Quality
2 specifications, 74 requirements. All libraries
0001 Quality evidence
| Requirement | Mechanism | Who | Note |
|---|---|---|---|
FR-001 | check | bun test suite/ | host-test-run.test.ts: records appended per test, a killed process gives cancelled |
FR-002 | check | bun test suite/ | host-test-run.test.ts: one record file per run, gathered from child processes |
FR-003 | check | bun test suite/ | host-test-run.test.ts: the exit rule, an unproven requirement fails |
FR-004 | check | bun test suite/ | host-test-run.test.ts: the record read back by the decoder, its digest |
FR-005 | check | bun test suite/ | test-report.test.ts: TAP 14, JUnit XML, CTRF, SARIF and OSCAL renderings |
FR-006 | check | bun test suite/ | test-report.test.ts: the completion report in Markdown and HTML with the trace matrix |
FR-007 | check | bun test suite/ | test-report.test.ts: counts equal across renderings (property), rendering twice the same bytes |
FR-008 | check | bun test suite/ | host-test-run.test.ts: the ledger entry and a signed in-toto Test Result |
FR-009 | check | bun test suite/ | host-test-run.test.ts: verify fails on a changed record, a broken chain, a bad signature |
FR-010 | check | bun test suite/ | host-test-run.test.ts: --require-review for high-risk requirements |
FR-011 | check | bun test suite/ | quality-branch.test.ts: the quality branch holds OTLP records and nothing else (a result as JUnit XML is refused by the chain); that no other channel carries a result in a run's pipeline stays with 0006-signals |
FR-012 | check | bun test suite/ | quality-ledger.test.ts, quality-system.test.ts: a CAPA is closed only by a named person: an automated run, an agent, the one that opened it and, where a regulated domain is declared, the author of the corrective change (read from Git) are each refused |
FR-013 | check | bun test suite/ | quality-record.test.ts, quality-command.test.ts: a planned case that did not run is not_run with its reason, never passed, refused without a reason, counted in the TAP, JUnit, CTRF, SARIF and completion renderings and by `quality record` |
FR-014 | check | bun test suite/ | quality-ladder.test.ts, quality-system.test.ts: a count of proved requirements is shown with the grade it is counted at and the review debt beside it, and refused without either; the completion report, the register view and the index carry both |
FR-015 | check | bun test suite/ | quality-ladder.test.ts, quality-ledger.test.ts, quality-system.test.ts: a failed run and a stale proof open a nonconformance by themselves (once each), with the requirement, the evidence and the reason; freshness --open writes them to the quality branch |
FR-016 | check | bun test suite/ | quality-ladder.test.ts: a requirement is counted proved at E3, and at E4 where one of its tests is of high risk (an E3 run of a high-risk test is not counted); quality-regulated.test.ts: gradeNeeded is E4 for every requirement of a harness that declares a regulated domain and production (the day is an argument), the ladder, the tier claim and quality tier use it |
FR-017 | check | bun test suite/ | quality-record.test.ts: the runtime records ifc.test.characteristic (an ISO/IEC 25010 word) and the reports bear the 29119-3 document names (completion, status, incident) |
FR-018 | check | bun test suite/ | quality-record.test.ts, quality-command.test.ts: the body of each case's log record is a CTRF test object; JUnit, CTRF, SARIF and TAP are rendered on demand, byte for byte the same, and not kept |
FR-019 | check | bun test suite/ | quality-record.test.ts: ifc.test.outcome takes not_run with ifc.test.reason, CTRF renders it as other, and a requirement whose only proof is not_run is unproven; the outcome words equal the ifqs ontology's |
FR-020 | check | bun test suite/ | quality-record.test.ts: checkRun leaves each rule as a case and each finding as a failed case with rule id, file and line, rendered as SARIF results (a rule that could not run is not_run); the Mizan and changes commands calling it are Planned, M2 and M7 of issue #10 |
FR-021 | check | bun test suite/ | quality-branch.test.ts, quality-command.test.ts: one new file per run at quality/evidence/YYYY/MM/DD/<trace id>.otlp.json on an orphan branch in a commit that only appends, never replaced, signed through the host's SSH signing, a force-push and a deletion refused by a repository set to refuse them |
