feat: agent skill injection via config (#1355)

Add agent_skills config section that maps agent types to skill directory paths. At spawn time, workflows load configured skills and inject them as <agent_skills> blocks in Task() prompts, giving subagents access to project-specific skill files. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-24 16:26:13 -04:00
parent 4ef309e0e4
commit db3eeb8fe4
22 changed files with 491 additions and 14 deletions
--- a/get-shit-done/workflows/audit-milestone.md
+++ b/get-shit-done/workflows/audit-milestone.md
@@ -18,6 +18,7 @@ Valid GSD subagent types (use exact names — do not fall back to 'general-purpo
 ```bash
 INIT=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" init milestone-op)
 if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+AGENT_SKILLS_CHECKER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-integration-checker 2>/dev/null)
 ```

 Extract from init JSON: `milestone_version`, `milestone_name`, `phase_count`, `completed_phases`, `commit_docs`.
@@ -78,7 +79,8 @@ Milestone Requirements:

 MUST map each integration finding to affected requirement IDs where applicable.

-Verify cross-phase wiring and E2E user flows.",
+Verify cross-phase wiring and E2E user flows.
+${AGENT_SKILLS_CHECKER}",
  subagent_type="gsd-integration-checker",
  model="{integration_checker_model}"
 )
--- a/get-shit-done/workflows/diagnose-issues.md
+++ b/get-shit-done/workflows/diagnose-issues.md
@@ -78,13 +78,19 @@ This runs in parallel - all gaps investigated simultaneously.
 </step>

 <step name="spawn_agents">
+**Load agent skills:**
+
+```bash
+AGENT_SKILLS_DEBUGGER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-debugger 2>/dev/null)
+```
+
 **Spawn debug agents in parallel:**

 For each gap, fill the debug-subagent-prompt template and spawn:

 ```
 Task(
-  prompt=filled_debug_subagent_prompt + "\n\n<files_to_read>\n- {phase_dir}/{phase_num}-UAT.md\n- .planning/STATE.md\n</files_to_read>",
+  prompt=filled_debug_subagent_prompt + "\n\n<files_to_read>\n- {phase_dir}/{phase_num}-UAT.md\n- .planning/STATE.md\n</files_to_read>\n${AGENT_SKILLS_DEBUGGER}",
  subagent_type="gsd-debugger",
  isolation="worktree",
  description="Debug: {truth_short}"
--- a/get-shit-done/workflows/discuss-phase-assumptions.md
+++ b/get-shit-done/workflows/discuss-phase-assumptions.md
@@ -66,6 +66,7 @@ Phase number from argument (required).
 ```bash
 INIT=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" init phase-op "${PHASE}")
 if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+AGENT_SKILLS_ANALYZER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-assumptions-analyzer 2>/dev/null)
 ```

 Parse JSON for: `commit_docs`, `phase_found`, `phase_dir`, `phase_number`, `phase_name`,
@@ -270,6 +271,8 @@ Return EXACTLY this structure:
 ## Needs External Research
 [Topics where codebase alone is insufficient — library version compatibility,
 ecosystem best practices, etc. Leave empty if codebase provides enough evidence.]
+
+${AGENT_SKILLS_ANALYZER}
 """)
 ```

--- a/get-shit-done/workflows/discuss-phase.md
+++ b/get-shit-done/workflows/discuss-phase.md
@@ -134,6 +134,7 @@ Phase number from argument (required).
 ```bash
 INIT=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" init phase-op "${PHASE}")
 if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+AGENT_SKILLS_ADVISOR=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-advisor 2>/dev/null)
 ```

 Parse JSON for: `commit_docs`, `phase_found`, `phase_dir`, `phase_number`, `phase_name`, `phase_slug`, `padded_phase`, `has_research`, `has_context`, `has_plans`, `has_verification`, `plan_count`, `roadmap_exists`, `planning_exists`.
@@ -498,7 +499,8 @@ After user selects gray areas in present_gray_areas, spawn parallel research age
     <project_context>{project name and brief description from PROJECT.md}</project_context>
     <calibration_tier>{resolved calibration tier: full_maturity | standard | minimal_decisive}</calibration_tier>

