I need to know to PLAN this phase well?"
</objective>

<phase_context>
IMPORTANT: If CONTEXT.md exists below, it contains user decisions from /gsd:discuss-phase.
- **Decisions** = Locked — research THESE deeply, no alternatives
- **Claude's Discretion** = Freedom areas — research options, recommend
- **Deferred Ideas** = Out of scope — ignore

{context_content}
</phase_context>

<additional_context>
**Phase description:** {phase_description}
**Requirements:** {requirements}
**Prior decisions:** {decisions}
</additional_context>

<output>
Write to: {phase_dir}/{phase_num}-RESEARCH.md
</output>
Task(
  prompt="First, read ~/.claude/agents/gsd-phase-researcher.md for your role and instructions.\n\n" + research_prompt,
  subagent_type="general-purpose",
  model="{researcher_model}",
  description="Research Phase {phase}"
)

Handle Researcher Return

  • ## RESEARCH COMPLETE: Display confirmation, continue to step 6
  • ## RESEARCH BLOCKED: Display blocker, offer: 1) Provide context, 2) Skip research, 3) Abort

6. Check Existing Plans

ls "${PHASE_DIR}"/*-PLAN.md 2>/dev/null

If exists: Offer: 1) Add more plans, 2) View existing, 3) Replan from scratch.

7. Use Context Files from INIT

All file contents are already loaded via --include in step 1 (@ syntax doesn't work across Task() boundaries):

# Extract from INIT JSON (no need to re-read files)
STATE_CONTENT=$(echo "$INIT" | jq -r '.state_content // empty')
ROADMAP_CONTENT=$(echo "$INIT" | jq -r '.roadmap_content // empty')
REQUIREMENTS_CONTENT=$(echo "$INIT" | jq -r '.requirements_content // empty')
RESEARCH_CONTENT=$(echo "$INIT" | jq -r '.research_content // empty')
VERIFICATION_CONTENT=$(echo "$INIT" | jq -r '.verification_content // empty')
UAT_CONTENT=$(echo "$INIT" | jq -r '.uat_content // empty')
CONTEXT_CONTENT=$(echo "$INIT" | jq -r '.context_content // empty')

8. Spawn gsd-planner Agent

Display banner:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 GSD ► PLANNING PHASE {X}
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

◆ Spawning planner...

Planner prompt:

<planning_context>
**Phase:** {phase_number}
**Mode:** {standard | gap_closure}

**Project State:** {state_content}
**Roadmap:** {roadmap_content}
**Requirements:** {requirements_content}

**Phase Context:**
IMPORTANT: If context exists below, it contains USER DECISIONS from /gsd:discuss-phase.
- **Decisions** = LOCKED — honor exactly, do not revisit
- **Claude's Discretion** = Freedom — make implementation choices
- **Deferred Ideas** = Out of scope — do NOT include

{context_content}

**Research:** {research_content}
**Gap Closure (if --gaps):** {verification_content} {uat_content}
</planning_context>

<downstream_consumer>
Output consumed by /gsd:execute-phase. Plans need:
- Frontmatter (wave, depends_on, files_modified, autonomous)
- Tasks in XML format
- Verification criteria
- must_haves for goal-backward verification
</downstream_consumer>

<quality_gate>
- [ ] PLAN.md files created in phase directory
- [ ] Each plan has valid frontmatter
- [ ] Tasks are specific and actionable
- [ ] Dependencies correctly identified
- [ ] Waves assigned for parallel execution
- [ ] must_haves derived from phase goal
</quality_gate>
Task(
  prompt="First, read ~/.claude/agents/gsd-planner.md for your role and instructions.\n\n" + filled_prompt,
  subagent_type="general-purpose",
  model="{planner_model}",
  description="Plan Phase {phase}"
)

9. Handle Planner Return

  • ## PLANNING COMPLETE: Display plan count. If --skip-verify or plan_checker_enabled is false (from init): skip to step 13. Otherwise: step 10.
  • ## CHECKPOINT REACHED: Present to user, get response, spawn continuation (step 12)
  • ## PLANNING INCONCLUSIVE: Show attempts, offer: Add context / Retry / Manual

10. Spawn gsd-plan-checker Agent

Display banner:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 GSD ► VERIFYING PLANS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

◆ Spawning plan checker...
PLANS_CONTENT=$(cat "${PHASE_DIR}"/*-PLAN.md 2>/dev/null)

Checker prompt:

<verification_context>
**Phase:** {phase_number}
**Phase Goal:** {goal from ROADMAP}

**Plans to verify:** {plans_content}
**Requirements:** {requirements_content}

**Phase Context:**
IMPORTANT: Plans MUST honor user decisions. Flag as issue if plans contradict.
- **Decisions** = LOCKED — plans must implement exactly
- **Claude's Discretion** = Freedom areas — plans can choose approach
- **Deferred Ideas** = Out of scope — plans must NOT include

{context_content}
</verification_context>

<expected_output>
- ## VERIFICATION PASSED — all checks pass
- ## ISSUES FOUND — structured issue list
</expected_output>
Task(
  prompt=checker_prompt,
  subagent_type="gsd-plan-checker",
  model="{checker_model}",
  description="Verify Phase {phase} plans"
)

11. Handle Checker Return

  • ## VERIFICATION PASSED: Display confirmation, proceed to step 13.
  • ## ISSUES FOUND: Display issues, check iteration count, proceed to step 12.

