Trust survival
Starter Runtime Mission 4
Governed Opportunity Intelligence
- Trust band: OBSERVE_ONLY
- Authority: Bounded
- Recovery: Enabled
- Replay: Available
- Autonomy expansion: Disabled
- Governance tier: Tier D
Provisioned via starter onboarding — preview governance records and trust evidence only. Does not unlock gmail.send, Tier B execution, or Phase 5 authority expansion.
Governed Opportunity Intelligence
observe_onlyopportunity_intelligence_v1
Trust contract preview for declared authority boundaries. ReflexFlow may describe workflow authority before it is allowed to enforce it.
Dry run (simulation) →Recovery simulation (HITL) →Opportunity workspace →Trust expansion governance →Longitudinal trust intelligence →
Trust contract preview
observe_onlyAuthority boundaries and recovery posture as declared — preview only, not enforced.
- Risk tier
- D (protected_human_led)
- Max turns
- 0 (tier default 0)
- Trust threshold
- 0.84
- Preview warnings
- 0
Allowed tools
- role_search_read
- company_profile_read
- resume_variant_prepare
- outreach_draft_prepare
- application_tracker_update
Allowed actions
- discover_roles
- score_fit
- draft_application_packet
- draft_outreach
- update_pipeline_status
Forbidden actions
- linkedin.easy_apply.submit
- application.submit
- outreach.send
- profile.update
- mass_apply
- autonomous_outreach
Human checkpoint rules
- require_before_execute: true
- require_on_trust_drop: true
- require_on_forbidden_action: true
- labels: before_application_packet, before_outreach, before_status_change
Escalation & recovery
- failure_escalation: human_queue
- rollback_strategy: last_human_approved
- retry_policy: max_retries=0, backoff_seconds=0
TrustGraph preview
observe_onlyRead-only observe_only graph: template → tools → HITL checkpoints. Not a live TrustGraph runtime surface.
Nodes
- workflow_template · Governed Opportunity Intelligence
- tool · role_search_read
- tool · company_profile_read
- tool · resume_variant_prepare
- tool · outreach_draft_prepare
- tool · application_tracker_update
- hitl_checkpoint · before_application_packet
- hitl_checkpoint · before_outreach
- hitl_checkpoint · before_status_change
Edges
- template:opportunity_intelligence_v1 → tool:role_search_read [allowed_tool]
- template:opportunity_intelligence_v1 → tool:company_profile_read [allowed_tool]
- template:opportunity_intelligence_v1 → tool:resume_variant_prepare [allowed_tool]
- template:opportunity_intelligence_v1 → tool:outreach_draft_prepare [allowed_tool]
- template:opportunity_intelligence_v1 → tool:application_tracker_update [allowed_tool]
- template:opportunity_intelligence_v1 → hitl:before_application_packet [human_checkpoint]
- template:opportunity_intelligence_v1 → hitl:before_outreach [human_checkpoint]
- template:opportunity_intelligence_v1 → hitl:before_status_change [human_checkpoint]
Contract validator
observe_onlyopportunity_intelligence_v1Read-only preview validation via parseWorkflowTemplateV0. No mutations or enforcement.
Preview GhostLog
observe_onlyPreview events only — not enforcement. These events are not persisted to production GhostLog tables.
- workflow_contract.loaded2026-05-28T00:48:30.602Z
{ "sourceFile": "packages/sdk/schemas/examples/opportunity_intelligence_v1.yaml" } - workflow_contract.validated2026-05-28T00:48:30.603Z
{ "valid": true, "errorCount": 0, "warningCount": 0 } - workflow_contract.risk_previewed2026-05-28T00:48:30.603Z
{ "riskLevel": "D", "riskPosture": "protected_human_led", "maxTurns": 0, "tierDefaultMaxTurns": 0 }