-     Research this gray area and return a structured comparison table with rationale.",
+     Research this gray area and return a structured comparison table with rationale.
+     ${AGENT_SKILLS_ADVISOR}",
     subagent_type="general-purpose",
     model="{ADVISOR_MODEL}",
     description="Research: {area_name}"
--- a/get-shit-done/workflows/execute-phase.md
+++ b/get-shit-done/workflows/execute-phase.md
@@ -63,6 +63,7 @@ Load all context in one call:
 ```bash
 INIT=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" init execute-phase "${PHASE_ARG}")
 if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+AGENT_SKILLS=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-executor 2>/dev/null)
 ```

 Parse JSON for: `executor_model`, `verifier_model`, `commit_docs`, `parallelization`, `branching_strategy`, `branch_name`, `phase_found`, `phase_dir`, `phase_number`, `phase_name`, `phase_slug`, `plans`, `incomplete_plans`, `plan_count`, `incomplete_count`, `state_exists`, `roadmap_exists`, `phase_req_ids`.
@@ -258,6 +259,8 @@ Execute each selected wave in sequence. Within a wave: parallel if `PARALLELIZAT
       - .claude/skills/ or .agents/skills/ (Project skills, if either exists — list skills, read SKILL.md for each, follow relevant rules during implementation)
       </files_to_read>

+       ${AGENT_SKILLS}
+
       <mcp_tools>
       If CLAUDE.md or project instructions reference MCP tools (e.g. jCodeMunch, context7,
       or other MCP servers), prefer those tools over Grep/Glob for code navigation when available.
@@ -580,6 +583,10 @@ Use AskUserQuestion to present the options.
 <step name="verify_phase_goal">
 Verify phase achieved its GOAL, not just completed tasks.

+```bash
+VERIFIER_SKILLS=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-verifier 2>/dev/null)
+```
+
 ```
 Task(
  prompt="Verify phase {phase_number} goal achievement.
@@ -588,7 +595,8 @@ Phase goal: {goal from ROADMAP.md}
 Phase requirement IDs: {phase_req_ids}
 Check must_haves against actual codebase.
 Cross-reference requirement IDs from PLAN frontmatter against REQUIREMENTS.md — every ID MUST be accounted for.
-Create VERIFICATION.md.",
+Create VERIFICATION.md.
+${VERIFIER_SKILLS}",
  subagent_type="gsd-verifier",
  model="{verifier_model}"
 )
--- a/get-shit-done/workflows/map-codebase.md
+++ b/get-shit-done/workflows/map-codebase.md
@@ -33,6 +33,7 @@ Load codebase mapping context:
 ```bash
 INIT=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" init map-codebase)
 if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+AGENT_SKILLS_MAPPER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-codebase-mapper 2>/dev/null)
 ```

 Extract from init JSON: `mapper_model`, `commit_docs`, `codebase_dir`, `existing_maps`, `has_maps`, `codebase_dir_exists`.
@@ -120,7 +121,8 @@ Write these documents to .planning/codebase/:
 - STACK.md - Languages, runtime, frameworks, dependencies, configuration
 - INTEGRATIONS.md - External APIs, databases, auth providers, webhooks

-Explore thoroughly. Write documents directly using templates. Return confirmation only."
+Explore thoroughly. Write documents directly using templates. Return confirmation only.
+${AGENT_SKILLS_MAPPER}"
 )
 ```

@@ -140,7 +142,8 @@ Write these documents to .planning/codebase/:
 - ARCHITECTURE.md - Pattern, layers, data flow, abstractions, entry points
 - STRUCTURE.md - Directory layout, key locations, naming conventions

-Explore thoroughly. Write documents directly using templates. Return confirmation only."
+Explore thoroughly. Write documents directly using templates. Return confirmation only.
+${AGENT_SKILLS_MAPPER}"
 )
 ```

@@ -160,7 +163,8 @@ Write these documents to .planning/codebase/:
 - CONVENTIONS.md - Code style, naming, patterns, error handling
 - TESTING.md - Framework, structure, mocking, coverage

-Explore thoroughly. Write documents directly using templates. Return confirmation only."
+Explore thoroughly. Write documents directly using templates. Return confirmation only.
+${AGENT_SKILLS_MAPPER}"
 )
 ```