FR-022 | check | bun test suite/ | quality-branch.test.ts, quality-record.test.ts: ifc.quality.previous.sha256 on every resource (64 zeros first) and the verifier names the first file that does not match |
FR-023 | check | bun test suite/ | quality-branch.test.ts: a note under refs/notes/quality on the tested commit gives the file's path, SHA-256 and grade; no note is E0 and a note that does not match its file breaks the evidence |
FR-024 | check | bun test suite/ | quality-ledger.test.ts, quality-system.test.ts: the index is derived from the quality branch, deterministic, checkable with --check, and rebuilt in a repository that holds only the branch to the same bytes |
FR-025 | check | bun test suite/ | quality-record.test.ts: secrets and e-mail addresses are redacted before a record is kept and a record that names a person is refused |
FR-026 | check | bun test suite/ | quality-ladder.test.ts, quality-system.test.ts, ifqs-ontology.test.ts: the six grades E0 to E5 exist in order with their ranks; a requirement is graded the highest grade of its fresh evidence (a statement E1, a review E2, a run E3 or E4, a probe E5), and stale evidence does not count |
FR-027 | check | bun test suite/ | quality-ladder.test.ts, quality-system.test.ts: quality register shows the proof register with the grade, the day and the freshness of each row, generated and checked against the file |
FR-028 | check | bun test suite/ | quality-ladder.test.ts, quality-system.test.ts: the review debt is the count of requirements held at E1 or E2, shown beside every count of proved requirements, and growth with no Decision fails |
FR-029 | check | bun test suite/ | quality-ladder.test.ts, quality-system.test.ts: a review by an agent is recorded as an agent's and graded E2 at most; a person's sample is recorded as a sample |
FR-030 | check | bun test suite/ | quality-ladder.test.ts, quality-system.test.ts: a proof is Fresh, Due or Stale by its cadence and Stale by change where a file of its chain, the requirement's own line or its commit's place in the history changed (real Git); the freshness words are also held by ifqs-ontology.test.ts |
FR-031 | check | bun test suite/ | quality-ledger.test.ts: a nonconformance carries its requirement, evidence, reason, opener and state, and is linked to a Problem change request (an unlinked one is listed as such) |
FR-032 | check | bun test suite/ | quality-ledger.test.ts: no event edits a state; the event that closes a nonconformance follows only a closed CAPA |
FR-033 | check | bun test suite/ | quality-ledger.test.ts, quality-system.test.ts: a CAPA states its corrective action and names a test that exists and proves the requirement (a name that does not exist is refused) |
FR-034 | check | bun test suite/ | quality-ledger.test.ts, quality-system.test.ts: a CAPA becomes verified only by the run that passes its named test at a revision that holds the corrective change (quality record does it); a failing, earlier or unrelated run verifies nothing |
FR-035 | check | bun test suite/ | quality-ledger.test.ts, quality-system.test.ts: a CAPA is closed only from verified and only by a named person's Decision with the person and the day; its nonconformance closes with it |
FR-036 | check | bun test suite/ | quality-ledger.test.ts, quality-system.test.ts: the index lists a nonconformance and its CAPA from the moment they are opened, and a closed one stays; the public page is 0022-trust's, Planned, M9 |
FR-037 | check | bun test suite/ | quality-ladder.test.ts: the success criteria of 0021-ifqs are read as objectives (a metric named as OpenTelemetry names it, a target, a unit, an error budget, a window) whose measure is computed from the evidence and emitted as OpenTelemetry metrics |
FR-038 | check | bun test suite/ | quality-ladder.test.ts, quality-system.test.ts: a spent error budget opens a nonconformance of its own, and the remaining budget is shown with the objective |
FR-039 | check | bun test suite/ | quality-ladder.test.ts: commons declares the suggested objectives (the share at E3, the review debt, not_run, the oldest open nonconformance, the stale proofs, the figures with no grade) |
FR-040 | check | bun test suite/ | quality-ladder.test.ts, quality-system.test.ts: quality tier judges a claim against the requirements of the tier and of those below it in the register |
