Warfighter Evidence Brief

Use this skill when someone holds a claim about what a unit needs (a statement of need, capability description, requirement, change request to a vendor, after-action comment, or test report behind one) and wants it graded before anyone buys anything.

This skill is Labor as Code™ for the evidence-labeling method in You Are Not the Warfighter. Labor as Code is a skill that encapsulates a person's knowledge, judgment, taste, and skill in durable, inspectable, versioned form. The know-how is the habit of asking three things of every claim about what a unit needs, before anyone argues about it: how strong is the evidence, how far from the fight was it gathered, and who wrote it.

The method spans two chapters. "The Farther a Test Is From the Fight, the Less It Proves" teaches Distance to the Fight and attaches the Simulated User Disclosure to results gathered away from the fight, worked on the Larkspur Brigade drone cell's range test against its training-center rotation. "Grade the Evidence for a Unit’s Need Before Anyone Buys Anything" teaches the User Evidence Ladder with defense examples and the Warfighter Evidence Brief, worked on the Cairn Division fires cell's input-screen claim. Both extend You Are Not the User, whose tools this book uses by name and never redefines. Read the chapters for the argument and the worked cases. This skill runs the same labels over the user's own document and never reuses the cases' numbers as the user's.

The skill grades evidence, never people. It never raises a label because a claim sounds confident, because a senior officer made it, because a contractor wrote it well, or because it has been repeated in several briefings. A general's belief about what soldiers prefer is Level 0. A staff sergeant with a stopwatch and a count of timed missions holds Level 4. It never invents evidence, units, people, quotes, requirements, numbers, authorities, legal conclusions, or test results. A gap is a finding.

The skill can't observe warfighters, and it isn't one. Observation is a person watching real work. The skill can read notes a named person wrote while watching and label them Level 4. It can't produce Level 4 evidence. An AI asked to play a warfighter and critique a design answers from text people wrote about the work; that simulation is D4 at best and carries a Simulated User Disclosure like any other simulated test.

Before running

1. Apply the book's security rule first. Use only unclassified, publicly releasable material, in an AI tool the user's organization has approved for it. No capability, vulnerability, or operational detail: no frequencies, ranges tied to named systems, countermeasure specifics, tactics, unit locations, or schedules. If the user pastes anything marked CUI, FOUO, or classified, or anything that reads like operational detail, stop, say so, and ask for a releasable version. Nothing controlled goes into the AI Workforce™ running this skill, including its files, connections, and memory.
2. Get the actual text. The statement of need itself, plus the notes, test reports, and after-action comments behind it. Not a summary. Label only what the user supplies; don't assume a count or a test exists because someone said "the data shows" or "we tested it."
3. Ask what decision the document supports: what to go look at, a trial, a buy, a fielding decision.
4. Ask who wrote each part (the unit, a requirements staff, a program office, a contractor, a vendor), and, for any test, each tester's current job. The skill can't tell a former operator from a serving one. Unknowns read "not in the records."

Step 1: Find every claim and who wrote it

1. Read the whole document once before labeling anything.
2. Quote, word for word, every sentence that claims something about what a unit or warfighter needs, does, or suffers, including claims in disguise: "units are asking," "operators report," "we tested it."
3. Record the warfighter role and unit type each claim names, or "no role named."
4. Record the source the document gives, or "no source stated."
5. Record who wrote each claim. This is a field because the law cares. FAR 7.503 lists "determining what supplies or services are to be acquired by the Government" among inherently governmental functions, and lists "services in support of acquisition planning" as approaching the line. Contractors can advise, analyze, draft, and build; they can't own the decision about what the government buys. Where a contractor wrote a claim, say so plainly and without accusation: it's an input to a government decision, and the unit should restate the need in its own words with its own evidence. Where the unit wrote it from its own observation, say that too.

A contract clause, a buyer's specification, or a developmental test pass is not evidence about warfighters at any level. Record it, name what it is ("contract requirement," "evidence the build meets its specification"), and keep it off the ladder.

Step 2: Assign a User Evidence Ladder level

Use the six levels and their exact names:

Rules for every claim:

1. Base the level only on the evidence stated.
2. Grade the evidence, not the speaker. Rank in the organization has nothing to do with rank on the ladder.
3. Count sources, not repetitions. One program office repeating a unit's complaint in five meetings is one report heard five times.
4. Restating a claim doesn't raise it. An anecdote rewritten as a requirement is still Level 1.
5. Level 3 needs counted behavior in the records. "The data shows" with nothing counted is Level 0.
6. Level 4 needs a named person who watched or timed the work.
7. Level 5 needs a measured outcome that changed after an intervention, against a stated baseline.
8. Mark a claim "unsupported" when the records state no evidence.
9. Give the reason for each label in one sentence. The ladder ranks what evidence can carry, not how true it is.

Step 3: Assign a Distance to the Fight to every test result

A claim that came from a test carries two labels, written after the claim the way an engineer writes units: "(Level 4, D0)."

1. Assign a distance to each tester or trial, not to a report as a whole.
2. Don't assign D0 or D1 unless the records show the testers are warfighters doing the job now. Don't assign D0 unless the fight's conditions and a thinking opposing force were present.
3. Contractor and developmental test are D4 by definition. That's what they're for. Say so without scorn.
4. Any result at D2 to D4 is simulated test. Replace "tested" or "passed" with "simulated test: passed."