@@ -179,7 +183,8 @@ Analyze this codebase for technical debt, known issues, and areas of concern.
 Write this document to .planning/codebase/:
 - CONCERNS.md - Tech debt, bugs, security, performance, fragile areas

-Explore thoroughly. Write document directly using template. Return confirmation only."
+Explore thoroughly. Write document directly using template. Return confirmation only.
+${AGENT_SKILLS_MAPPER}"
 )
 ```

--- a/get-shit-done/workflows/new-milestone.md
+++ b/get-shit-done/workflows/new-milestone.md
@@ -146,6 +146,9 @@ node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" commit "docs: start milesto
 ```bash
 INIT=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" init new-milestone)
 if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+AGENT_SKILLS_RESEARCHER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-project-researcher 2>/dev/null)
+AGENT_SKILLS_SYNTHESIZER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-synthesizer 2>/dev/null)
+AGENT_SKILLS_ROADMAPPER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-roadmapper 2>/dev/null)
 ```

 Extract from init JSON: `researcher_model`, `synthesizer_model`, `roadmapper_model`, `commit_docs`, `research_enabled`, `current_milestone`, `project_exists`, `roadmap_exists`, `latest_completed_milestone`, `phase_dir_count`, `phase_archive_path`.
@@ -222,6 +225,8 @@ Focus ONLY on what's needed for the NEW features.
 - .planning/PROJECT.md (Project context)
 </files_to_read>

+${AGENT_SKILLS_RESEARCHER}
+
 <downstream_consumer>{CONSUMER}</downstream_consumer>

 <quality_gate>{GATES}</quality_gate>
@@ -256,6 +261,8 @@ Synthesize research outputs into SUMMARY.md.
 - .planning/research/PITFALLS.md
 </files_to_read>

+${AGENT_SKILLS_SYNTHESIZER}
+
 Write to: .planning/research/SUMMARY.md
 Use template: ~/.claude/get-shit-done/templates/research-project/SUMMARY.md
 Commit after writing.
@@ -370,6 +377,9 @@ Task(prompt="
 - .planning/config.json
 - .planning/MILESTONES.md
 </files_to_read>
+
+${AGENT_SKILLS_ROADMAPPER}
+
 </planning_context>

 <instructions>
--- a/get-shit-done/workflows/new-project.md
+++ b/get-shit-done/workflows/new-project.md
@@ -59,6 +59,9 @@ The document should describe what you want to build.
 ```bash
 INIT=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" init new-project)
 if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+AGENT_SKILLS_RESEARCHER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-project-researcher 2>/dev/null)
+AGENT_SKILLS_SYNTHESIZER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-synthesizer 2>/dev/null)
+AGENT_SKILLS_ROADMAPPER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-roadmapper 2>/dev/null)
 ```

 Parse JSON for: `researcher_model`, `synthesizer_model`, `roadmapper_model`, `commit_docs`, `project_exists`, `has_codebase_map`, `planning_exists`, `has_existing_code`, `has_package_file`, `is_brownfield`, `needs_codebase_map`, `has_git`, `project_path`.
@@ -630,6 +633,8 @@ What's the standard 2025 stack for [domain]?
 - {project_path} (Project context and goals)
 </files_to_read>

+${AGENT_SKILLS_RESEARCHER}
+
 <downstream_consumer>
 Your STACK.md feeds into roadmap creation. Be prescriptive:
 - Specific libraries with versions
@@ -668,6 +673,8 @@ What features do [domain] products have? What's table stakes vs differentiating?
 - {project_path} (Project context)
 </files_to_read>

+${AGENT_SKILLS_RESEARCHER}
+
 <downstream_consumer>
