Affected Verification

Affected Verification deterministically selects the smallest verification workload whose sufficiency can be defended from the available change-impact, dependency, coverage, policy, and risk evidence.

The operative claim is minimum defensible verification, not mathematical global minimality. Unknown impact is never permission to skip work.

This repository contains a dependency-free Node.js 20 planner. It consumes normalized change, impact, check-level dependency-completeness, verification-catalog, and policy data and emits an opsle.affected-verification.plan.v2 argument containing selected checks, skipped checks, boundary evidence, exact reasons, provenance hashes, uncertainty, escalation, and sufficiency. It plans work; it does not run commands.

opsle/tasks is a production consumer of this public plan contract. Tasks captures the staged BUILD tree, derives normalized input from a base-revision manifest, invokes the CLI without a shell, validates the returned action IDs and commands against that immutable input, and retains execution authority. AV output never creates an unrestricted command surface. An empty change set and an empty verification catalog are valid inputs so consumers can represent unchanged builds and repositories with no automated checks without inventing work; the result remains subject to the ordinary sufficiency and uncertainty rules.

Try it

node bin/affected-verification.js \
  fixture unrelated-large-suite \
  --receipt /tmp/av-receipt.json

Canonical plan JSON is written to stdout. The opsle.value-receipt.v1 is written only to the requested sidecar, and one [Affected Verification] indicator is written to stderr. The fixture reports exactly 14 of 1,043 test executions selected, 1,029 skipped, plus one lint and one typecheck; the test-execution reduction is an EXACT calculation, not a time, cost, token, or correctness claim.

npm run verify

Contract and evidence

• SPEC.md — normative prototype contract and sufficiency states
• PRIOR_ART.md — source-linked reconciliation with existing selectors
• ARCHITECTURE.md — adapters, project boundaries, shadow mode, and trust ramp
• BENCHMARK.md — controlled research method and calibration results
• benchmark/av-exp-001/REPORT.md — preregistered real-repository shadow calibration
• benchmark/av-exp-002/REPORT.md — cross-ecosystem Python shadow calibration and observed AV miss
• benchmark/av-exp-003/REPORT.md — opaque dependency boundary repair, adversarial corpus, regression matrix, and precision cost
• LIMITATIONS.md — current claim ceiling and non-goals
• fixtures/scenarios.js and fixtures/negative-cases.js — twelve positive/boundary scenarios plus explicit conflicting, malformed, impossible, and tampered cases
• schemas/plan-v2.schema.json — current check-level dependency-completeness plan shape
• schemas/plan-v1.schema.json — immutable historical plan shape

Status

AV's research evidence remains narrow and does not establish general selector completeness. AV-EXP-001 observed no AV miss in its frozen JavaScript corpus; AV-EXP-002 permanently observed one AV miss in its frozen Python corpus; AV-EXP-003 selected that known check under the repair and observed zero repaired misses in its generalized and frozen replay corpora. Tasks therefore treats targeted selection as authoritative only when the repository manifest declares complete impact, catalog, and check-boundary evidence. Unknown or incomplete evidence broadens to the full configured command or stops; it never silently becomes a passing verification result.

Apache-2.0.

Opsle Tasks capability package

The independently versioned Tasks capability package ships the executable planning/capture adapter, AV runtime, and explicit schemas. Build an installable artifact with npm run pack:tasks-capability. Generic trusted discovery and operator project grants activate it; installation defaults to disabled. AV remains independent and owns verification planning, while Tasks owns command execution and release.