icdev 0.0.3__py3-none-any.whl

args/project_defaults.yaml

@@ -0,0 +1,235 @@
+# SPARKPILOT Project Defaults
+# Change behavior without editing goals or tools
+
+project:
+  classification: "CUI"
+  distribution: "Distribution D: Authorized DoD Personnel Only"
+  cui_category: "CTI"
+  default_backend: "python"
+  default_frontend: "javascript"
+  default_database: "postgresql"
+  supported_languages:
+    - "python"
+    - "java"
+    - "javascript"
+    - "typescript"
+    - "go"
+    - "rust"
+    - "csharp"
+  language_registry_path: "context/languages/language_registry.json"
+
+tdd:
+  test_framework: "pytest"
+  bdd_framework: "behave"
+  feature_format: "gherkin"
+  coverage_threshold: 80
+  require_tests_first: true
+
+compliance:
+  framework: "NIST 800-53"
+  revision: "Rev 5"
+  impact_level: "Moderate"
+  sbom_format: "cyclonedx"
+  stig_severity_threshold: "CAT2"
+
+security:
+  sast_tools:
+    - "bandit"
+    - "eslint-plugin-security"
+  dependency_audit_tools:
+    - "pip-audit"
+    - "npm-audit"
+  secret_detection: "detect-secrets"
+  container_scanner: "trivy"
+  max_critical_vulns: 0
+  max_high_vulns: 0
+
+infrastructure:
+  cloud: "aws-govcloud"
+  region: "us-gov-west-1"
+  orchestrator: "kubernetes"
+  iac_tool: "terraform"
+  config_tool: "ansible"
+  secrets_manager: "aws-secrets-manager"
+  container_registry: "ecr"
+
+cicd:
+  platform: "gitlab"
+  stages:
+    - "lint"
+    - "test"
+    - "security-scan"
+    - "build"
+    - "compliance-check"
+    - "deploy-staging"
+    - "deploy-prod"
+  require_review: true
+  auto_rollback: true
+  rollback_on:
+    - "health_check_fail"
+    - "cat1_finding"
+    - "critical_vuln"
+
+monitoring:
+  log_aggregator:
+    - "elk"
+    - "splunk"
+  metrics: "prometheus"
+  dashboards: "grafana"
+  alert_channels:
+    - "slack"
+    - "email"
+  self_healing: true
+
+cssp:
+  framework: "DoD Instruction 8530.01"
+  auto_generate: true
+  siem_targets:
+    - "splunk"
+    - "elk"
+  vuln_scan_cadence_days: 30
+  ir_reporting_timelines:
+    critical: 1    # hours
+    high: 24
+    moderate: 72
+  certification_validity_years: 3
+  pki_required: true
+
+xacta:
+  api_base_url: "https://xacta.govcloud.local/api/v1"
+  auth_method: "pki"
+  client_cert_secret: "xacta/client-cert"
+  client_key_secret: "xacta/client-key"
+  ca_bundle_secret: "xacta/ca-bundle"
+  sync_mode: "hybrid"
+  export_format: "oscal"
+  export_dir: "compliance/xacta-exports"
+  auto_sync: true
+  sync_on_events:
+    - ssp_generated
+    - poam_generated
+    - stig_checked
+    - cssp_assessed
+    - sbom_generated
+
+sbd:
+  framework: "CISA Secure by Design"
+  dodi_reference: "DoDI 5000.87"
+  auto_assess: true
+  cisa_commitments_tracked: 7
+  memory_safe_languages:
+    - "rust"
+    - "go"
+    - "java"
+    - "python"
+    - "csharp"
+  required_security_headers:
+    - "Content-Security-Policy"
+    - "Strict-Transport-Security"
+    - "X-Frame-Options"
+    - "X-Content-Type-Options"
+  threat_model_methodology: "STRIDE"
+  sbom_freshness_days: 30
+  default_credential_patterns:
+    - "admin/admin"
+    - "root/root"
+    - "password123"
+
+ivv:
+  framework: "IEEE 1012-2016"
+  auto_assess: true
+  independence_authority: "SPARKPILOT IV&V Engine"
+  min_test_coverage: 80
+  rtm_required: true
+  process_areas_tracked: 9
+  certification_validity_months: 12
+  tdd_strict: true
+  require_security_tests: true
+
+maintenance:
+  config_path: "args/maintenance_config.yaml"
+  auto_audit: true
+  audit_cadence_days: 7
+  sla_enforcement: true
+  auto_remediate: true