 Your FEATURES.md feeds into requirements definition. Categorize clearly:
 - Table stakes (must have or users leave)
@@ -706,6 +713,8 @@ How are [domain] systems typically structured? What are major components?
 - {project_path} (Project context)
 </files_to_read>

+${AGENT_SKILLS_RESEARCHER}
+
 <downstream_consumer>
 Your ARCHITECTURE.md informs phase structure in roadmap. Include:
 - Component boundaries (what talks to what)
@@ -744,6 +753,8 @@ What do [domain] projects commonly get wrong? Critical mistakes?
 - {project_path} (Project context)
 </files_to_read>

+${AGENT_SKILLS_RESEARCHER}
+
 <downstream_consumer>
 Your PITFALLS.md prevents mistakes in roadmap/planning. For each pitfall:
 - Warning signs (how to detect early)
@@ -779,6 +790,8 @@ Synthesize research outputs into SUMMARY.md.
 - .planning/research/PITFALLS.md
 </files_to_read>

+${AGENT_SKILLS_SYNTHESIZER}
+
 <output>
 Write to: .planning/research/SUMMARY.md
 Use template: ~/.claude/get-shit-done/templates/research-project/SUMMARY.md
@@ -984,6 +997,8 @@ Task(prompt="
 - .planning/config.json (Granularity and mode settings)
 </files_to_read>

+${AGENT_SKILLS_ROADMAPPER}
+
 </planning_context>

 <instructions>
@@ -1078,6 +1093,8 @@ Use AskUserQuestion:
  - .planning/ROADMAP.md (Current roadmap to revise)
  </files_to_read>

+  ${AGENT_SKILLS_ROADMAPPER}
+
  Update the roadmap based on feedback. Edit files in place.
  Return ROADMAP REVISED with changes made.
  </revision>
--- a/get-shit-done/workflows/plan-phase.md
+++ b/get-shit-done/workflows/plan-phase.md
@@ -24,6 +24,9 @@ Load all context in one call (paths only to minimize orchestrator context):
 ```bash
 INIT=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" init plan-phase "$PHASE")
 if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+AGENT_SKILLS_RESEARCHER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-researcher 2>/dev/null)
+AGENT_SKILLS_PLANNER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-planner 2>/dev/null)
+AGENT_SKILLS_CHECKER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-checker 2>/dev/null)
 ```

 Parse JSON for: `researcher_model`, `planner_model`, `checker_model`, `research_enabled`, `plan_checker_enabled`, `nyquist_validation_enabled`, `commit_docs`, `text_mode`, `phase_found`, `phase_dir`, `phase_number`, `phase_name`, `phase_slug`, `padded_phase`, `has_research`, `has_context`, `has_reviews`, `has_plans`, `plan_count`, `planning_exists`, `roadmap_exists`, `phase_req_ids`.
@@ -298,6 +301,8 @@ Answer: "What do I need to know to PLAN this phase well?"
 - {state_path} (Project decisions and history)
 </files_to_read>

+${AGENT_SKILLS_RESEARCHER}
+
 <additional_context>
 **Phase description:** {phase_description}
 **Phase requirement IDs (MUST address):** {phase_req_ids}
@@ -485,6 +490,8 @@ Planner prompt:
 - {UI_SPEC_PATH} (UI Design Contract — visual/interaction specs, if exists)
 </files_to_read>

+${AGENT_SKILLS_PLANNER}
+
 **Phase requirement IDs (every ID MUST appear in a plan's `requirements` field):** {phase_req_ids}

 **Project instructions:** Read ./CLAUDE.md if exists — follow project-specific guidelines
@@ -582,6 +589,8 @@ Checker prompt:
 - {research_path} (Technical Research — includes Validation Architecture)
 </files_to_read>

+${AGENT_SKILLS_CHECKER}
+
 **Phase requirement IDs (MUST ALL be covered):** {phase_req_ids}

 **Project instructions:** Read ./CLAUDE.md if exists — verify plans honor project guidelines
@@ -628,6 +637,8 @@ Revision prompt:
 - {context_path} (USER DECISIONS from /gsd:discuss-phase)
 </files_to_read>

+${AGENT_SKILLS_PLANNER}
+
 **Checker issues:** {structured_issues_from_checker}
 </revision_context>

--- a/get-shit-done/workflows/quick.md
+++ b/get-shit-done/workflows/quick.md
@@ -118,6 +118,10 @@ If `$FULL_MODE` only:
 ```bash
 INIT=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" init quick "$DESCRIPTION")
 if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+AGENT_SKILLS_PLANNER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-planner 2>/dev/null)
+AGENT_SKILLS_EXECUTOR=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-executor 2>/dev/null)
+AGENT_SKILLS_CHECKER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-checker 2>/dev/null)
+AGENT_SKILLS_VERIFIER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-verifier 2>/dev/null)
 ```

 Parse JSON for: `planner_model`, `executor_model`, `checker_model`, `verifier_model`, `commit_docs`, `branch_name`, `quick_id`, `slug`, `date`, `timestamp`, `quick_dir`, `task_dir`, `roadmap_exists`, `planning_exists`.
@@ -327,6 +331,8 @@ Task(
 ${DISCUSS_MODE ? '- ' + QUICK_DIR + '/' + quick_id + '-CONTEXT.md (User decisions — research should align with these)' : ''}
 </files_to_read>

+${AGENT_SKILLS_PLANNER}
+
 </research_context>

 <focus>
@@ -381,6 +387,8 @@ ${DISCUSS_MODE ? '- ' + QUICK_DIR + '/' + quick_id + '-CONTEXT.md (User decision
 ${RESEARCH_MODE ? '- ' + QUICK_DIR + '/' + quick_id + '-RESEARCH.md (Research findings — use to inform implementation choices)' : ''}
 </files_to_read>

+${AGENT_SKILLS_PLANNER}
+
 **Project skills:** Check .claude/skills/ or .agents/skills/ directory (if either exists) — read SKILL.md files, plans should account for project skill rules

 </planning_context>
@@ -438,6 +446,8 @@ Checker prompt:
 - ${QUICK_DIR}/${quick_id}-PLAN.md (Plan to verify)
 </files_to_read>

+${AGENT_SKILLS_CHECKER}
+
 **Scope:** This is a quick task, not a full phase. Skip checks that require a ROADMAP phase goal.
 </verification_context>

@@ -490,6 +500,8 @@ Revision prompt:
 - ${QUICK_DIR}/${quick_id}-PLAN.md (Existing plan)
 </files_to_read>

+${AGENT_SKILLS_PLANNER}
+
 **Checker issues:** ${structured_issues_from_checker}

 </revision_context>
@@ -536,6 +548,8 @@ Execute quick task ${quick_id}.
 - .claude/skills/ or .agents/skills/ (Project skills, if either exists — list skills, read SKILL.md for each, follow relevant rules during implementation)
 </files_to_read>

+${AGENT_SKILLS_EXECUTOR}
+
 <constraints>
 - Execute all tasks in the plan
 - Commit each task atomically
@@ -586,6 +600,8 @@ Task goal: ${DESCRIPTION}
 - ${QUICK_DIR}/${quick_id}-PLAN.md (Plan)
 </files_to_read>

+${AGENT_SKILLS_VERIFIER}
+
 Check must_haves against actual codebase. Create VERIFICATION.md at ${QUICK_DIR}/${quick_id}-VERIFICATION.md.",
  subagent_type="gsd-verifier",
  model="{verifier_model}",
--- a/get-shit-done/workflows/research-phase.md
+++ b/get-shit-done/workflows/research-phase.md
@@ -42,6 +42,7 @@ If exists: Offer update/view/skip options.
 INIT=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" init phase-op "${PHASE}")
 if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
 # Extract: phase_dir, padded_phase, phase_number, state_path, requirements_path, context_path
+AGENT_SKILLS_RESEARCHER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-researcher 2>/dev/null)
 ```