12. Revision Loop (Max 3 Iterations)

Track iteration_count (starts at 1 after initial plan + check).

If iteration_count < 3:

Display: Sending back to planner for revision... (iteration {N}/3)

PLANS_CONTENT=$(cat "${PHASE_DIR}"/*-PLAN.md 2>/dev/null)

Revision prompt:

<revision_context>
**Phase:** {phase_number}
**Mode:** revision

**Existing plans:** {plans_content}
**Checker issues:** {structured_issues_from_checker}

**Phase Context:**
Revisions MUST still honor user decisions.
{context_content}
</revision_context>

<instructions>
Make targeted updates to address checker issues.
Do NOT replan from scratch unless issues are fundamental.
Return what changed.
</instructions>
Task(
  prompt="First, read ~/.claude/agents/gsd-planner.md for your role and instructions.\n\n" + revision_prompt,
  subagent_type="general-purpose",
  model="{planner_model}",
  description="Revise Phase {phase} plans"
)

After planner returns -> spawn checker again (step 10), increment iteration_count.

If iteration_count >= 3:

Display: Max iterations reached. {N} issues remain: + issue list

Offer: 1) Force proceed, 2) Provide guidance and retry, 3) Abandon

13. Present Final Status

Route to <offer_next> OR auto_advance depending on flags/config.

14. Auto-Advance Check

Check for auto-advance trigger:

  1. Parse --auto flag from $ARGUMENTS
  2. Read workflow.auto_advance from config:
    AUTO_CFG=$(node ~/.claude/get-shit-done/bin/gsd-tools.cjs config get workflow.auto_advance 2>/dev/null || echo "false")
    

If --auto flag present OR AUTO_CFG is true:

Display banner:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 GSD ► AUTO-ADVANCING TO EXECUTE
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Plans ready. Spawning execute-phase...

Spawn execute-phase as Task:

Task(
  prompt="Run /gsd:execute-phase ${PHASE}",
  subagent_type="general-purpose",
  description="Execute Phase ${PHASE}"
)

Handle execute-phase return:

  • PHASE COMPLETE → Display final summary:
    ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
     GSD ► PHASE ${PHASE} COMPLETE ✓
    ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
    
    Auto-advance pipeline finished.
    
    Next: /gsd:discuss-phase ${NEXT_PHASE} --auto
    
  • GAPS FOUND / VERIFICATION FAILED → Display result, stop chain:
    Auto-advance stopped: Execution needs review.
    
    Review the output above and continue manually:
    /gsd:execute-phase ${PHASE}
    

If neither --auto nor config enabled: Route to <offer_next> (existing behavior).

<offer_next> Output this markdown directly (not as a code block):

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ GSD ► PHASE {X} PLANNED ✓ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Phase {X}: {Name} — {N} plan(s) in {M} wave(s)

Wave Plans What it builds
1 01, 02 [objectives]
2 03 [objective]

Research: {Completed | Used existing | Skipped} Verification: {Passed | Passed with override | Skipped}

───────────────────────────────────────────────────────────────

▶ Next Up

Execute Phase {X} — run all {N} plans

/gsd:execute-phase {X}

/clear first → fresh context window

───────────────────────────────────────────────────────────────

Also available:

  • cat .planning/phases/{phase-dir}/*-PLAN.md — review plans
  • /gsd:plan-phase {X} --research — re-research first

─────────────────────────────────────────────────────────────── </offer_next>

<success_criteria>

  • .planning/ directory validated
  • Phase validated against roadmap
  • Phase directory created if needed
  • CONTEXT.md loaded early (step 4) and passed to ALL agents
  • Research completed (unless --skip-research or --gaps or exists)
  • gsd-phase-researcher spawned with CONTEXT.md
  • Existing plans checked
  • gsd-planner spawned with CONTEXT.md + RESEARCH.md
  • Plans created (PLANNING COMPLETE or CHECKPOINT handled)
  • gsd-plan-checker spawned with CONTEXT.md
  • Verification passed OR user override OR max iterations with user decision
  • User sees status between agent spawns
  • User knows next steps </success_criteria>