Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
A direct command skips the review prompt. Inspect the source before running it.
Identifies missing prerequisites (blockers) for each /architect: command based on _state.json field requirements
Blocker Detection Skill
Identifies missing prerequisites (blockers) for each /architect: command based on _state.json field requirements. Used by /architect:next-steps, /architect:check-state, and validation commands to determine which commands can run and which are blocked.
When to Use
Invoke this skill to determine if a command can execute or what prerequisites must be completed first. Blocker detection:
Prevents wasted time on commands that will fail due to missing inputs
Shows users exact remediation steps ("Run X first, then retry Y")
{"blocker_id":"B-1xx","severity":"critical","type":"missing_input","field":"solution.sdl.yaml OR _state.json","reason":"Cannot proceed without architecture definition","fix":"Run /architect:blueprint or /architect:sdl first","fix_eta_minutes":20}
Commands affected:
Any command that reads SDL (scaffold, design-system, generate-data-model, etc.)
Any command that depends on state (check-state, next-steps, production-readiness, etc.)
2. Missing Scaffold (CRITICAL for code-generation commands)
Blocker structure:
{"blocker_id":"B-2xx","severity":"critical","type":"missing_scaffold","field":"components[].name","reason":"Command needs scaffolded components to generate code for","fix":"Run /architect:scaffold first","fix_eta_minutes":15}
Commands affected:
generate-tests (needs components to test)
security-scan (needs code to scan)
setup-monitoring (needs service structure to instrument)
generate-docs (needs files to document)
accessibility-audit (needs rendered components)
3. Missing Data Model (MEDIUM for features that need entity definitions)
Blocker structure:
{"blocker_id":"B-3xx","severity":"medium","type":"missing_data_model","field":"entities","reason":"Command generates tests/docs that reference entity types","fix":"Run /architect:generate-data-model first","fix_eta_minutes":10}
Commands affected:
generate-tests (better test fixtures with entity schema)
generate-data-model (needs entities first)
load-test (needs entity sizes for payload generation)
4. Missing Design Tokens (MEDIUM for UI commands)
Blocker structure:
{"blocker_id":"B-4xx","severity":"medium","type":"missing_design_tokens","field":"design (primary color, fonts, etc.)","reason":"Command needs design direction to apply tokens consistently","fix":"Run /architect:design-system first","fix_eta_minutes":15}
Commands affected:
scaffold-component (applies design tokens to new components)
prototype (needs design palette)
wireframes (needs color/font specs)
5. Missing Blueprint (LOW-MEDIUM for planning commands)
Blocker structure:
{"blocker_id":"B-5xx","severity":"low","type":"missing_blueprint","field":"blueprint","reason":"Blueprint provides architecture decisions and patterns","fix":"Run /architect:blueprint first","fix_eta_minutes":20}
Commands affected:
cost-estimate (better estimates with architecture detail)
load-test (needs service interaction diagram)
technical-roadmap (needs architecture context)
6. Missing Compliance/Monitoring (LOW-MEDIUM for launch gates)
Blocker structure:
{"blocker_id":"B-6xx","severity":"medium","type":"missing_compliance","field":"monitoring OR compliance","reason":"Launch gates require observability and regulatory sign-off","fix":"Run /architect:setup-monitoring and /architect:compliance first","fix_eta_minutes":60}
Commands affected:
launch-check (gates on monitoring + compliance)
production-readiness (gates on enterprise readiness)
Blocker Severity Levels
Level
Meaning
Action
critical
Command CANNOT run at all without this prerequisite
Block immediately, show remediation path
medium
Command can run with degraded output; results less useful without this
Warn user, allow proceed with caveats
low
Command runs fully; output would be richer with this
Cascading effects (what else becomes unblocked after this)
Commands and Their Blockers
Blueprint / Architecture Commands
Command
Blockers
Severity
Fix
blueprint
None (can run on empty state)
—
—
sdl
None (can run on empty state)
—
—
design-system
design field from blueprint
low
Run /architect:blueprint first
generate-data-model
entities in SDL; components in scaffold
