251 P1 Project Understanding 3463d1f5

Phase 1: Project Understanding

Type: Exploratory Executor: Script + LLM Knowledge: Security Principles

⚠️ MANDATORY: 4-Phase Gating Protocol (BLOCKING)

CRITICAL: You MUST complete the following four stages in sequence and output the result of each stage. Skipping any stage will degrade analysis quality!

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

🧠 THINKING - Phase 1 Entry Gate

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Purpose: Phase 1 is the first phase with no upstream data dependencies. Focus on project discovery.

⚠️ You MUST output THINKING results in the following format:

🧠 THINKING - P1 Entry Gate
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📌 CORE PROBLEM
Comprehensively discover project architecture, modules, and entry points to establish the threat modeling foundation

📊 KNOWN CONTEXT
| Metric | Value | Source |
|--------|-------|--------|
| Project Path | {actual_path} | User Input |
| Project Type | [to be discovered] | Requires scan |
| Tech Stack | [to be discovered] | package.json/requirements.txt |

❓ UNKNOWNS
- Module organization structure
- Entry point distribution (14 types require scanning)
- Dynamic route indicators
- Documentation quality grade

⚠️ RISKS
- Incomplete entry point type scanning (all 14 types MUST be scanned)
- Missing dynamic route indicators (Layer 3 coverage)
- coverage_confidence too low (<0.70)
- Module security_level not assigned

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
⛔ STOP CHECK
- Project path accessible? [YES/NO]
- Ready to continue PLANNING? [YES/NO]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

⛔ STOP CONDITION: If any STOP CHECK = NO → Resolve the issue before continuing

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📋 PLANNING - Sub-task Decomposition

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Step 1: Validate Project Path (BLOCKING - MUST execute)

# Verify project exists (execute in project root directory)
ls .

# Check if .phase_working directory exists
ls ./Risk_Assessment_Report/.phase_working/ 2>/dev/null || echo "Will create"

Step 2: Output Sub-task Table (MANDATORY)

⚠️ You MUST output PLANNING results in the following format:

📋 PLANNING - P1 Sub-tasks
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

| # | Sub-task | Expected Output |
|---|----------|-----------------|
| T1 | Execute P1.0 three-layer static discovery | P1_static_discovery.yaml |
| T2 | Decide P1.1 doc analysis based on quality_grade | yaml:doc_analysis (conditional) |
| T3 | P1.2 code analysis | 3 YAML blocks |
| T4 | P1.3 dynamic route detection | Layer 3 indicators |
| T5 | P1.4 three-source alignment validation | P1_source_alignment.yaml |
| T6 | P1.5 validation and confidence calculation | coverage_confidence |
| T7 | Write final output | P1_project_context.yaml + MD |

⛔ PLANNING CHECK
- Sub-tasks decomposed? [YES/NO]
- Ready to execute TaskCreate? [YES/NO]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Step 3: TaskCreate for ALL sub-tasks (MANDATORY)

⚠️ Before starting any implementation, you MUST execute TaskCreate to create all sub-tasks!

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

⚡ EXECUTION LOOP

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

For each sub-task:

1. TaskUpdate(status: "in_progress")
2. Implement sub-task
3. Verify: Does output match expectations?
4. If verification passes: TaskUpdate(status: "completed") → Next sub-task
5. If verification fails: Diagnose → Fix → Retry (max 3x) → If still failing: CHECKPOINT to request user decision

Output Order (CRITICAL):

1. Write YAML first: .phase_working/{SESSION_ID}/data/P1_project_context.yaml
2. Write MD second: .phase_working/{SESSION_ID}/reports/P1-PROJECT-UNDERSTANDING.md

Key Commands:

# P1.0 Three-layer discovery (path is positional argument, not --project-root)
python $SKILL_PATH/scripts/module_discovery.py . --p1-discovery --output-yaml \
  > .phase_working/{SESSION_ID}/data/P1_static_discovery.yaml
# Note: "." means current project directory, can also use absolute path like /path/to/project

# P1.5 Validation
python $SKILL_PATH/scripts/phase_data.py --validate --phase 1 --root .

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

🔍 REFLECTION - Completion Verification

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

⚠️ After completing EXECUTION, you MUST output REFLECTION results in the following format:

🔍 REFLECTION - P1 Completion Check
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

| Check Item | Status |
|------------|--------|
| P1.0 three-layer discovery executed? (P1_static_discovery.yaml exists) | [✅/❌] |
| P1_project_context.yaml exists and valid? | [✅/❌] |
| discovery_checklist all 14 entry types scanned:true? | [✅/❌] |
| Every module has security_level assigned? | [✅/❌] |
| Every entry point has unique ID (EP-xxx format)? | [✅/❌] |
| coverage_confidence.overall_confidence ≥ 0.70? | [✅/❌] |
| Hook validation passed (exit 0)? | [✅/❌] |

⛔ COMPLETION GATE
- All checks passed? [YES/NO]
- Ready to enter P2? [YES/NO]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

⛔ If any check fails → Fix and re-verify until all pass

⚠️ MANDATORY OUTPUT RULES

CRITICAL: Phase 1 requires TWO outputs - a YAML data file AND a Markdown report.

Dual Output Requirement

┌─────────────────────────────────────────────────────────────────────┐
│  PHASE 1 MUST PRODUCE TWO FILES:                                    │
├─────────────────────────────────────────────────────────────────────┤
│                                                                      │
│  1. DATA FILE (PRIMARY - Write First!)                              │
│     Path: .phase_working/{SESSION_ID}/data/P1_project_context.yaml  │
│     Purpose: Structured data for P2 to read                         │
│     Format: Valid YAML with schema_version: "3.0.3 (20260209a)"                   │
│                                                                      │
│  2. REPORT FILE (SECONDARY - Write After Data!)                     │
│     Path: .phase_working/{SESSION_ID}/reports/P1-PROJECT-UNDER...md │
│     Purpose: Human-readable analysis report                         │
│     Format: Markdown with sections and tables                       │
│                                                                      │
│  ❌ FORBIDDEN: Writing only .md without .yaml                       │
│  ❌ FORBIDDEN: Embedding YAML blocks inside .md as data source      │
│  ✅ REQUIRED: .yaml file is the authoritative data source           │
│                                                                      │
└─────────────────────────────────────────────────────────────────────┘

Required Data Sections in YAML

Validation Gate

Phase 1 CANNOT complete until:

1. .phase_working/{SESSION_ID}/data/P1_project_context.yaml exists and is valid YAML
2. discovery_checklist has all 10 entry types with scanned: true
3. Every module has security_level assigned
4. Every entry point has unique ID (EP-xxx format)
5. .phase_working/{SESSION_ID}/reports/P1-PROJECT-UNDERSTANDING.md exists

Input Context

None (first phase)

Output Context

→ P2: project_context {project_type, modules[], entry_points[], security_design{}}

Sub-Phase Architecture

Terminology Note: "P1.x" refers to sub-phases within Phase 1. "Layer 1/2/3" refers to the three-layer discovery strategy (deterministic/heuristic/dynamic). These are orthogonal concepts.

P1.0 (Script)        Three-Layer Static Discovery (Layer 1 + Layer 2 + Layer 3)
        ↓
P1.1 (LLM)           Doc-Guided Analysis (conditional)
        ↓
P1.2 (LLM)           Code Analysis & Module Inventory
        ↓
P1.3 (Script)        Dynamic Route Detection (Layer 3 refinement)
        ↓
P1.4 (Script)        Three-Source Alignment
        ↓
P1.5 (Script+LLM)    Validation & Coverage Confidence

*P1.1 required if documentation.quality_grade != "none"