+  auto_remediate_max_severity: "medium"
+  report_format: "markdown"
+  cui_markings: true
+  trend_lookback_days: 90
+
+mbse:
+  xmi_import_dir: "./models/xmi"
+  reqif_import_dir: "./models/reqif"
+  export_dir: "./models/export"
+  auto_sync: false
+  sync_on_build: true
+  default_language: "python"
+  digital_thread_auto_link: true
+  pi_snapshot_on_pi_close: true
+  des_assessment_frequency: "per_pi"
+  des_gate_critical: true
+  thread_coverage_warn_threshold: 0.60
+
+modernization:
+  supported_legacy_languages:
+    - "python"
+    - "java"
+    - "csharp"
+  supported_legacy_frameworks:
+    python: ["django-1", "flask-0"]
+    java: ["struts", "ejb", "jsp", "spring-3"]
+    csharp: ["wcf", "webforms", "net-framework-4"]
+  target_frameworks:
+    python: ["django-4", "flask-3", "fastapi"]
+    java: ["spring-boot-3"]
+    csharp: ["aspnet-core", "blazor", "grpc"]
+  target_databases: ["postgresql", "aurora"]
+  default_migration_approach: "strangler_fig"
+  default_target_architecture: "microservices"
+  seven_r_weights:
+    technical_fitness: 0.25
+    business_value: 0.20
+    risk: 0.20
+    cost: 0.15
+    ato_impact: 0.10
+    timeline: 0.10
+  compliance_coverage_threshold: 0.95
+  pi_gate_required: true
+  read_only_analysis: true
+  max_complexity_threshold: 50
+  tech_debt_productivity_rate: 20
+
+agents:
+  bedrock_model: "anthropic.claude-3-5-sonnet-20241022-v2:0"
+  bedrock_region: "us-gov-west-1"
+  max_concurrent_tasks: 10
+  task_timeout_minutes: 30
+  heartbeat_interval_seconds: 30
+
+# Phase 19: Agentic generation defaults
+agentic:
+  default_port_offset: 1000
+  default_cloud_provider: aws
+  default_cloud_region: us-gov-west-1
+  default_impact_level: IL4
+  core_agents:
+    - orchestrator
+    - architect
+    - builder
+    - knowledge
+    - monitor
+  conditional_agents:
+    compliance:
+      trigger: ato_required
+    security:
+      trigger: overall_score >= 5
+  essential_goals:
+    - build_app
+    - tdd_workflow
+    - compliance_workflow
+    - security_scan
+    - deploy_workflow
+    - monitoring
+    - self_healing
+    - agent_management
+  grandchild_prevention: true

args/prompt_chains.yaml

@@ -0,0 +1,163 @@
+# Prompt Chain Templates — LLM-to-LLM sequential reasoning
+# Each chain defines a series of steps where each agent processes
+# the previous agent's output through its domain lens.
+#
+# Variable substitution:
+#   $INPUT    — previous step's output (first step gets the user's input)
+#   $ORIGINAL — the initial user input (available in every step)
+#   $STEP{step_id} — output from a specific named step
+#
+# Decision D-PC-1: YAML-driven prompt chains — add new chains without code changes (D26 pattern)
+# Decision D-PC-2: LLM reasoning chains, not tool execution — uses LLMRouter.invoke(), not A2A dispatch
+# Decision D-PC-3: Sequential execution only — no DAG parallelism (that's team_orchestrator.py's job)
+
+prompt_chains:
+  plan_critique_refine:
+    description: "Iterative planning with adversarial review"
+    max_iterations: 1
+    steps:
+      - step_id: initial_plan
+        agent: architect
+        prompt: |
+          Plan the implementation for the following requirement.
+          Output a structured plan with phases, components, and risks.
+
+          Requirement: $INPUT
+        timeout_s: 120
+      - step_id: compliance_critique
+        agent: compliance
+        prompt: |
+          Review this implementation plan for NIST 800-53 compliance gaps,
+          FedRAMP requirements, and security control coverage.