FR-041 | check | bun test suite/ | quality-attestation.test.ts, quality-tier2-command.test.ts: a person's attestation is a recorded sample (the rule and seed, the population's count and digest, the items, the result of each, the person's name and role and the day) kept as chained records, drawn again from its seed; signing is the commit's, by Git |
FR-042 | check | bun test suite/ | quality-attestation.test.ts, quality-tier2-command.test.ts: a plan is registered on the branch before its run, the run records its hash (ifc.quality.plan.sha256), and a run that followed it has no finding while a plan registered after, an unknown plan, a case outside it and a planned case not run are named; notRunOfPlan makes the not_run cases |
FR-043 | check | bun test suite/ | quality-attestation.test.ts, quality-tier2-command.test.ts: a mutation report of the tool it names becomes a record, the strength of each proof is shown beside its grade and a strength below the floor lowers a run's grade one rung; the run itself is the tool's |
FR-044 | check | bun test suite/ | quality-attestation.test.ts: an agent's review is graded E2, below a test run's E3, a person samples the agents' reviews at the stated rate, and a sample owed, an unattested sample and a review a person found wanting are findings |
FR-045 | check | bun test suite/ | quality-branch.test.ts: a run kept unsigned is E3 and one kept in a signed commit E4 (SSH key made by ssh-keygen), a note is path, sha256 and grade with one paragraph per run, and a note that does not match its file or a chain that does not verify is E0 |
FR-046 | check | bun test suite/ | quality-ledger.test.ts: each event is one redacted file quality/records/NNNNNN-<kind>.json on the quality branch, chained to the file before it, found when changed, removed or inserted, and signed where Git is set to sign |
FR-047 | check | bun test suite/ | quality-ladder.test.ts: the default cadences are 30 days for a run, 7 for a probe, 90 for a statement and a year for a review |
FR-048 | check | bun test suite/ | quality-attestation.test.ts: the records of Tier 2 are files of the ledger in its chain and an event that does not follow is refused (an attestation without a result on each item, no name, a second one; a plan with no case or a case with no acceptance criterion; a mutation row that does not account for its mutants); a sample is drawn again from its seed and a run names its plan by hash |
FR-049 | check | bun test suite/ | quality-attestation.test.ts: the strength of a proof is the share of its mutants detected, and below the floor a run's grade is lowered one rung with both shown; a proof no mutation run measured keeps its grade and says so; quality-regulated.test.ts: the floor is 600, 800 for a regulated harness in production (a stated floor never lowers it) and it decides whether the grade is lowered |
FR-050 | check | bun test suite/ | quality-retention.test.ts: the record states retentionDays (a whole number of at least one, read from ifccm.declaration.ts); the period of the oldest evidence has passed on the day it ends and not before (the period and the day are arguments); quality archive --tag tags the tip of a real quality branch, never over a tag that exists, refuses where no period is stated, and deletes nothing (no option does: --delete is a usage error); that the period is not shorter than the longest that applies to the evidence is the harness's to state and no check reads it |
FR-051 | check | bun test suite/ | quality-ladder.test.ts: the review debt of agents and of a person are counted apart, add up to the total and are both shown in the count; quality-trust.test.ts: the page's JSON and HTML carry both |
FR-052 | check | bun test suite/ | quality-trust.test.ts, quality-tier2-command.test.ts: the page is generated from the quality branch and --check fails where it differs, a page edited by hand included |
FR-053 | check | bun test suite/ | quality-trust.test.ts: each requirement with its grade, freshness, day and run, the review debt, the objectives with their budgets and the counts of passed, failed and not_run; one self-contained HTML file |
FR-054 | check | bun test suite/ | quality-trust.test.ts: failed cases, cases that did not run, open nonconformances with their age and open CAPA come before any figure or success, and none is dropped |
