AI Workflow Policy (MEGA16B)

mega16b_ai_workflow_policyMEGA16B_AI_WORKFLOW_POLICY_READY
Generated at
2026-05-29T10:07:47Z
Snapshot version
1
Redaction
APPLIED
Source paths
  • config/ai_gateway/mega16b/controlled_ai_workflow_policy.json
Warnings
  • MEGA16B — controlled AI participation across 10 advisory workflows; global live AI still gated
  • 3 live grouped OpenAI calls + 1 deterministic lexical mapping; total cost ~$0.0014
  • All AI outputs PENDING_REVIEW / DRAFT / NON_FINAL · human review mandatory
  • Knowledge Brain: LEXICAL_INTERNAL_RETRIEVAL ready; vector embeddings deferred
  • NO owner testing · NO public launch · NO final approval · NO external export
  • Loopback only · Caddy/DNS untouched · no dependency on external AI cores

Top-level metrics

FieldValue
scope_labelCONTROLLED_AI_WORKFLOW_PARTICIPATION_ONLY
max_live_calls_for_mega16b4
per_call_cost_cap_usd0.05
total_phase_cost_cap_usd1
max_output_tokens700
request_timeout_seconds30
max_retries0
workflows.length10
rollback_on_any_cap_breachtrue

workflows (10)

workflow_idcanonical_keygrouplive_call_required
WF-MEGA16B-01diagnostic_interpretationAtrue
WF-MEGA16B-02ten_layer_assessmentAtrue
WF-MEGA16B-03rise_gate_analysisAtrue
WF-MEGA16B-04strategy_options_generationBtrue
WF-MEGA16B-05execution_readiness_analysisBtrue
WF-MEGA16B-06decision_memo_draftingBtrue
WF-MEGA16B-07evaluation_scoring_assistanceCtrue
WF-MEGA16B-08report_drafting_assistanceCtrue
WF-MEGA16B-09human_review_item_generationCtrue
WF-MEGA16B-10knowledge_evidence_mappingDfalse