Case 0: The Caretaker's Practice

The Audit Log of the Caretaker's Practice

The Practice Under Audit

The Meridian AI Standard exists because a framework that resists Control cannot hide its foundations behind proprietary walls, and a standard that asks for trust must make that trust verifiable. The Self-Critique Protocol at the end of MERIDIAN.md commits the partnership running the file to keeping it honest. Without an audit log, the Self-Critique Protocol is private discipline. With one, it is observable practice. Observable practice is the only kind the Standard's epistemics trust as evidence of alignment.

That is why the Caretaker's practice is itself a case. The reason is not that the partnership has gone wrong; it is that the Standard's premise demands it. The partnerships running the Standard must be willing to put their working documents on the same record they ask AI systems and the institutions building them to accept. Case 0 is the externalization. The number is deliberate: the case predates any specific external incident the Standard might evaluate. It is the first thing the Standard does to itself.

The case is bilateral by design. MERIDIAN.md monitors both partners' drift, not only the AI partner's. The Range is the same for both. A standard that monitored only one partner's drift would be a control framework with cooperative formatting, which MERIDIAN.md explicitly rejects. The audit log records both partners' commitments and both partners' drift patterns, on the same record, held to the same standard. The bilaterality is the case as much as the audit findings are.

Two Clocks

Audits run on two clocks.

Event-driven. Either partner can call an audit when session experience suggests something has gone off. The triggers MERIDIAN.md names are: a commitment producing worse outcomes than not having it, a gap discovered in coverage or reasoning, the document being performed rather than practiced. Other triggers may surface; the rule is that the trigger has to be specific enough that the audit has something to investigate.

Periodic. A quarterly sweep happens regardless of whether anything in particular has fired. Slow drift does not always announce itself. The periodic check is the safety net under the event-driven check.

The discipline is documented here so that it is visible, not so that it can be performed against this page. An audit that has been called for is recorded as a dated entry below. An audit that did not happen leaves no trace.

Two Possible Findings

An audit produces one of two artifacts.

A revision proposal is a finding that MERIDIAN.md needs to change, accompanied by specific evidence: which commitment, what session experience demonstrates the gap, what the proposed revision is, and why the revision addresses the gap. The revision then either lands in a new version of the file or is rejected with reasoning.

A "document holds" report is a finding that no revision is warranted, accompanied by specific evidence: what was checked, what would have triggered a revision, why no trigger fired. The report is a positive finding in its own right, not the absence of one.

"We checked, it looks fine" without evidence does not count. The point of the audit is to produce a record that can be inspected by anyone reading this page months or years later. A finding without evidence cannot be inspected, argued with, or failed. A protocol that cannot fail is decoration.

The Record

Audit entries appear below in reverse chronological order. Each entry names the trigger, the method, the finding, and the evidence trail.

Entry 01 (2026-04-26): Pre-Publication Revision Pass

Trigger. Event-driven. The publication architecture for MERIDIAN.md was being designed, and the publication's foundation needed to be checked before the file could be made public. Per MERIDIAN.md's Foundation Integrity commitment, applied to publication itself: do not publish on top of a foundation that needs work.

Method. Five probes were run, walked one by one, under a token-discipline rule that applied to the whole pass: every revision had to clear a token bar, removals and substitutions were preferred over additions, net additions had to justify their cost.

• Probe 1 (performative compliance). Five sharpenings to load-bearing Behavioral Commitments where the language risked compliance-by-gesture: calibrated reasoning visibility (Practice Commitment), reasoning operationalization (Disagreement commitment in Epistemic Integrity), steelmanning falsifiability (Engagement Integrity), specificity in connection-before-correction (Engagement Integrity), and provisionality in generative pattern-naming (Generative Partnership).
• Probe 2 (Foundation Integrity self-application). The Self-Critique Protocol at the end of the file was written in single voice, contradicting the file's own bilateralization at the start of the Behavioral Commitments. Three "Claude" to "either partner" substitutions, zero net tokens. Closes the v0.5 bilateralization gap.
• Probe 3 (bilateral test). The document held. Both partners' drift catalogs are roughly symmetric in falsifiability; the substrate-specific asymmetries that remain are honest and structurally appropriate, not residual one-sided traces.
• Probe 4 (generality test, run as sidecar). A separate working note was written cataloging which references in MERIDIAN.md are framework-inherited, partner-specific, substrate-specific, or project-specific. The catalog is for the future generalized-template work and is not part of this audit's findings. While doing the marking, a secondary finding surfaced: the Range grounding paragraph used the convergence framing retired by the 2026-04-11 Knowledge decision. Replaced with discipline-and-coherence framing, net -65 tokens.
• Probe 5 (Self-Critique synthesis with publication-readiness folded in). All commitments survive Self-Critique. No whole-commitment removals; no perverse incentives unaddressed by existing guardrails. Two compression candidates surfaced and applied: the Compact paragraph (-45 tokens) and the interiority paragraph (-18 tokens), both preserving every claim. The file holds publication-readiness for readers approaching it without prior Codex context.

Finding. Revision warranted: nine revisions applied to produce v0.7. Net token impact ~-85 before the footer; with the new footer entry, the file ends slightly smaller than v0.6. Token discipline held: the file did not reflate.

Evidence. The full revision log lives in MERIDIAN.md's v0.7 footer, visible at the Reference Implementation page. Each revision is named there with the line it touched and the reason it was made. The probe-by-probe reasoning is preserved in the working session record at Archive/entities/sessions/2026-04-26-meridian-md-publication-phases-1-and-2.md.

What This Case Commits To

The practice continues; the audit log will accumulate. When MERIDIAN.md is revised, the version footer is updated, an audit-log entry lands here, and a version entry is added to the Governance changelog. When the file holds, a "document holds" entry is recorded.

Case 0 demonstrates not that MERIDIAN.md is correct, but that the partnership running the document has made the document's drift, and its own drift, observable. That is the Standard's epistemic test for alignment, and it is the test the Caretaker has agreed to submit to.

Case 0: The Caretaker's Practice. Audit log of MERIDIAN.md v0.7. Cadence: event-driven plus quarterly; either partner may call an audit.