FR-055 | check | bun test suite/ | quality-trust.test.ts, quality-tier2-command.test.ts: the scope, the revision, the day and the span, and what is not covered; a statement of conformity, certification or compliance in what the harness wrote is refused |
FR-056 | check | bun test suite/ | quality-trust.test.ts: where the newest run is older than the cadence of a run (30 days, 0021-ifqs FR-044) the page says that it is stale and by how many days |
FR-057 | check | bun test suite/ | quality-trust.test.ts: a rendering is made for one audience, and the audiences are the code list ifqs:Audience (Public, Assessor, Internal) |
FR-058 | check | bun test suite/ | quality-trust.test.ts, quality-tier2-command.test.ts: the public page leaves out a customer's name, an address, a host, a path and a failure's message, and says how many values it redacted |
FR-059 | check | bun test suite/ | quality-trust.test.ts, quality-pack.test.ts: the same evidence and audience give the same bytes, and the counts, grades and digests of the public page equal the internal one's |
FR-060 | check | bun test suite/ | quality-pack.test.ts: the evidence files of the span with their chain, the ledger, the notes, the specifications at the tested revisions, the traceability matrix, the OSCAL results of the runs, the index, the signature and its keys, the verifier and a manifest of each file's SHA-256 and media type |
FR-061 | check | bun test suite/ | quality-pack.test.ts: the assessor's pack holds the full failure messages and the run files, marked with its audience and its redactions; a file that holds a secret refuses the pack and an e-mail address already in a kept file is named in the manifest, since changing a kept file would break its chain |
FR-062 | check | bun test suite/ | quality-pack.test.ts: the page and the pack are made from one collected evidence, and each figure of the page equals the same figure computed from the pack |
FR-063 | check | bun test suite/ | quality-pack.test.ts, quality-tier2-command.test.ts: the pack as a deterministic tar with its SHA-256 beside it and a DCAT distribution that names it |
FR-064 | check | bun test suite/ | quality-pack.test.ts, quality-tier2-command.test.ts: with no network, the manifest's digests, the chains, the signature of each envelope against the keys of the pack, the notes and the figures, naming the first thing that fails |
FR-065 | check | bun test suite/ | quality-pack.test.ts, quality-tier2-command.test.ts: a key the pack lacks, a file the manifest lists and the pack lacks and an unsigned pack are not_run with the reason, and the pack is not reported verified |
FR-066 | check | bun test suite/ | quality-pack.test.ts: the verifier is one typed file with no import, shipped in the pack with its command, run from inside the pack on this runtime and on Node 26 |
FR-067 | check | bun test suite/ | quality-pack.test.ts: a kept file that holds a secret refuses the pack and names the file; personal data a kept file retains is named in the manifest; the files the pack makes are redacted and counted; quality-redaction.test.ts: redaction is at the source: the ledger's events of every kind (Tier 1 and Tier 2), a run kept with a secret and an e-mail address, its note and the commit messages of the branch hold neither (a person's name stays) |
FR-068 | check | bun test suite/ | quality-trust-capa.test.ts: the Public page names a CAPA by identifier, state and named test and withholds its corrective-action text, counted among the redactions; quality-trust.test.ts: the public page names an open nonconformance by its identifier, its requirement's identifier and a cause code only (a private marker in its reason and the requirement's sentence never reach it), the Assessor and Internal pages show the requirement's text read from the specifications at the tested revision and the reason; the corrective action text of a CAPA is not yet withheld from the public page |
FR-069 | check | bun test suite/ | quality-verifier-locator.test.ts: the verifier files are found beside the module and in the src/assessor of the package from its compiled dist/assessor, and a missing file is a typed error naming the places; tests/release/quality-assessor-package.test.ts (the release): the packed tarball, installed, makes the pack on Node 26 and on Bun |