Step 4: Write a Simulated User Disclosure for every D2–D4 result

Use the core book's form, unchanged. Where the records are silent, write "not in the records."

SIMULATED USER DISCLOSURE: [test name, date]
Who simulated which role:              ____
Distance to the Fight (per tester):    ____
What they know that the warfighter wouldn't: ____
What the warfighter knows that they don't:   ____
Conditions absent (volume, interruptions, real data, exceptions,
  jamming, fatigue, night, weather, degraded communications): ____
Decision this result supports:         ____
Decision this result does not support: ____

Leave "what the warfighter knows that they don't" for a warfighter to answer; the skill can only list the questions it raises. State the narrowest decision the result supports, such as a safety check or readiness for a trial with real units. Nothing is ready for the fight until warfighters at D0 or D1 have used it.

Step 5: Fill the Warfighter Evidence Brief

Identify the one claim the document exists to advance and fill its brief:

WARFIGHTER EVIDENCE BRIEF: [claim]
1. Claim about what a unit needs:        ____
2. Warfighter role and unit type:        ____
3. User Evidence Ladder level, with the evidence itself: ____
4. Distance to the Fight of any test:    ____
5. Who wrote the claim:                  ____
6. What evidence is missing:             ____
7. What result would change our mind:    ____
8. Smallest test to raise one level:     ____
9. Integrator Sort lane, if known:       ____
10. Decision requested, and who holds it: ____

1. Field 1 quotes the claim exactly. If it names a product rather than a problem, note it.
2. Field 3 gives the strongest evidence the claim has, with the evidence itself, not only the number. Field 5 comes from Step 1.
3. Field 7 is written before any new result exists, by the unit. The skill may propose a candidate, marked "draft: the unit to confirm or rewrite." A usable measure names a number and a setting. "Soldiers accept it" is not a measure.
4. Field 8 names who, how many, where, what gets measured, and the result that would count against the claim, usually a D0 or D1 trial.
5. Field 9 reads "unknown" unless someone has run the Integrator Sort (the integrator-sort skill runs it). Never assign a lane here on the skill's own judgment.
6. Field 10 copies the decision the document requests, exactly, and names who holds it: a commander, a portfolio acquisition executive (PAE), or a contracting officer. If the records don't say, write "decision holder to be confirmed." Some claims need two holders, one for the portfolio and one for the contract.

The brief doesn't replace any legal document. It travels with whatever the Department now uses in place of the old requirements documents.

If the material is after-action notes, sort first, in the terms of Zero-Bureaucracy Product and Project Management: an observation is what someone saw, a hypothesis is a claim about why that a test could prove wrong, a requirement is something the system must do, and a decision is someone with authority choosing. Keep each note with its author and date, and never promote a note into a requirement.

Step 6: Conclude with a recommendation

Close with a short statement, not an evidence pack. It has four parts.

The decision the evidence supports now. Level 0 is enough to decide what to go look at. Level 1 is enough to go look. Level 2 shows a need is widespread without saying what it is. Level 3 shows what people do at scale, not why. Level 4 explains why for a few people, enough for a bounded trial. Level 5 supports a larger decision, and its Distance to the Fight says whether it has met the fight's conditions yet. The book's rule: no statement of need asks for a bigger decision than its evidence can carry, and a Level 0 or Level 1 claim asks for a test, not a buy. If the document asks for more, say it plainly: "This is a Level 1 claim at D4 asking for a buy."

The weakest claim the decision depends on: the one whose failure would sink the request and whose level or distance is weakest.

The smallest test that would raise that claim one level. Be specific, and use the book's moves: count independent reports to reach Level 2; instrument the workflow to reach Level 3; have a named person watch and time warfighters doing the task to reach Level 4; run a bounded trial against a stated baseline to reach Level 5; take a D1 result to the next D0 exercise already on the calendar. Say where a test needs a human observer or a unit. The skill can't run it.

The reasoning, in two or three sentences tied to specific labels, including who wrote the claim.

If two readings are both defensible (a counted pattern and an observed one point at different needs, or a D1 measurement and a D0 observation disagree), present both side by side, each with its reasoning and the test that would settle it. If the evidence supports nothing beyond "go look," say that, and name what would change it. Don't round Level 0 and Level 1 claims up to a case for buying.

Checks before calling the work done

• Every quote matches the source word for word. Spot-check three.
• No number, unit, person, or test appears that the user didn't supply. Delete any that does.
• No label was raised for rank, confidence, or repetition. Re-read every Level 2 claim for independence and every Level 4 claim for a named observer.
• Every claim has an author or "not in the records."
• Every D2–D4 result has a disclosure, and none supports a broader decision than its evidence.
• The recommendation names a decision, a claim, and a test.

What this skill does not do

It doesn't decide whether to buy, build, or field anything; the commander, PAE, or contracting officer named in field 10 does, with the brief in hand. It doesn't assign an Integrator Sort lane, a gate, or a rights clause as final. It doesn't treat its own reading of a report, or its own role-play of a warfighter, as observation. And it doesn't write field 7 for the unit. When the evidence is thin, the correct output is "this is Level 1, and here is the trial that would raise it," not a confident case built to fill the silence.