 ## Step 4: Spawn Researcher
@@ -58,6 +59,8 @@ Research implementation approach for Phase {phase}: {name}
 - {state_path} (Project decisions and history)
 </files_to_read>

+${AGENT_SKILLS_RESEARCHER}
+
 <additional_context>
 Phase description: {description}
 </additional_context>
--- a/get-shit-done/workflows/ui-phase.md
+++ b/get-shit-done/workflows/ui-phase.md
@@ -21,6 +21,8 @@ Valid GSD subagent types (use exact names — do not fall back to 'general-purpo
 ```bash
 INIT=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" init plan-phase "$PHASE")
 if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+AGENT_SKILLS_UI=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-ui-researcher 2>/dev/null)
+AGENT_SKILLS_UI_CHECKER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-ui-checker 2>/dev/null)
 ```

 Parse JSON for: `phase_dir`, `phase_number`, `phase_name`, `phase_slug`, `padded_phase`, `has_context`, `has_research`, `commit_docs`.
@@ -122,6 +124,8 @@ Answer: "What visual and interaction contracts does this phase need?"
 - {research_path} (Technical Research — stack decisions)
 </files_to_read>

+${AGENT_SKILLS_UI}
+
 <output>
 Write to: {phase_dir}/{padded_phase}-UI-SPEC.md
 Template: ~/.claude/get-shit-done/templates/UI-SPEC.md
@@ -180,6 +184,8 @@ Check all 6 dimensions. Return APPROVED or BLOCKED.
 - {research_path} (Technical Research — check stack alignment)
 </files_to_read>

+${AGENT_SKILLS_UI_CHECKER}
+
 <config>
 ui_safety_gate: {ui_safety_gate config value}
 </config>
--- a/get-shit-done/workflows/ui-review.md
+++ b/get-shit-done/workflows/ui-review.md
@@ -18,6 +18,7 @@ Valid GSD subagent types (use exact names — do not fall back to 'general-purpo
 ```bash
 INIT=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" init phase-op "${PHASE_ARG}")
 if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+AGENT_SKILLS_UI_REVIEWER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-ui-reviewer 2>/dev/null)
 ```

 Parse: `phase_dir`, `phase_number`, `phase_name`, `phase_slug`, `padded_phase`, `commit_docs`.
@@ -85,6 +86,8 @@ Conduct 6-pillar visual audit of Phase {phase_number}: {phase_name}
 - {context_path} (User decisions, if exists)
 </files_to_read>

+${AGENT_SKILLS_UI_REVIEWER}
+
 <config>
 phase_dir: {phase_dir}
 padded_phase: {padded_phase}
--- a/get-shit-done/workflows/validate-phase.md
+++ b/get-shit-done/workflows/validate-phase.md
@@ -18,6 +18,7 @@ Valid GSD subagent types (use exact names — do not fall back to 'general-purpo
 ```bash
 INIT=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" init phase-op "${PHASE_ARG}")
 if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+AGENT_SKILLS_AUDITOR=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-nyquist-auditor 2>/dev/null)
 ```

 Parse: `phase_dir`, `phase_number`, `phase_name`, `phase_slug`, `padded_phase`.
@@ -95,7 +96,8 @@ Task(
    "<files_to_read>{PLAN, SUMMARY, impl files, VALIDATION.md}</files_to_read>" +
    "<gaps>{gap list}</gaps>" +
    "<test_infrastructure>{framework, config, commands}</test_infrastructure>" +
-    "<constraints>Never modify impl files. Max 3 debug iterations. Escalate impl bugs.</constraints>",
+    "<constraints>Never modify impl files. Max 3 debug iterations. Escalate impl bugs.</constraints>" +
+    "${AGENT_SKILLS_AUDITOR}",
  subagent_type="gsd-nyquist-auditor",
  model="{AUDITOR_MODEL}",
  description="Fill validation gaps for Phase {N}"
--- a/get-shit-done/workflows/verify-work.md
+++ b/get-shit-done/workflows/verify-work.md
@@ -32,6 +32,8 @@ If $ARGUMENTS contains a phase number, load context:
 ```bash
 INIT=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" init verify-work "${PHASE_ARG}")
 if [[ "$INIT" == @file:* ]]; then INIT=$(cat "${INIT#@file:}"); fi
+AGENT_SKILLS_PLANNER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-planner 2>/dev/null)
+AGENT_SKILLS_CHECKER=$(node "$HOME/.claude/get-shit-done/bin/gsd-tools.cjs" agent-skills gsd-checker 2>/dev/null)
 ```

 Parse JSON for: `planner_model`, `checker_model`, `commit_docs`, `phase_found`, `phase_dir`, `phase_number`, `phase_name`, `has_verification`, `uat_path`.
@@ -431,6 +433,8 @@ Task(
 - .planning/ROADMAP.md (Roadmap)
 </files_to_read>

+${AGENT_SKILLS_PLANNER}
+
 </planning_context>

 <downstream_consumer>
@@ -477,6 +481,8 @@ Task(
 - {phase_dir}/*-PLAN.md (Plans to verify)
 </files_to_read>

+${AGENT_SKILLS_CHECKER}
+
 </verification_context>

 <expected_output>
@@ -517,6 +523,8 @@ Task(
 - {phase_dir}/*-PLAN.md (Existing plans)
 </files_to_read>

+${AGENT_SKILLS_PLANNER}
+
 **Checker issues:**
 {structured_issues_from_checker}