+          List specific gaps with severity (critical/high/medium/low).
+
+          Plan to review:
+          $INPUT
+
+          Original requirement: $ORIGINAL
+        timeout_s: 90
+      - step_id: security_critique
+        agent: security
+        prompt: |
+          Review this implementation plan for security vulnerabilities,
+          attack surface concerns, and OWASP risks.
+          List specific concerns with severity.
+
+          Plan to review:
+          $INPUT
+
+          Original requirement: $ORIGINAL
+        timeout_s: 90
+      - step_id: refined_plan
+        agent: architect
+        prompt: |
+          Revise your plan addressing ALL compliance and security gaps below.
+          For each gap, explain how the revised plan addresses it.
+
+          Compliance gaps:
+          $STEP{compliance_critique}
+
+          Security concerns:
+          $STEP{security_critique}
+
+          Original requirement: $ORIGINAL
+        timeout_s: 120
+
+  scout_analyze_recommend:
+    description: "Scout codebase, analyze patterns, recommend improvements"
+    steps:
+      - step_id: scout
+        agent: knowledge
+        prompt: |
+          Analyze the following codebase area for patterns, anti-patterns,
+          technical debt, and architectural concerns.
+
+          Area to analyze: $INPUT
+        timeout_s: 90
+      - step_id: analyze
+        agent: architect
+        prompt: |
+          Given this codebase analysis, identify the top 5 improvement
+          opportunities ranked by impact and effort.
+
+          Analysis: $INPUT
+          Original request: $ORIGINAL
+        timeout_s: 90
+      - step_id: recommend
+        agent: builder
+        prompt: |
+          For each improvement opportunity, provide a concrete implementation
+          plan with specific files, functions, and code changes needed.
+
+          Opportunities: $INPUT
+          Original request: $ORIGINAL
+        timeout_s: 120
+
+  security_review_chain:
+    description: "Multi-perspective security review"
+    steps:
+      - step_id: threat_model
+        agent: security
+        prompt: |
+          Perform STRIDE threat modeling on the following component.
+          Identify threats, attack vectors, and mitigations.
+
+          Component: $INPUT
+        timeout_s: 120
+      - step_id: compliance_check
+        agent: compliance
+        prompt: |
+          Map the identified threats to NIST 800-53 controls.
+          Identify any controls not adequately addressed.
+
+          Threat model: $INPUT
+          Component: $ORIGINAL
+        timeout_s: 90
+      - step_id: remediation_plan
+        agent: builder
+        prompt: |
+          Generate a remediation implementation plan addressing
+          all identified threats and compliance gaps.
+
+          Compliance gaps: $INPUT
+          Threat model: $STEP{threat_model}
+          Component: $ORIGINAL
+        timeout_s: 120
+
+  build_review_iterate:
+    description: "Build code then review and iterate"
+    steps:
+      - step_id: initial_build
+        agent: builder
+        prompt: |
+          Implement the following feature using TDD (RED-GREEN-REFACTOR).
+          Include test code and implementation code.
+
+          Feature: $INPUT
+        timeout_s: 180
+      - step_id: code_review
+        agent: architect
+        prompt: |
+          Review this code for architecture, patterns, maintainability,
+          and adherence to SOLID principles. Provide specific feedback.
+
+          Code: $INPUT
+          Feature requirement: $ORIGINAL
+        timeout_s: 90
+      - step_id: final_build
+        agent: builder
+        prompt: |
+          Revise the implementation addressing ALL review feedback.
+          Show the final, production-ready code.
+
+          Review feedback: $INPUT
+          Original code: $STEP{initial_build}
+          Feature: $ORIGINAL
+        timeout_s: 180
+
+# Default settings
+defaults:
+  timeout_s: 120
+  model_effort: high
+  record_chain: true
+  audit_trail: true

args/rag_config.yaml

@@ -0,0 +1,167 @@
+# CUI // SP-CTI
+# Phase 64 — Universal RAG Subsystem Configuration
+# All RAG settings in one place. Follows D26 declarative config pattern.