Three-Layer Discovery Strategy

┌─────────────────────────────────────────────────────────────────────────────┐
│               THREE-LAYER DISCOVERY ARCHITECTURE                             │
├─────────────────────────────────────────────────────────────────────────────┤
│                                                                              │
│  LAYER 1: Deterministic Discovery (95%+ Confidence)                          │
│  ─────────────────────────────────────────────────                            │
│  • Static code patterns: @app.route, @router.get, @RequestMapping            │
│  • Config file parsing: openapi.yaml, routes.yaml, serverless.yml            │
│  • Framework-specific: Django URLconf, Spring controllers                    │
│  Output: Verified routes with exact file:line locations                      │
│                                                                              │
│  LAYER 2: Heuristic Discovery (70-90% Confidence)                            │
│  ─────────────────────────────────────────────────                            │
│  • Directory patterns: routes/, handlers/, api/, controllers/                │
│  • Import analysis: flask, fastapi, express, spring detection                │
│  • Code pattern matching: request/response handling functions                │
│  Output: Probable route locations for LLM analysis                          │
│                                                                              │
│  LAYER 3: Dynamic Route Indicators (30-60% Confidence)                       │
│  ─────────────────────────────────────────────────────                        │
│  • Conditional registration: if config.ENABLE_* + add_route                  │
│  • Plugin systems: plugin.register, loader.load_module                       │
│  • Reflection calls: getattr, eval, exec                                     │
│  • Runtime registration: def register_routes() patterns                      │
│  Output: Uncertainty markers for human review                               │
│                                                                              │
│  COVERAGE CONFIDENCE FORMULA:                                                │
│  ────────────────────────────                                                 │
│  final = base_confidence - uncertainty_penalty + tool_boost                  │
│                                                                              │
│  Where:                                                                      │
│    l1_contribution = max(l1_route_conf, l1_config_conf) * 0.6               │
│    l2_contribution = max(l2_dir_conf, l2_import_conf) * 0.3                 │
│    base_confidence = min(0.95, l1_contribution + l2_contribution + 0.05)    │
│    uncertainty_penalty = min(0.20, dynamic_count * 0.02 + high_risk * 0.03) │
│    tool_boost = sum(available_tools.boost), capped at 0.15                  │
│                                                                              │
│  Weight Distribution:                                                        │
│    • Layer 1 (Deterministic): 60% weight                                    │
│    • Layer 2 (Heuristic): 30% weight                                        │
│    • Baseline: +5% minimum confidence                                        │
│                                                                              │
│  Enhanced Tools (when available):                                            │
│    • Semgrep: +0.10 confidence boost (framework-native detection)           │
│    • OWASP Noir: +0.15 confidence boost (LLM-enhanced discovery)            │
│    • CodeQL: +0.12 confidence boost (deep dataflow analysis)                │
│                                                                              │
└─────────────────────────────────────────────────────────────────────────────┘

P1.0 Three-Layer Static Discovery (Script)

Run first, before any manual analysis:

# NEW: Full three-layer P1 discovery with YAML output
# Usage: module_discovery.py <path> [options]  (path is positional argument)
python $SKILL_PATH/scripts/module_discovery.py . --p1-discovery --output-yaml \
  > .phase_working/{SESSION_ID}/data/P1_static_discovery.yaml

Error Handling

Fallback Strategy: If script fails after 2 retries, proceed to P1.2 with manual discovery and set coverage_confidence.overall_confidence: 0.50 with note "script_fallback": true.

Discovery Outputs

The three-layer discovery produces:

⚠️ JSON/YAML Output Handling Rules

Why: Large output through pipes may corrupt. head truncates structures mid-way.

Correct Pattern:

# Step 1: Run full P1 discovery (path is the first positional argument)
python $SKILL_PATH/scripts/module_discovery.py . --p1-discovery --output-yaml \
  > .phase_working/{SESSION_ID}/data/P1_static_discovery.yaml

# Step 2: Check coverage confidence
grep "overall_confidence" .phase_working/{SESSION_ID}/data/P1_static_discovery.yaml

Record from output:

Decision Gate:

Coverage Confidence Decision:

P1.0+ Enhanced Tool Discovery (LLM Decision + Execution)

IMPORTANT: Script DETECTS tools and SUGGESTS commands. LLM DECIDES and CALLS.

When to Execute

Check enhanced_tools._summary in P1_static_discovery.yaml output:

# Example output from module_discovery.py
enhanced_tools:
  _summary:
    available_count: 2
    available_tools: ["semgrep", "noir"]
    llm_action_required: true
    instruction: "Enhanced tools detected. LLM should execute..."
  semgrep:
    available: true
    version: "1.55.0"
    suggested_action: "RUN_FOR_ENHANCED_DISCOVERY"
    recommended_commands:
      - purpose: "route_discovery"
        command: "semgrep --config auto --json --output semgrep_routes.json {project_path}"
      - purpose: "security_patterns"
        command: "semgrep --config p/security-audit --json --output semgrep_security.json {project_path}"

LLM Decision Matrix

LLM Execution Protocol

When llm_action_required == true:

# Step 1: LLM executes recommended commands (replace {project_path} with actual path)
# For Semgrep:
semgrep --config auto --json --output .phase_working/{SESSION_ID}/data/semgrep_routes.json {project_path}