medium
Run /architect:scaffold and provide entities in SDL
visualise
SDL or state file
critical
Run /architect:blueprint or /architect:sdl
Code Generation Commands
Command
Blockers
Severity
Fix
scaffold
SDL (solution.sdl.yaml or state)
critical
Run /architect:blueprint first
scaffold-component
Existing scaffold
critical
Run /architect:scaffold first
implement
Existing scaffold
critical
Run /architect:scaffold first
generate-tests
Components (scaffold) + test framework choice
critical
Run /architect:scaffold first
review
Source files
critical
Run /architect:scaffold first
Production & Observability Commands
Command
Blockers
Severity
Fix
setup-monitoring
Service structure (scaffold)
critical
Run /architect:scaffold first
setup-cicd
Scaffold + git repo
critical
Run /architect:scaffold and git init
compliance
Scaffold (for code scanning)
medium
Run /architect:scaffold first (can scan SDL for policy gaps)
security-scan
Source files
medium
Run /architect:scaffold first
load-test
Scaffold + service endpoints
medium
Run /architect:scaffold and /architect:implement
launch-check
Scaffold + build artifacts
critical
Run /architect:scaffold and build locally
production-readiness
Monitoring + compliance + tests
critical
Run monitoring, compliance, and generate-tests first
Documentation & Planning Commands
Command
Blockers
Severity
Fix
generate-docs
Scaffold (optional)
low
Run /architect:scaffold for better docs, but can generate from SDL alone
cost-estimate
Tech stack + architecture details
low
Run /architect:blueprint for better estimates
technical-roadmap
Blueprint
low
Run /architect:blueprint first
user-journeys
Personas + use cases
low
Provide in SDL or use /architect:user-personas
problem-validation
Problem statement
low
Define in SDL
Usage in Commands
From /architect:next-steps:
for each candidate_command in [top_10_commands]:
blockers = blocker_detection(candidate, state)
if blockers.can_execute == false:
score -= dependency_cost(blockers.blocking_count)
show blocker count in recommendation
show unblock_eta in time estimate
else:
score += points (command is ready)
From /architect:check-state:
for each command:
blockers = blocker_detection(command, state)
if blockers.critical_count > 0:
report "Command X blocked by Y. Run Z to unblock."
From user error handling:
if command fails with "missing prerequisite":
blockers = blocker_detection(command, state)
show user: "I need X to continue. Run Y first (est. N minutes). See dependency chain above."
Edge Cases
Command is blocked by itself
Example:scaffold-component exists but no initial scaffold
{"command":"scaffold-component","can_execute":false,"blockers":[{"reason":"Need initial scaffold with base structure first","fix":"Run /architect:scaffold to create the initial project structure"}]}
Multiple independent paths to unblock
Example:production-readiness needs both monitoring AND compliance
{"command":"production-readiness","can_execute":false,"blockers":[{"fix":"Run /architect:setup-monitoring (30 min)"},{"fix":"Run /architect:compliance (20 min)"}],"note":"Paths are independent; can run in parallel"}
Optional blocker
Example:generate-tests works without entities, but better with them
{"command":"generate-tests","can_execute":true,"blockers":[{"severity":"low","type":"missing_entities","reason":"Tests will be generic without entity type information","fix":"Run /architect:generate-data-model for richer fixtures"}],"note":"Can proceed, but recommend completing blocker first"}
Implementation Notes
Fast lookup: Blockers are statically defined per command (not computed from graph analysis)
No cascades beyond 3 levels: If unblocking X requires Y which requires Z, show Z but don't compute full graph beyond 3 levels
Optimistic: if a command might work without a blocker (e.g., generate-tests can work with just SDL), mark blocker as "can_skip": true with a warning
Audit trail: every blocker detection gets a timestamp and can be logged to _activity.jsonl for investigation
Related Commands
/architect:next-steps — uses blockers to calculate recommendation scores and unblock ETAs
/architect:check-state — uses blockers to recommend prerequisite commands
/architect:production-readiness — uses blockers to gate launch (critical blockers = launch not ready)
skills/prerequisite-graph — visualizes full command dependency topology
