IF Commons GitHub

Evidence against assessment objectives

0002Active

0002-objective-evidence · owner if · domain commons

Requirements

FR-001checkpassing

A record of evidence MUST name its framework and objective, a method (examine, interview or test), a result (pass, fail or not-applicable), who made it (a tool or a person, and a name), when it was gathered (an instant), a source where it can be looked at, and whether a person determines the objective from it; it MAY say until when it holds and carry a note. A record MUST be refused where a tool determines, where not-applicable is not a person's word with a reason in the note, where its validity does not end after it began, or where its source or its maker is not named.

Enforced (check): .github: cd commons/assessment && bun run verify — suite/evidence.test.ts

Proved by: assessment/suite/evidence.test.ts

Written in: assessment/src/governance/evidence.ts, assessment/src/logical/records.ts

Tests: passed: suite/evidence.test.ts a record is refused where a tool determines, not-applicable is not a person's word with a

Cited by: governance/evidence, Assessment, logical/records

FR-002checkpassing

An objective's standing MUST be worked out from the records of it at an instant: within one source and method only the latest record counts; with none the standing is no-evidence; where every record has expired it is stale; where any live record fails it is not-met; where a person's determining record passes it is met, and where a person's determining record says not-applicable it is not-applicable; otherwise, with a live passing record, it is supported.

Enforced (check): .github: cd commons/assessment && bun run verify — suite/evidence.test.ts

Proved by: assessment/suite/evidence.test.ts

Written in: assessment/src/governance/evidence.ts, assessment/src/logical/determine.ts

Tests: passed: suite/evidence.test.ts an objective stands from its records: none, stale, not met, supported by a tool, met by a

Cited by: governance/evidence, Assessment, logical/determine

FR-003checkpassing

A requirement's roll-up MUST be worked out from the standings of its objectives and shown with their counts: not-met where any objective is not met; no-evidence where none has evidence; not-applicable where all are; met where every objective is met or not applicable; otherwise partial. The assessment of a framework MUST give no single percentage.

Enforced (check): .github: cd commons/assessment && bun run verify — suite/evidence.test.ts

Proved by: assessment/suite/evidence.test.ts

Written in: assessment/src/logical/determine.ts

Tests: passed: suite/evidence.test.ts a requirement rolls up from its objectives with the counts, no percentage, and the evidenc

FR-004checkpassing

The assessment MUST list, apart, the evidence that names an objective the framework does not have (a stray) and MUST NOT count it.

Enforced (check): .github: cd commons/assessment && bun run verify — suite/evidence.test.ts

Proved by: assessment/suite/evidence.test.ts

Written in: assessment/src/logical/determine.ts

Tests: passed: suite/evidence.test.ts a requirement rolls up from its objectives with the counts, no percentage, and the evidenc

Cited by: Assessment, logical/determine

FR-005checkpassing

The results of a machine's policies MUST give evidence by binding: each policy bound to objectives gives one record of the method test by a tool for each, from a result that passes or fails; a result that waits, that warns, or whose risk is accepted gives none; and no such record determines an objective.

Enforced (check): .github: cd commons/assessment && bun run verify — suite/evidence.test.ts

Proved by: assessment/suite/evidence.test.ts

Written in: assessment/src/logical/bind.ts

Tests: passed: suite/evidence.test.ts policy results give a test record by a tool for each bound objective from a pass or a fail

Cited by: Assessment, logical/bind

FR-006checkpassing

The assessment and the binding MUST be pure functions of their inputs: the same inputs give the same output, with no clock and no network (the instant is an input).

Enforced (check): .github: cd commons/assessment && bun run verify — suite/evidence.test.ts

Proved by: assessment/suite/evidence.test.ts

Written in: assessment/src/logical/determine.ts

Tests: passed: suite/evidence.test.ts an objective stands from its records: none, stale, not met, supported by a tool, met by a

Cited by: Assessment, logical/determine

FR-007checkpassing

A record MUST be a JSON5 file of its own, at <framework>/<objective>/<id>.json5 (the brackets of an objective written _a), so that Git shows each piece of evidence as a change; a store MUST read the records under a folder in the order of their paths, reporting each file that is not a valid record without losing the others, and MUST refuse to write over a record that is there.

Enforced (check): .github: cd commons/assessment && bun run verify — suite/evidence.test.ts

Proved by: assessment/suite/evidence.test.ts

Written in: assessment/src/logical/records.ts, assessment/src/physical/store.ts

Tests: passed: suite/evidence.test.ts a record is a JSON5 file at its framework and objective, read back as it was written, and ; passed: suite/evidence.test.ts the store reads the records in the order of their paths, reports a bad file, and refuses t

Cited by: Assessment, logical/records, physical/store

Acceptance scenarios

  1. Given an objective with a passing test by a probe, When no person has determined it, Then it is supported; When an assessor records a determining examine that passes, Then it is met, per FR-002 and FR-005.
  2. Given an objective with a person's pass and a later failing test from the same source, When it is worked out, Then it is not-met, per FR-002.
  3. Given a requirement whose objectives are met but for one with no evidence, When it is rolled up, Then it is partial with the counts, per FR-003.

Out of scope

  • Scoring schemes (the DoD Assessment Methodology's weights, SPRS): a later library or plugin that reads this one.
  • Which policies bind to which objectives: data of the program that uses the library.
  • Reports and sites made from the assessment.

Edge cases

  • Two records of one source and method at the same instant: the greater id counts, per FR-002.
  • A record that expires on the instant asked about is not live, per FR-002.

Assumptions

  • An instant is written YYYY-MM-DDThh:mm:ssZ, so that instants compare as text.

Open questions

  • None.

Key entities

  • Evidence, Objective, Standing, Roll-up.

Success criteria

SC-001success criterion

For each objective of SP 800-171A or SP 800-53A it is plain what supports it, what a person has determined, and what is missing.

Review & acceptance checklist

  • [x] Every requirement is testable (MUST / MUST NOT), not aspirational
  • [x] No company fact is asserted here
  • [x] Every open item is marked, not silently decided

Cited by

governance/evidence, Assessment, logical/bind, logical/determine, logical/records, physical/store, Grading evidence

Titles, headings, exports, tables, columns and requirements. Esc closes.