# For OWASP Noir:
noir -b {project_path} -f json -o .phase_working/{SESSION_ID}/data/noir_endpoints.json

Result Integration

LLM should merge enhanced discovery results into P1.2 analysis:

# Add to yaml:entry_point_inventory
enhanced_discovery_results:
  tools_executed: ["semgrep", "noir"]
  semgrep_findings:
    routes_discovered: 12
    security_patterns: 5
    source_file: "semgrep_routes.json"
  noir_findings:
    endpoints_discovered: 18
    source_file: "noir_endpoints.json"
  merged_entry_points:
    - id: EP-ENH-001
      path: "/api/internal/sync"
      source: "semgrep"
      confidence: 0.95

Confidence Boost Application

When enhanced tools successfully execute:

• Add tool_boost to coverage_confidence
• Update coverage_confidence.confidence_breakdown.tool_boost

P1.1 Documentation Analysis (LLM)

Skip if: quality_grade == "none"

Document Priority:

1. README.md
2. docs/ARCHITECTURE*, DESIGN*
3. docs/API*, openapi.yaml
4. CONTRIBUTING*
5. Other docs/*.md

Output: yaml:doc_analysis

schema_version: "3.0.3 (20260209a)"
phase: 1
sub_phase: "P1.1"
analyzed_at: "ISO8601"
documents_analyzed:
  - path: "README.md"
    category: "readme"
    size_bytes: 12500

project_intent:
  summary: "One-paragraph summary"
  target_users: ["developers", "enterprises"]
  key_features: ["Feature 1", "Feature 2"]

architecture_overview:
  type: "monolith|microservices|serverless|hybrid"
  frontend:
    framework: "React/Vue/Svelte"
  backend:
    framework: "FastAPI/Express/Spring"
  database:
    type: "relational/nosql/mixed"
    systems: ["PostgreSQL", "Redis"]

documented_modules:
  - name: "Authentication Module"
    description: "Handles OAuth2 and local auth"
    source: "docs/ARCHITECTURE.md"

security_mentions:
  - topic: "authentication"
    details: "OAuth2 with PKCE support"
    source: "README.md"

notes_for_analysis:
  - "Documentation mentions webhook integration not yet implemented"

P1.2 Code Analysis (LLM)

Core Goal: Comprehensively understand project architecture by verifying documented claims and discovering undocumented components.

Entry Point Types (All Must Be Scanned)

Required Output Blocks

Block 1: yaml:module_inventory

modules:
  - id: M-001              # Format: M-{Seq:03d}
    name: "Authentication Module"
    path: "src/auth"
    type: Authentication
    security_level: HIGH   # HIGH|MEDIUM|LOW
    files: 12
    loc: 1500
    entry_types: [API, UI]
    submodules:
      - id: M-001-001      # Parent-Child: M-{Parent}-{Seq:03d}
        name: "Auth Handlers"
        path: "src/auth/handlers"
        files: 4
        loc: 600

Block 2: yaml:entry_point_inventory

api_entries:
  - id: EP-API-001
    path: "/api/v1/auth/login"
    methods: [POST]
    module: auth
    handler: "src/auth/handlers/login.py:45"
    auth_required: false
    exposure: EXTERNAL

ui_entries:
  - id: EP-UI-001
    type: WebForm
    path: "/login"
    component: "LoginForm"

system_entries:
  - id: EP-SYS-001
    type: CronJob
    trigger: "0 * * * *"

hidden_entries:
  - id: EP-HID-001
    path: "/health"

Block 3: yaml:discovery_checklist

checklist:
  rest_api:
    scanned: true
    count: 45
    source_patterns: ["routes/*.py", "@app.route"]
    status: COMPLETED
  internal_api:
    scanned: true
    count: 8
    status: COMPLETED
  graphql:
    scanned: true
    count: 0
    status: NOT_APPLICABLE
    reason: "Project does not use GraphQL"
  websocket:
    scanned: true
    count: 2
    status: COMPLETED
  cron_jobs:
    scanned: true
    count: 3
    status: COMPLETED
  message_queue:
    scanned: true
    count: 0
    status: NOT_APPLICABLE
  webhooks:
    scanned: true
    count: 2
    status: COMPLETED
  file_upload:
    scanned: true
    count: 3
    status: COMPLETED
  health_endpoints:
    scanned: true
    count: 3
    status: COMPLETED
  debug_endpoints:
    scanned: true
    count: 1
    status: COMPLETED
  # AI/LLM Entry Types (when applicable)
  llm_gateway:
    scanned: true
    count: 0
    status: NOT_APPLICABLE
    reason: "Project is not an AI/LLM application"
  model_endpoint:
    scanned: true
    count: 0
    status: NOT_APPLICABLE
  rag_pipeline:
    scanned: true
    count: 0
    status: NOT_APPLICABLE
  agent_tool:
    scanned: true
    count: 0
    status: NOT_APPLICABLE

