Phase 1 of 6
Scoping & Target-Engagement Authority
Define the mission envelope, release authority, rules-of-engagement boundaries, and LOAC posture before any targeting-support capability is designed. This is a decision-support system with a human in the release chain — that framing must be established at scope time, not bolted on later.
0/11
Phase Progress
Required Recommended Optional Open-Source Proprietary Trinidy
Mission Scope & Release Authority
Confirm the system is targeting decision support, not an autonomous weapon
Why This Matters
DoD Directive 3000.09 (updated 25 January 2023) is the authoritative US policy on autonomy in weapon systems. It draws a hard legal and procedural line between human-in-the-loop decision support, semi-autonomous weapons, and autonomous weapons that select and engage targets without further human action. Autonomous and certain semi-autonomous systems require a senior review by the USD(P), USD(R&E), and Vice Chairman prior to formal development and again before fielding. Mislabeling a system at scoping is the single fastest way to fail a weapons-review legal check later.
Note prompts — click to add
+ Has the program formally classified this system under the DoDD 3000.09 taxonomy in writing?+ If semi-autonomous or autonomous, has the senior-review package been initiated?+ Does every document, brief, and UI screen describe this as decision support with human release authority?Explicitly classify the system against DoDD 3000.09 categories before any architecture decision.
Single choice
Identify the release authority and engagement-approval chain
Why This Matters
The CJCSI 3121.01B Standing Rules of Engagement define who may authorize what level of force under which conditions, and those authorities flow differently in deliberate targeting (Joint Targeting Cycle, ATO-driven) versus dynamic targeting (F2T2EA — find, fix, track, target, engage, assess) versus time-sensitive targeting. The system must make the release authority explicit on every targeting recommendation — a target folder that does not name its required approver is a target folder that cannot be legally released.
Note prompts — click to add
+ Does every target nomination surface the required release authority on the UI?+ Have we mapped our targeting workflow to both deliberate and dynamic targeting authority chains?+ Who holds release authority for time-sensitive targets, and how is that authority delegated in practice?Specify who holds release authority and where the human decision point is enforced in the system.
Select all that apply
Define rules-of-engagement scope and special instructions (SPINS) ingestion
Why This Matters
ROE are not static — theater commanders tailor standing rules with supplemental measures that change by mission, phase of operation, and political guidance. Protected sites under the 1954 Hague Convention on Cultural Property and the No-Strike List must be treated as hard constraints, not soft preferences. A targeting-support model that cannot ingest the current NSL and today's SPINS cannot provide legally defensible recommendations.
Confirm how current ROE, SPINS, and No-Strike List data reach the targeting support system.
Select all that apply
Specify target category scope (what this system recommends on)
Why This Matters
Every target category carries a different LOAC risk profile and a different political-military authority threshold. Dual-use infrastructure (power, comms, transport) is governed by very narrow proportionality reasoning under AP-I Article 52 and usually requires elevated approval. Personnel targeting support is a category most programs deliberately exclude from the initial validated envelope because the identification error margin is much less forgiving.
Define which target categories are within the system's validated envelope.
Select all that apply
Set latency envelope by targeting mode
Why This Matters
The 3–5 minute TST window drives real architectural consequence: if target-folder assembly, CDE, and ROE analysis take 20 minutes, the TST has already moved. Terminal ATR seekers on JAGM, LRASM, StormBreaker, and JDAM-class weapons operate on sub-second budgets with no reachback at all. A single system rarely covers both ends — scope the latency envelope honestly.
Select the latency budget the decision-support pipeline must hold for each mode.
Single choice
Trinidy — For TST and dynamic targeting, seconds of latency are mission-critical. On-platform ATR inference and on-node target-folder assembly keep the full pipeline inside a classified enclave — no cloud round-trip, no link dependency, and predictable latency when the sensor platform is forward of reachback.
Confirm positive identification (PID) standard
Why This Matters
Positive identification is the LOAC distinction principle operationalized. The DoD Law of War Manual (updated 2023) treats PID as the prerequisite to lawful engagement against anything that is not clearly and continuously a lawful target. AI classification confidence is not PID on its own — it is a contributor to a PID judgment that remains a human responsibility. Programs that conflate "high classifier confidence" with PID create an unreviewable liability surface.
Note prompts — click to add
+ Is our PID standard documented and accepted by the legal advisor and the operational commander?+ Does the UI clearly distinguish AI classification confidence from PID status?+ When PID cannot be established, does the workflow force a hard stop rather than a soft warning?Document the PID standard the system must support.
Single choice
Specify classification level and enclave residency
Confirm which classified networks and enclaves the system operates within.
Select all that apply
Trinidy — Target libraries, intent signatures, and ROE are typically SECRET to TS/SCI. On-node residency inside the classified enclave — with no cross-domain flow to commercial cloud — is the only deployment posture that avoids cross-domain solution overhead and ITAR exposure.
Document weapons-review legal gate (DoDD 2311.01 / DoDI 5500.15)
Why This Matters
DoDD 2311.01 (DoD Law of War Program) and DoDI 5500.15 require a legal review of the acquisition or procurement of weapons and weapon systems to ensure consistency with US obligations under the law of war, including Geneva Conventions Additional Protocol I Article 36 where applicable. AI-enabled targeting support falls under this review, and the review must be completed before fielding. Programs that treat the weapons review as a paperwork step at the end of development frequently discover design choices that the reviewer cannot clear.
Note prompts — click to add
+ Has the cognizant Service JAG / General Counsel initiated the weapons review?+ Are design decisions being captured in a format the reviewer can cite (inputs, outputs, human control points)?+ What is the reviewer's position on the specific AI components being fielded?Confirm the Article 36 / DoDD 2311.01 weapons-review pathway is engaged.
✓ savedAlign with the DoD Political Declaration on Responsible Military Use of AI
Why This Matters
The Political Declaration on Responsible Military Use of AI and Autonomy was launched by the US in February 2023 and had 50+ state endorsers by early 2024. It establishes principles including auditability, human responsibility, explicit engagement boundaries, and testing of AI-enabled military capabilities. While non-binding, it is the governing statement of US military-AI responsibility norms and US partners expect interoperable systems to align with it. Alignment is a design artifact, not a slide.
Confirm system design aligns with the Political Declaration (February 2023).
✓ savedDefine acquisition pathway under DoDI 5000.02
Select the Adaptive Acquisition Framework pathway governing development.
Single choice
Coalition and releasability posture
Specify ITAR/EAR and allied-sharing constraints that shape the deployable configuration.
Select all that apply