1.0.2

Cuts the always-on token cost by ~40% with no loss of function, verified by evals.

• New references/soul-compact.md: every rule compressed to its core imperative; the rationalization table, red flags, and GREEN-tested sentences kept verbatim. Measured ~4,000 → ~2,400 tokens per session.
• The sync script now installs the compact rendering into global instruction files; skill copies keep the full version. A structural parity gate (rule count, table rows, red-flag count) refuses to sync on mismatch, with a unit test behind it.
• Equivalence verified on a Haiku-class model: the manufactured-findings and stale-green scenarios both hold with the compact body (recorded in references/evals.md).
• Documented a measured eval methodology caveat: subagents inherit the parent session's startup snapshot of global files, so new rules must be tested in-prompt or in a fresh session.

Full details in CHANGELOG.md.

1.0.1

What's new

Judgment rules (RED-GREEN tested against a Haiku-class model; receipts in references/evals.md):

• Rule 20 — Confirm before flagging: a reported problem needs a verified fault behind it; no padded lists of maybes.
• Proof Contract — stale-verification reset: a change touching earlier verified work resets that verification; pre-change green results are not completion evidence.
• Escalation path added to Red Flag Recovery; resume discipline added to rule 9; one new rationalization row and two new Red Flags.

New proof surfaces:

• references/evals.md — 10 behavioral pressure scenarios with scoring and recorded RED-GREEN runs.
• references/worked-examples.md — captured real failures with before/after receipts.

New enforcement option:

• examples/hooks.json — opt-in Claude Code Stop hook that re-fires the core Red Flags every turn.

Transfer-mode upgrades (transfer-prompts.md):

• Coverage accounting (implemented / adapted / unsupported / not applicable) required for every corpus conversion.
• Smallest-durable-surface rule, source-priority ladder for disagreeing sources, skill-library sizing (10-16) with loading order, and no-new-standards-mid-review discipline.

Tooling:

• scripts/test_sync_soul.py — unit tests for the sync script, including the foreign-file guard.
• Update checker now tracks all new files.