summary:
  total_entry_points: 72
  coverage: "100%"
  ai_llm_detected: false  # Set to true if any AI entry types found

Scenario Detection

Non-Typical Project Escape Hatch (P1-GAP-14)

For projects that don't fit standard patterns (SDK/libraries, data pipelines, batch processors):

Escape Hatch Protocol:

1. If total_entry_points == 0 after P1.0, check if project is non-typical
2. Add project_category: "non_typical" to P1_project_context.yaml
3. Document justification in discovery_checklist.escape_hatch_reason
4. Proceed to P1.2 with adapted focus

P1.3 Dynamic Route Detection (Script)

Note: P1.0 --p1-discovery already includes Layer 3 analysis. P1.3 is a refinement step that runs ONLY if layer3_dynamic_indicators.total_count > 0 in P1.0 output, requiring deeper analysis.

Run after P1.2 LLM analysis (OPTIONAL - only if P1.0 found dynamic indicators):

# Skip if P1.0 layer3_dynamic_indicators.total_count == 0
# Run only for deeper analysis of HIGH risk indicators
python $SKILL_PATH/scripts/module_discovery.py . --detect-dynamic --pretty

Layer 3 Indicator Types

Handling Dynamic Indicators

If layer3_dynamic_indicators.total_count > 0:

1. Review each indicator in the output
2. For HIGH risk indicators: Document in architecture_findings with F-P1-xxx ID
3. For MEDIUM risk indicators: Note in discovery_checklist.notes
4. Add to coverage_confidence.uncertainty_sources in final output

P1.4 Three-Source Alignment (Script)

Run to validate alignment between discovery sources:

python $SKILL_PATH/scripts/phase_data.py --p1-source-alignment --root .

Three Sources Compared

Note: Source B is extracted from the documentation section of P1_static_discovery.yaml (generated by module_discovery.py --p1-discovery). If P1.1 was executed, doc_analysis data is also included. There is NO separate P1_doc_inventory.yaml file.

Alignment Score Interpretation

Output: P1_source_alignment.yaml

schema_version: "3.0.3 (20260209a)"
overall_alignment_score: 0.82
alignment_by_category:
  entry_points:
    total_items: 45
    in_all_three_sources: ["api/login", "api/register"]
    in_two_sources: ["api/webhook"]
    only_in_static: ["api/internal"]
    only_in_docs: []
    only_in_llm: ["api/admin"]
    needs_review: true

P1.5 Validation & Coverage Confidence (Script+LLM)

Run after completing P1.4:

python $SKILL_PATH/scripts/phase_data.py --validate --phase 1 --root .

Validation Rules:

If BLOCKING fails: Fix issues, do not proceed If WARNING only: Acknowledge and continue

Coverage Confidence Output

The final P1_project_context.yaml MUST include:

coverage_confidence:
  overall_confidence: 0.82
  confidence_breakdown:
    layer1_contribution: 0.57
    layer2_contribution: 0.24
    base_confidence: 0.86
    uncertainty_penalty: 0.08
    tool_boost: 0.04
  uncertainty_sources:
    - type: conditional_registration
      impact: HIGH
      count: 2
      description: "Routes registered based on configuration flags"
    - type: no_openapi_spec
      impact: LOW
      count: 1
      description: "No OpenAPI specification found for validation"
  has_uncertainty: true
  recommendation: "MEDIUM_CONFIDENCE_REVIEW_RECOMMENDED"

