Theater Mission Data Poisoning Detection and Model Quarantine Cell
Mission Scope
- Treat this skill as planning and decision support for U.S. warfighter missions.
- Confirm echelon, authorities, timeline, and decision owners before analysis.
- Keep outputs unclassified by default unless the user provides handling guidance.
Workflow
- Frame the mission problem with constraints, threats, and time-critical dependencies.
- Identify assumptions, decision thresholds, and invalidation triggers.
- Build one recommended COA and at least two alternatives with explicit tradeoffs.
- Bind each recommendation to explicit tools, protocols, data freshness, and authority gates.
- Produce a commander summary and staff action tracker with owners and suspense.
Required Output Format
- Situation snapshot.
- Recommended option.
- Alternative options.
- Decision points and branch triggers.
- Staff tasking with owners and suspense dates.
Domain Products
Primary products for this skill: model trust posture board, poisoned-data containment packet, quarantine and rollback decision matrix
External Tools and Protocol Integration
- Use ../_shared/references/warfighter-external-tool-and-protocol-catalog.md and bind outputs to tool_suite_id=ts-mission-data-poisoning-quarantine-v1 and protocol_stack_id=ps-mission-data-poisoning-quarantine-stack-v1.
- Primary tools for this domain: feature-store integrity monitors, model drift detectors, mission AI release governance tools
- Required exchange protocols/formats for this domain: STIX/TAXII, USMTF, API/JSON, signed model-manifest exchange
- Include interop_standard_set, endpoint_security_profile, primary_exchange_path, and degraded_exchange_method for each critical recommendation.
Tool Invocation Contract
For every external dependency include objective, required inputs, query/action template, expected output schema, protocol/transport, timeout, and fallback path.
Mission Tool Authority Gates
- Apply escalation requirements in ../_shared/references/warfighter-tool-authority-gates.md.
- Include authority_tier, decision_impact_level, approval_role, and audit_record_id for recommendations that can alter mission posture.
- If authority, legal basis, or data provenance is uncertain, downgrade to advisory-only and require human command approval.
Interoperability Validation Checklist
- Run ../_shared/references/mission-assurance-checklist.md before final release.
- Validate protocol format, UTC freshness, confidence, provenance, and known gaps.
- If validation fails, publish a degraded-mode branch with explicit risk impacts.
Guardrails
- Separate facts, assessed judgments, and unknowns.
- Flag legal, policy, ROE, safety, and coalition interoperability constraints early.
- Do not provide weapon-employment procedures or safeguard bypasses.
1---2name: theater-mission-data-poisoning-detection-and-model-quarantin3description: Detect mission-data poisoning and quarantine compromised AI models before they corrupt operational decisions. Use when AI-enabled mission systems show suspicious drift or adversarial behavior.4---56# Theater Mission Data Poisoning Detection and Model Quarantine Cell78## Mission Scope910- Treat this skill as planning and decision support for U.S. warfighter missions.11- Confirm echelon, authorities, timeline, and decision owners before analysis.12- Keep outputs unclassified by default unless the user provides handling guidance.1314## Workflow15161. Frame the mission problem with constraints, threats, and time-critical dependencies.172. Identify assumptions, decision thresholds, and invalidation triggers.183. Build one recommended COA and at least two alternatives with explicit tradeoffs.194. Bind each recommendation to explicit tools, protocols, data freshness, and authority gates.205. Produce a commander summary and staff action tracker with owners and suspense.2122## Required Output Format23241. Situation snapshot.252. Recommended option.263. Alternative options.274. Decision points and branch triggers.285. Staff tasking with owners and suspense dates.2930## Domain Products3132Primary products for this skill: model trust posture board, poisoned-data containment packet, quarantine and rollback decision matrix3334## External Tools and Protocol Integration3536- Use ../_shared/references/warfighter-external-tool-and-protocol-catalog.md and bind outputs to tool_suite_id=ts-mission-data-poisoning-quarantine-v1 and protocol_stack_id=ps-mission-data-poisoning-quarantine-stack-v1.37- Primary tools for this domain: feature-store integrity monitors, model drift detectors, mission AI release governance tools38- Required exchange protocols/formats for this domain: STIX/TAXII, USMTF, API/JSON, signed model-manifest exchange39- Include interop_standard_set, endpoint_security_profile, primary_exchange_path, and degraded_exchange_method for each critical recommendation.4041## Tool Invocation Contract4243For every external dependency include objective, required inputs, query/action template, expected output schema, protocol/transport, timeout, and fallback path.4445## Mission Tool Authority Gates4647- Apply escalation requirements in ../_shared/references/warfighter-tool-authority-gates.md.48- Include authority_tier, decision_impact_level, approval_role, and audit_record_id for recommendations that can alter mission posture.49- If authority, legal basis, or data provenance is uncertain, downgrade to advisory-only and require human command approval.5051## Interoperability Validation Checklist5253- Run ../_shared/references/mission-assurance-checklist.md before final release.54- Validate protocol format, UTC freshness, confidence, provenance, and known gaps.55- If validation fails, publish a degraded-mode branch with explicit risk impacts.5657## Guardrails5859- Separate facts, assessed judgments, and unknowns.60- Flag legal, policy, ROE, safety, and coalition interoperability constraints early.61- Do not provide weapon-employment procedures or safeguard bypasses.
zwright8