Skip to main content
generate-verified-docs Generates audit-grade, hallucination-free documentation from codebase features with full traceability. Use when documenting features, creating technical specs, or generating verified documentation with code references.
Aller à l'installation Skills Marketplace Découvrez et explorez les compétences IA créées par la communauté.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Afficher les détails du prompt Une commande directe contourne le prompt de vérification. Examinez la source avant de l'exécuter.
npx skills add https://github.com/quran/quran.com-frontend-next --skill generate-verified-docsLa commande reste sur une seule ligne. Faites défiler horizontalement pour la vérifier avant de la copier.
Vous préférez une copie locale ? Téléchargez les fichiers actuellement disponibles dans SkillsMP.
Télécharger Zip Téléchargement... Plus depuis ce dépôt Reviews PR(s) using comprehensive review guidelines including security, correctness, clean code, TypeScript, React patterns, i18n/RTL, performance, and accessibility. Use when reviewing pull requests or code changes.
Generates a comprehensive testing plan based on the current branch changes or a specific PR. Use when creating QA checklists, test plans, or verifying PR readiness.
Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.
Basé sur la classification professionnelle SOC
name generate-verified-docs description Generates audit-grade, hallucination-free documentation from codebase features with full traceability. Use when documenting features, creating technical specs, or generating verified documentation with code references.
Verified Documentation Generator
Generate audit-grade documentation from a codebase feature with full traceability and zero
hallucination.
Arguments
$ARGUMENTS should be:
Feature name and description (e.g., "User Registration", "Bookmark System", "Reading Progress")
Prerequisites
Clear feature name and description from user
Access to the codebase root directory
Hard Rules
❌ No assumptions or inferred behavior
❌ No undocumented states or transitions
✅ Every claim must link to file:line
✅ Unknowns must be explicitly reported
✅ Halt if verification fails
Stage 1: Scope Resolution
Goal : Define feature boundaries and reject ambiguous scopes.
Parse the feature description to identify expected entry points (API routes, UI components, CLI
commands)
Search codebase for matching routes, exported functions/classes, and test files
Validate at least one concrete code location exists
Gate :
If no code locations found → HALT with failure report
If scope is ambiguous → Ask user to narrow scope
If confirmed → Proceed with entry_point_candidates list
Output : scope_definition, entry_point_candidates, scope_confidence
Stage 2: Codebase Discovery
Goal : Map all code artifacts related to the feature.
Verify each entry point exists and classify as: API | UI | JOB | EVENT | CLI
Trace dependencies (imports, injected dependencies, runtime-resolved modules)
Build dependency graph (DAG)
Classify artifacts: Controllers, Services, Repositories, Models, Utils, External
Extract data models with types, relationships, and validation rules
Every entry point must be verifiable (file_exists, line_exists, callable)
If no verifiable entry points → HALT
Output : entry_points, dependency_graph, artifacts_by_type, data_models,
external_integrations, discovery_gaps
Stage 3: Execution Flow Tracing Goal : Trace complete execution paths from entry to exit.
Synchronous paths : Follow function call chains, record decision branches, track variable
transformationsAsync boundaries : Identify Promise/async-await, event emitters, message queues, webhooksSide effects : Catalog DATABASE_WRITE, DATABASE_READ, EXTERNAL_CALL, FILE_SYSTEM, CACHE_OP,
EVENT_EMIT, LOGGINGFlow enumeration : Build all paths, tag as HAPPY_PATH | ERROR_PATH | EDGE_CASE
Every flow step must have a code reference
If coverage below threshold → Ask user to accept partial or HALT
Output : execution_flows, async_boundaries, side_effects, flow_coverage,
untraced_branches
Stage 4: State & Transition Extraction Goal : Extract explicit state machines; reject inferred states.
Search for status/state enums, FSM patterns, status columns in schemas
Map transitions: STATE_A --[condition]--> STATE_B @ file:line
Verify each state value exists in code (enum, constant, or literal)
Identify orphan states (no inbound) and terminal states (no outbound)
Every state must have a code reference
Every transition must link verified states
If undocumented states found → HALT with failure report
Output : states, transitions, state_machine, orphan_states, terminal_states
Stage 5: Failure & Edge Case Analysis Goal : Identify error handling coverage and missing failure modes.
Locate try/catch blocks, error callbacks, error boundaries, middleware handlers
Trace exception propagation paths
Detect edge cases: NULL_INPUT, EMPTY_ARRAY, BOUNDARY_VALUES, PERMISSION, TIMEOUT, CONFLICT
Score risk: LOW | MEDIUM | HIGH | CRITICAL
No hard gate – outputs are informational.
Output : handled_failures, unhandled_failures, silent_failures, edge_cases, risk_areas
Stage 6: Diagram Generation Goal : Produce diagrams strictly matching verified flows.
Generate State Diagram (Mermaid stateDiagram-v2) from state_machine
Generate Sequence Diagrams (Mermaid sequenceDiagram) from execution_flows
Generate Dependency Diagram (Mermaid flowchart) from dependency_graph
Cross-check every diagram element against source stage outputs
Remove any unverified elements
Log warnings for removed elements
Output : state_diagram, sequence_diagrams, dependency_diagram, diagram_verification_log
Stage 7: Documentation Assembly Goal : Produce structured documentation.
Generate markdown with these sections:
# [Feature Name] Documentation
## Overview
## Entry Points
## Data Models
## Execution Flows
## State Machine
## Error Handling
## Known Gaps & Limitations
## Appendix: Code References
Every claim gets [ref: file:line] annotation
Build clickable traceability index
Generate two views: Technical (full detail) and Non-Technical (simplified)
Output : documentation_technical, documentation_summary, traceability_index,
confidence_report
Stage 8: Verification & Validation Goal : Cross-check all claims; fail if verification cannot pass.
Extract all factual claims from documentation
Verify each claim against traceability_index:
File exists
Line content matches claim basis
Check diagram-flow consistency
Verify all gaps appear in Known Limitations
Hallucination detection : Flag claims without code references or contradicting code
If hallucination detected → HALT immediately
If too many unverifiable claims → HALT
If passed → Output final documentation
Output : verification_passed, verified_claims, failed_claims, hallucination_report,
final_documentation or failure_report
Failure Behavior When any stage encounters a blocking failure:
STOP execution immediately
PRESERVE all partial outputs
Generate failure report:
{
"stage_failed" : "<stage_name>" ,
"failure_type" : "<failure_enum>" ,
"failure_reason" : "<explanation>" ,
"partial_outputs" : { ... } ,
"recovery_suggestions" : [ ...]
}
AWAIT user intervention
Example Usage User: Generate documentation for the "User Registration" feature
Stage 1: Found POST /auth/register, /signup page
Stage 2: Traced AuthController → AuthService → UserRepository, EmailService
Stage 3: Mapped registration flow with email verification async boundary
Stage 4: Extracted UserStatus enum { PENDING, ACTIVE, SUSPENDED }
Stage 5: Identified missing timeout handling on EmailService
Stage 6: Generated state diagram and sequence diagram
Stage 7: Assembled full technical documentation
Stage 8: Verified 24/24 claims, no hallucinations → SUCCESS
