Schema Discover AI

Convention: Values shown as X|Y in examples indicate allowed alternatives — use exactly one value per field, not the literal pipe character.

ai-workload-profile.json (Phase 1 output)

Focused profile of AI/ML workloads including models, capabilities, integration patterns, and supporting infrastructure.

Profile source and IaC inference:

• metadata.profile_source — "application_code" (full profile from code path), "iac_vertex" (minimal profile from Terraform Vertex-strong discovery only), or "merged" (code + pre-existing IaC profile combined in discover-app-code.md).
• summary.inferred_from_iac — true when the profile was produced or substantially completed without application code analysis (IaC-only or unknown integration until Clarify). false when code analysis drove model/SDK detection.

CRITICAL Field Names (use EXACTLY these keys):

• model_id — Model identifier string (NOT model_name, name)
• service — GCP service category (NOT service_type, gcp_service)
• detected_via — Array of detection sources (NOT detection_method, source)
• capabilities_used — Array of capability strings per model (NOT capabilities, features)
• usage_context — Human-readable description of what the model does (NOT description, purpose)
• pattern — Integration pattern in integration object: direct_sdk, framework, rest_api, mixed, or unknown when not inferable (typical for IaC-only profiles) (NOT integration_type, method)
• gateway_type — Gateway/router type in integration object: "llm_router", "api_gateway", "voice_platform", "framework", "direct", or null
• capabilities_summary — Boolean map in integration object (NOT capabilities, feature_flags)
• ai_source — Source AI provider in summary object: "gemini", "openai", "anthropic", "both", or "other"

Key Fields:

• metadata.sources_analyzed — Which data sources were provided (affects which sections are populated)
• summary.overall_confidence — Combined detection confidence from all signals
• models[] — Each distinct AI model/service detected, with evidence and capabilities. May be empty for profile_source: "iac_vertex" when no model IDs can be inferred from Terraform alone; Clarify and Design must not assume non-empty.
• integration.pattern — How the app connects to AI (direct_sdk, framework, rest_api, mixed, or unknown for IaC-only)
• integration.capabilities_summary — Union of all capabilities across all models
• infrastructure[] — Terraform resources related to AI (empty array if no Terraform provided)
• current_costs — Present ONLY if billing data OR OpenAI usage API data was provided; omitted entirely otherwise. source records provenance. Merge is provider-aware: billing CSVs measure GCP/Vertex spend, the usage API measures OpenAI spend — when both exist for DIFFERENT providers, monthly_ai_spend is their SUM with source: "mixed" and the per-provider split in breakdown[] (never max/pick-one — that drops a provider). Same-provider overlap: the usage API wins and the displaced figure lands in conflicting_sources[] (never silently resolved). breakdown is present only for source: "mixed".
• detection_signals[] — Raw signals from AI detection for transparency. method values include terraform, code, live_gcloud, and openai_usage_api.

Conditional sections:

• current_costs — Include ONLY if billing data was provided (billing discovery ran) or OpenAI usage API discovery ran (discover-openai-api.md). Omit entirely if neither.
• infrastructure — Set to [] if no Terraform files were provided (IaC discovery did not run).
• agentic_profile — Include ONLY if agentic signals detected (is_agentic: true). Omit entirely otherwise.
• tool_manifest — Include ONLY if agentic_profile exists. Set to [] if agentic but no tools detected.
• workloads — Always present. Empty array [] if no AI call sites detected.

workloads[] (per-workload disambiguation)

Generated by discover-app-code.md Step 5B. Splits the detected AI usage into distinct workloads — one per unique (model_id, sdk_method, structured_output) tuple. Enables downstream phases to produce one Bedrock recommendation per workload.

CRITICAL Field Names (use EXACTLY these keys):

• workload_id — Deterministic hash: "wl_" + sha256(model_id + "|" + sdk_method + "|" + structured_flag)[:6] (NOT id, name)
• model_id — Same model_id as in models[] (NOT model_name, source_model)
• sdk_method — Fully qualified SDK method name (NOT method, api_call)
• capability — Inferred capability: "text_generation", "structured_output", "image_generation", "embedding", "speech_to_text", "text_to_speech", "document_extraction", "image_analysis", "speech_transcription", "unknown" (NOT capability_inferred, type)
• capability_confidence — "high", "medium", or "low" (NOT confidence, confidence_level)
• structured_output — Boolean: true iff structured-output indicators detected in call arguments
• call_sites — Array of { "file": string, "line": integer } — repo-relative POSIX paths

Capability assignment rules (from references/vendored/ai/sdk-capability-map.json):

Validation rules:

• workloads[] is always present (empty array if no AI detected)
• Every workloads[].model_id MUST exist in models[].model_id
• workload_id is unique within the array and deterministic across re-runs
• call_sites is non-empty for every workload entry
• call_sites[].file uses repo-relative POSIX paths (forward slashes, no absolute paths)

agentic_profile (conditional — present only when agentic signals detected)

Generated by discover-app-code.md Step 3.5 when agentic framework signals are found. Captures agent architecture, orchestration patterns, and memory usage.

CRITICAL Field Names (use EXACTLY these keys):

• agent_id — Agent identifier (NOT name, agent_name, id)
• framework — Agentic framework detected (NOT agentic_framework, agent_framework, orchestration_framework)
• orchestration_pattern — How agents coordinate (NOT pattern, orchestration_type, coordination)
• memory_type — Per-agent memory type (NOT memory, memory_kind)
• memory_backend — System-level memory storage (NOT memory_store, backend)
• has_human_in_loop — Boolean (NOT human_in_loop, hitl)

Allowed values:

• framework: "langgraph", "crewai", "autogen", "openai_agents", "strands", "custom", "none"
• orchestration_pattern: "single", "hierarchical", "swarm", "graph", "sequential", "unknown"
• agents[].memory_type: "conversation_buffer", "rag", "none", "unknown"
• memory_backend: "redis", "postgres", "in_memory", "vector_store", "unknown"

Classification rules for orchestration_pattern:

Key fields:

• is_agentic — Always true when this section exists (redundant but explicit for downstream gates)
• framework — Primary agentic framework; if multiple detected, use the one with the most agent definitions
• agents[] — Each distinct agent/role detected; include all that can be identified from code
• agents[].model_id — The LLM model this agent uses; "unknown" if not extractable from code
• agents[].tools — Tool names attached to this agent; empty array [] if no tools detected
• agent_count — Must equal length of agents[] array
• tool_count — Total unique tools across all agents (deduplicated)

tool_manifest (conditional — present only when agentic_profile exists)

Generated by discover-app-code.md Step 6.5. Inventories tool definitions found in agent code. This is discovery only — no migration recommendations.

CRITICAL Field Names (use EXACTLY these keys):

• name — Tool name as defined in code (NOT tool_name, tool_id, id)
• transport — How the tool communicates (NOT transport_type, type, protocol)
• auth_hint — Detected authentication pattern (NOT auth, auth_type, authentication)
• used_by_agents — Array of agent_id values that reference this tool (NOT agents, consumers)

Allowed values:

• transport: "function" (local Python/JS/Go function), "api" (HTTP/REST call), "mcp" (MCP server), "unknown"
• auth_hint: "none", "api_key", "oauth", "iam", "unknown"

Classification rules for transport:

Classification rules for auth_hint:

Validation rules:

• Every agent_id in used_by_agents MUST exist in agentic_profile.agents[].agent_id
• tool_manifest length MUST equal agentic_profile.tool_count (deduplicated tool count)
• If agentic_profile exists but no tools detected: set tool_manifest: [] and agentic_profile.tool_count: 0