v2.1.1

Repository rename — no scoring or tooling changes; RULESET_VERSION stays
2.0.0.

Identity

• The repository is now
  emmanuel-a-otchere/measuring-and-managing-autonomy (lowercase,
  hyphenated). All live references updated: README, package metadata
  (project.urls), schema $id, the site's method link, package
  docstrings, and release/security/governance docs. The distribution name
  is unchanged (measuring-autonomy); the method name is unchanged
  (mama); the umbrella organization is unchanged.
• Historical changelog entries keep the URLs they were published with
  (GitHub redirects renamed repositories and Pages sites).
• The GitHub Pages site moves to
  https://emmanuel-a-otchere.github.io/measuring-and-managing-autonomy/
  (the old address redirects).

v2.1.0

Establishment follow-through (CR MAMA-AUT-CR-001, Phases 2 & 4). Additive
only — no scoring changes; RULESET_VERSION stays 2.0.0.

Added

• Crosswalks (docs/crosswalks.md): relation to TM Forum's same-named MAMA
  programme stated as independent and unaffiliated, with the TM Forum
  reference policy (GB1040 / GB1042A / IG1301 only; no IG1403 or
  Autonomous Networks / ANLET content); coverage-and-gaps mappings to NIST
  AI RMF 1.0, ISO/IEC 42001:2023, ISO/IEC 23894:2023, OECD AI Principles,
  OWASP GenAI Top 10, and MITRE ATLAS — no equivalence claims
• Shared scoring corpus (tests/corpus.json, 9 cases + invalid inputs)
  with per-engine checkers (tools/corpus_check.py / corpus_check.js)
  enforced in CI — the Python and JavaScript engines cannot drift silently
• Machine-readable scorecard export: docs/export-format.md +
  schema/mama-scorecard-export.schema.json (JSON Schema draft 2020-12)
• docs/facilitator-guide.md — running assessment sessions
• docs/training.md — self-guided ~60-minute training module
• MAMA-consumption sections in both worked examples: how a consuming MAMA
  project links use-case records and joins automation/value measures
  beside — never blended with — autonomy measures

v2.0.1

Establishment release (CR MAMA-AUT-CR-001, Phase 0). No scoring changes —
RULESET_VERSION stays 2.0.0.

License (corrective)

• The LICENSE file was BSD-3-Clause (© 2022) while every doc and the
  package metadata said MIT — a contradiction present since the original
  import. The intended license is MIT: the LICENSE file now carries the
  MIT text, and repository metadata, README, package, and distribution
  artifacts are reconciled. Prior distributions (v1.x–v2.0.0) shipped the
  BSD-3 file with MIT-claiming metadata; treat MIT as intended and
  re-download current artifacts where the distinction matters.

Identity

• Naming established: mama (lowercase) = measuring and managing
  autonomy, this method and toolkit (canonical home: this repository);
  MAMA (uppercase) = Measuring Automation & Measuring Autonomy, the
  umbrella framework that uses mama. Naming box added to README and the
  assessment site; integration guide published
  (docs/mama-integration.md).

Method governance & documentation

• GOVERNANCE.md (stewardship roles, decision rights, appeals, COI) and
  SECURITY.md (private vulnerability reporting)
• docs/handbook.md — the end-to-end operating guide
• docs/migration-v1-to-v2.md — re-baselining guide + historical-score
  policy (supersedes reading migration only from the changelog)
• docs/limitations.md + docs/validation-plan.md — the method is
  labeled pilot / under validation until the validation program
  reports
• README: current status (was stale "v0.1 draft"), intended users, offline
  use, links to all governance artifacts; roadmap reconciled with a v2.1
  establishment section

v2.0.0

MAJOR — re-baseline required. The level scale changed; assessments and
ledgers on the retired scale are not comparable with v2 scores.

Level scale (scoring change, RULESET_VERSION 1.0.0 → 2.0.0)

• Definitive autonomy levels are now L1–L5; L0 is the Submarine Line
  — no autonomy at all, nothing granted, nothing to review or revoke — kept
  as the grounding floor beneath the scale.
• Mapping from v1.x (old level n → new level n+1): v1 L0 Ask → v2 L1;
  v1 L1 Inform-after → v2 L2; v1 L2 Guardrailed → v2 L3; v1 L3 Supervised →
  v2 L4; v1 L4 Accountable → v2 L5. Coarse trend comparisons across the
  re-baseline may use this mapping; formal comparability starts with
  re-scored v2 baselines.
• Dimension scores are now 0–5 (was 0–4): a dimension with not even the L1
  (Ask-arrangement) item true scores 0 — Submarine Line for that dimension.
• The questionnaire gains an L1 item per dimension describing the Ask
  arrangement; former items shift up one level label.
• Weakest-link rule unchanged: min(D3, D4, D6). A safe level of 0 means
  nothing is grantable yet; the tools now say so explicitly (submarine move).

Ledger format (schema 1 → 2)

• schema: 2 = the v2 level scale. ledger_health.py validates levels 0–5.
• Schema-1 files still validate but warn: "retired 0–4 scale — re-baseline
  and set schema: 2". Missing schema: field = current supported schema.
• Reference ledgers (examples/ledger.yaml, examples/agents-ledger.yaml)
  re-baselined to schema 2 with levels shifted accordingly.

Docs & tooling

• All framework docs re-based: levels tables, maturity grid (now six
  columns), questionnaire, governance-loop thresholds, contract template,
  worked examples re-scored (payments team: safe L3, hold at L3; refund
  agent: safe L3, expand from L2).
• Site (site/) and CLI engines updated in lockstep; parity-tested at 2.0.0.

Migration (v1.x → v2.0.0)

1. Re-score each domain with the v2 questionnaire (autonomy-score prints
   ruleset_version: 2.0.0).
2. Rewrite ledger contracts onto the new scale (mapping above) and set
   schema: 2; version the contract id (-v<n+1>) rather than mutating
   history in place.
3. Do not compare v1.x scorecards to v2 scorecards directly — use the
   mapping for coarse trends only.

v1.1.1

Metadata only — no code or scoring changes (RULESET_VERSION stays 1.0.0).

Packaging

• PyPI project description: the README's repo-relative links rendered as
  broken pypi.org/project/... paths there; all links now use absolute
  GitHub URLs (identical behavior on GitHub and PyPI)
• Wheel classifiers corrected for PyPI's server-side validation —
  Intended Audience :: Management is not a valid trove classifier; replaced
  with Developers + Information Technology (the PyPI 1.1.0 artifacts
  already carried this fix; the GitHub Release artifacts rebuild with it here)

v1.1.0

Toolkit release: the framework tools are now installable and distributable.

Tooling

• Packaging: measuring-autonomy installs with autonomy-score and
  ledger-health console commands (src/ layout; the tools/ scripts
  remain as zero-install shims over the package)
• Release pipeline: pushing a v* tag builds the wheel + sdist and
  standalone one-file binaries (macOS arm64, Linux x86_64, Windows x86_64),
  gates every artifact on its --self-test and a CLI smoke, then publishes
  a GitHub Release with notes taken from this changelog (RELEASE.md)
• Optional PyPI publishing workflow (publish-pypi.yml, manual dispatch,
  requires one-time trusted-publisher setup)
• CI: install PyYAML for the ledger tooling — the runner's toolcache Python
  does not bundle it, which had left every CI run red (PR #6)
• Pages: the deploy workflow enables GitHub Pages itself; the assessment
  site is live at https://emmanuel-a-otchere.github.io/MeasuringAndManagingAutonomy/

Scoring semantics

None — RULESET_VERSION stays 1.0.0; no scored output changed.