+
+rag:
+  enabled: true
+
+  vector_store:
+    backend: auto                    # auto, sqlite, chromadb, faiss
+    sqlite:
+      db_path: data/rag/rag_vectors.db
+    chromadb:
+      persist_dir: data/rag/chromadb/
+    faiss:
+      index_dir: data/rag/faiss/
+      index_type: IVF256_HNSW32
+
+  embedding:
+    batch_size: 20
+    dimensions: 768                  # nomic-embed-text via Ollama
+    cache_ttl_seconds: 300
+
+  chunking:
+    short_threshold_tokens: 500      # < 500 tok: store whole
+    chunk_size_tokens: 2000          # > 2000 tok: sliding window
+    overlap_pct: 0.10                # 10% overlap at sentence boundaries
+
+  retrieval:
+    vector_top_k: 50                 # Initial vector retrieval count
+    final_top_k: 5                   # After re-ranking
+    fusion_method: rrf               # rrf or weighted_sum (D-RAG-19)
+    rrf_k: 60                        # RRF constant (higher = more weight to top ranks)
+    bm25_boost_weight: 0.3           # BM25 keyword boost weight (used by weighted_sum)
+    time_decay_enabled: true
+    min_score_threshold: 0.1         # Minimum score to include in results
+
+  rerank:
+    enabled: true
+    method: bge                      # bge, llm, or hybrid (bge→llm fallback) (D-RAG-20)
+    bge_model: bge-reranker-v2-m3   # Ollama cross-encoder model
+    rerank_weight: 0.6               # Weight of rerank score in final blend
+    llm_model: qwen3-local           # Fallback LLM reranker (scanner_function)
+    max_chunk_preview_chars: 400     # Truncate chunk content for re-rank prompt
+
+  injection:
+    enabled: true
+    injection_top_k: 5              # How many chunks to inject into LLM context
+    max_injection_chars: 4000       # Max total chars of injected RAG context
+    injection_position: system      # Inject into system prompt
+    citation_enabled: true           # Enable [SOURCE-N] tagging (D-RAG-21)
+    citation_instruction: true       # Append citation instruction to system prompt
+    citation_validation: true        # Post-generation validation logging
+    function_denylist:
+      - attachment_analysis          # Vision tasks don't benefit from text RAG
+
+  ingestion:
+    realtime_sources:
+      - innovation_signals
+      - innovation_solutions
+      - creative_pain_points
+      - creative_feature_gaps
+      - creative_specs
+      - research_dossiers
+      - research_challenges
+      - research_forecasts
+      - compliance_artifacts
+      - memory_entries
+      - atlas_critique_sessions      # ATLAS-CR critique findings
+      # GovProposal (Phase 59/60/62)
+      - proposal_knowledge_base      # Reusable content blocks
+      - proposal_section_drafts      # HITL-approved drafts only
+      - rfp_shall_statements         # Extracted requirements
+      - cpmp_deliverables            # Contract deliverables
+    batch_sources:
+      - audit_trail
+      - ai_telemetry
+      - mbse_model_elements
+      - supply_chain_dependencies
+      - agent_memory
+      - code_quality_metrics
+      - innovation_trends
+      - creative_trends
+      # GovProposal batch (Phase 59/60/62)
+      - rfp_requirement_patterns     # Clustered requirement patterns
+      - sam_gov_opportunities        # SAM.gov opportunities
+      - proposal_compliance_matrix   # Compliance coverage
+      - govcon_awards                # Competitive intelligence
+      - cpmp_contracts               # Contract records
+    batch_schedule_minutes: 60
+    dedup_enabled: true              # SHA-256 content hash dedup (D-RAG-5)
+
+  retention:
+    hot_days: 30                     # Full float32 embeddings, <50ms
+    warm_days: 365                   # float16 compressed, ~200ms
+    warm_compression: float16
+    # cold: 365+ days, metadata only, re-embed on demand
+
+  provenance:
+    enabled: true
+    record_query_content: false      # D282 content tracing policy
+    # Set ICDEV_CONTENT_TRACING_ENABLED=true for plaintext query logging
+
+  child_app:
+    copy_to_children: true           # Include tools/rag/ in child app generation
+    parent_cache_ttl_seconds: 3600   # Child cache TTL for parent RAG queries
+    critique_history_federation: true # Child apps can query parent critique findings
+
+  critique_integration:
+    enabled: true                    # RAG-augmented ATLAS-CR critiques
+    source_types:                    # Source types to search for critique context
+      - atlas_critique_sessions
+      - compliance_artifacts
+      - innovation_signals
+    max_critique_context_chars: 3000 # Max RAG context for critique phase
+
+  # --- D-RAG-15: PDF Multi-Provider Ingestion ---
+  pdf:
+    enabled: true
+    provider_chain:                  # Fallback order (D-RAG-15)
+      - anthropic
+      - google
+      - vision_llava
+      - pypdf_text
+    max_pages: 200
+    max_file_size_mb: 50
+    extract_images: false            # Extract embedded images (future)
+    anthropic:
+      api_key: "env:ANTHROPIC_API_KEY"
+      model: claude-sonnet-4-20250514
+    google:
+      api_key: "env:GOOGLE_API_KEY"
+      model: gemini-2.0-flash
+    vision_llava:
+      model: llava:latest
+
+  # --- D-RAG-16: Intelligence Mode Auto-Indexing ---
+  auto_indexer:
+    enabled: false                   # Opt-in (user must explicitly enable)
+    watch_extensions:
+      - .md
+      - .txt
+      - .py
+      - .yaml
+      - .yml
+      - .json
+      - .pdf
+    excluded_dirs:
+      - .git
+      - node_modules
+      - __pycache__
+      - .venv
+      - .tmp
+    max_file_size_mb: 10
+    cui_boundary_enforcement: true   # Block files above project classification
+    scan_interval_seconds: 300       # 5 minutes between scans
+
+  # --- D-RAG-22: RAGAS Evaluation ---
+  evaluation:
+    enabled: true
+    test_set_path: data/rag/evaluation_set.json
+    metrics:
+      - ndcg_at_k
+      - mrr
+      - context_precision
+      - faithfulness
+      - answer_relevancy
+    llm_judge_model: qwen3-local     # scanner_function for LLM metrics

args/research_config.yaml

@@ -0,0 +1,89 @@
+# CUI // SP-CTI
+# Research Engine Configuration — sparkpilot
+# Controls vertical definitions, scoring weights, scan limits, and cache TTL.
+
+verticals:
+  defense:
+    name: "Defense & Intelligence"
+    slug: defense
+    description: "Defense software factories, DevSecOps platforms, and compliance automation tools"
+    keywords:
+      - software_factory
+      - devsecops
+      - compliance_automation
+      - cato
+      - ato
+      - cmmc
+      - fedramp
+      - sbom
+      - zero_trust
+      - mbse
+      - digital_thread
+      - air_gap
+      - platform_one
+      - iron_bank
+    regulatory_bodies:
+      - NIST
+      - DoD CIO
+      - CMMC-AB
+      - FedRAMP PMO
+      - OMB
+      - CISA
+      - DISA
+      - DoDI
+    focus_areas:
+      - ai_powered_development
+      - automated_compliance
+      - supply_chain_security
+      - zero_trust_architecture
+      - embedded_firmware
+
+  cybersecurity:
+    name: "Cybersecurity"
+    slug: cybersecurity
+    focus_areas:
+      - siem_soar
+      - threat_intelligence
+      - vulnerability_management
+
+  fintech:
+    name: "Financial Technology"
+    slug: fintech
+    focus_areas:
+      - payment_processing
+      - regulatory_compliance
+      - fraud_detection
+
+scoring:
+  weights:
+    market_demand: 0.25
+    regulatory_pressure: 0.20
+    technical_complexity: 0.15
+    competitive_saturation: 0.15
+    icdev_readiness: 0.15
+    compliance_alignment: 0.10
+
+  severity_thresholds:
+    critical: 0.75
+    notable: 0.50
+
+build_buy:
+  weights:
+    compliance_control: 0.30
+    integration_effort: 0.25
+    cost: 0.20
+    time_to_market: 0.15
+    competitive_advantage: 0.10
+
+pipeline:
+  max_signals_per_stage: 500
+  max_active_sessions: 5
+  session_timeout_hours: 72
+
+cache:
+  ttl_hours: 24
+  max_entries: 10000
+
+rate_limits:
+  web_requests_per_minute: 30
+  api_requests_per_minute: 60