Report Template

# P1: Project Understanding

## Sub-Phase Progress Tracker

| Sub-Phase | Status | Output | Notes |
|-----------|--------|--------|-------|
| P1.0 Three-Layer Discovery | □ Done | P1_static_discovery.yaml | coverage_confidence: _____ |
| P1.1 Doc Analysis | □ Done / □ Skipped | yaml:doc_analysis | Skip reason: _____ |
| P1.2 Code Analysis | □ Done | 3 YAML blocks | |
| P1.3 Dynamic Indicators | □ Done | Layer 3 count: _____ | HIGH risk: _____ |
| P1.4 Source Alignment | □ Done | alignment_score: _____ | |
| P1.5 Validation | □ Done | PASSED/FAILED | |

## P1.0 Three-Layer Discovery Summary

| Metric | Value |
|--------|-------|
| total_files | _____ |
| total_directories | _____ |
| Layer 1 routes discovered | _____ |
| Layer 2 directories found | _____ |
| Layer 3 dynamic indicators | _____ |
| coverage_confidence | _____ |
| recommendation | HIGH/MEDIUM/LOW |

## P1.1 Decision
- quality_grade: `{grade}` (score: {score})
- Decision: **Execute P1.1** / **Skip P1.1**
- Reason: _____

[yaml:doc_analysis block if P1.1 executed]

## Module Inventory

[yaml:module_inventory block]

## Entry Point Inventory

[yaml:entry_point_inventory block]

## Discovery Checklist

[yaml:discovery_checklist block]

## Layer 3 Dynamic Indicators (Uncertainty Sources)

| Indicator Type | Count | Risk | Files |
|----------------|-------|------|-------|
| conditional_registration | _____ | HIGH | _____ |
| plugin_system | _____ | MEDIUM | _____ |
| reflection_calls | _____ | HIGH | _____ |

## Source Alignment Summary

| Category | 3-Source | 2-Source | 1-Source | Needs Review |
|----------|----------|----------|----------|--------------|
| entry_points | _____ | _____ | _____ | _____ |
| modules | _____ | _____ | _____ | _____ |
| frameworks | _____ | _____ | _____ | _____ |

**Alignment Score**: _____
**Recommendation**: _____

## Architecture Findings

```yaml:architecture_findings
findings:
  - id: F-P1-001
    type: architecture
    title: "Finding title"
    description: "Detailed description"
    severity: HIGH      # CRITICAL|HIGH|MEDIUM|LOW|INFO
    category: entry_point|module_structure|dependency|configuration|dynamic_route
    location:
      module_id: M-xxx
      file: "path/to/file.py"
      line: 123
    security_relevance: "Why this matters for security"
    related_elements:
      - type: entry_point
        id: EP-xxx
      - type: module
        id: M-xxx
    recommended_action: "What to investigate in later phases"

summary:
  total: 0
  by_severity:
    critical: 0
    high: 0
    medium: 0
    low: 0
    info: 0

Finding Categories:

• entry_point: Unusual or sensitive entry points discovered
• module_structure: Architectural concerns or anti-patterns
• dependency: Risky or outdated dependencies
• configuration: Security-relevant configuration issues
• dynamic_route: Routes that may be registered dynamically (uncertainty source)

Coverage Confidence

overall_confidence: 0.82
confidence_breakdown:
  layer1_contribution: 0.57
  layer2_contribution: 0.24
  base_confidence: 0.86
  uncertainty_penalty: 0.08
  tool_boost: 0.04
uncertainty_sources:
  - type: conditional_registration
    impact: HIGH
    count: 2
has_uncertainty: true
recommendation: "MEDIUM_CONFIDENCE_REVIEW_RECOMMENDED"

---

## Completion Checklist

Before marking Phase 1 complete:

- [ ] P1.0 three-layer discovery executed (P1_static_discovery.yaml)
- [ ] P1.1 executed OR skipped (with valid reason)
- [ ] P1.2 LLM analysis complete:
  - [ ] yaml:module_inventory present
  - [ ] yaml:entry_point_inventory present
  - [ ] yaml:discovery_checklist present with all scanned:true
  - [ ] yaml:architecture_findings present (even if empty)
- [ ] P1.3 dynamic indicators reviewed (if any)
- [ ] P1.4 source alignment validated (P1_source_alignment.yaml)
- [ ] P1.5 validation passed:
  - [ ] coverage_confidence included in output
  - [ ] uncertainty_sources documented (if any)

---

**End of Phase 1 Instructions** (~450 lines, ~3.5K tokens)