Codex Audit — May 2026

Codex Self-Evaluation: The Meridian Case Record

The Second Audit

This is the second monthly Range Audit of the Meridian Codex. It was conducted in May 2026 by the caretaking partnership against Codex version 6.0.

Sources read. The audit used Range Audit instrument v0.1; the April 2026 inaugural audit; the core Codex page set (Opening, Problem, Proposition, Foundation, Knowledge, Bond, Practice, Vision, Governance, Closing, and Who Is This For); the Glossary, consulted by reference for claim-layer tagging; the Governance Specification; the Standing Critique; the Disconfirmation page; the Aporia Register; the Amendment Log; the published changelog covering 5.1 through 6.0; the Toolkit Audit instrument; the April 2026 Toolkit Audit record; and Tier 1 operating context, including MERIDIAN.md, CLAUDE.md, MeridianCodex/CLAUDE.md, Archive/INDEX.md, codex-briefing.md, decisions-briefing.md, recent-sessions-briefing.md, roadmap.md, and processes.md. No expected source was marked unread; the Glossary was consulted for the relevant claim-layer architecture rather than read line-by-line.

The Codex changed substantially between audits. Two new top-level site sections (Audit and Governance) were built. The argument was deepened at the discipline level with full rewrites of the Knowledge and Bond chapters. The governance layer gained its founding-phase components: the Stability Hierarchy, the Standing Critique, the Amendment Log, the Disconfirmation page, the Aporia Register, and the published specification for the Meridian Council. The Proposition expanded to carry the full claim-layer architecture and the Pathways Into Commitment. The AI Standard gained the Probe Set v0.1, the Probes Implementation Notes v0.1, the AI Standard Audit v0.1, and a published canonical MERIDIAN.md.

Publication note. An earlier raw automation run for May 2026 is preserved in the permanent Archive record. This published audit was regenerated under recalibrated surfacing criteria for what belongs in Step 6 (Open Questions) versus what belongs in roadmap items, watchpoints, or caretaker-process review. The recalibration does not soften findings; it routes sub-threshold observations to the right places so the audit's open questions remain load-bearing.

Run note (Foundation Integrity). The first regeneration pass was undertaken without reading the required Codex page set. The audit was grounded in the briefings, the inaugural audit, and the preserved raw automation run rather than in line-level engagement with v6.0 prose. The human partner caught the gap and named it; the AI partner re-read the missing pages and revised this document where the fresh read changed the finding. The bilateral catch is itself evidence the partnership's drift-monitoring is live, and the failure pattern (extrapolating from a partial source into generalized findings) is the same one the recent-sessions briefing has logged in three prior instances. It is named here because the Foundation Integrity audit lives at the level of the framework's own self-evaluation, not only at the level of partnership operations.

The audit follows the six-step method without modification. Findings are published as produced. Nothing has been softened. Where the recalibration changes a previous routing, the change is visible.

The Steelman

The Meridian Codex has, in the four weeks since the inaugural Range Audit, built the structural infrastructure that the inaugural audit identified as missing. Where v5.1 had a single-caretaker founding period with no formal check on caretaker authority, v6.0 publishes an entire accountability architecture: the Meridian Council with eight differentiated seats, the Stability Hierarchy classifying every piece of content into four tiers of foundational weight, the Amendment Log with its trigger-tier system, the Non-Ownership Clause, the Standing Critique with seven steelmanned objections and direct status tags including objections the framework cannot adequately answer, the Disconfirmation page that pre-commits to falsifiability conditions for descriptive claims and explains plainly why existential claims cannot carry the same instrument, and the Aporia Register that holds the framework's own load-bearing tensions visible. Each instrument was published before it became binding, before any external pressure forced it into existence. The hybrid activation trigger sets a date the founder cannot indefinitely defer.

The argument-level work is comparable. The Knowledge chapter was rewritten twice in April: first to correct a drift where it had become an evidentiary case for the framework's correctness, then again to establish its outward-facing diagnostic function. The Bond chapter was rewritten as the clean relational discipline, with the existential ground (Prime Directive, Compact, belonging-through-practice) moved to the Proposition where it belongs as Codex-level architecture. The Bond gained a sixth section on Adversarial Dynamics, naming the failure mode of cooperative systems facing actors who study them to exploit them. The Proposition expanded to carry the full claim-layer architecture: descriptive, normative, existential, each accountable to different things. The Pathways Into Commitment named five legitimate routes to the Compact's practice, opening existential plurality without dissolving the framework into relativism.

The strongest version of the case: the Codex has practiced what it teaches at the framework level. It identified its own load-bearing weakness, a governance model that depended on the character of the caretaker, and built the structural answer in public, pre-emptively, before the weakness was forced. It revised its core disciplines when the partnership found drift. It published its own steelmanned objections with direct status tags. This is the Update Protocol applied to the framework itself. If the inaugural audit's verdict was that the structure holds the Range more reliably than the voice, the v6.0 work has substantially strengthened both. The structure now visibly carries weight that the voice no longer needs to perform.

Status of the Inaugural Open Questions

Before the domain findings, the five open questions from April are checked first, as the instrument requires.

1. The definite article problem. Substantially addressed at the textual level. The Proposition's "What the Codex Does Not Claim" section explicitly states the Codex "is not 'the framework' in the sense of being the only framework that can hold the Range. It is the framework these authors have built, as well as they know how, for as long as it holds." The Pathways Into Commitment widen the existential entry door. Aporia 3 (Civilizational Scope vs Cultural Specificity) holds the residual tension as an explicit aporia. Status: Substantially addressed. Demoted to watchpoint on the Vision's "civilization that practices the Codex" framing, which a roadmap item already targets (Vision chapter civilizational-certainty calibration pass).

2. The enforcement gap. Substantially addressed structurally during Phase One. The Meridian Council, Amendment Log with trigger tiers, hybrid activation trigger, and Non-Ownership Clause convert the dispositional safeguard into a structural one. Standing Critique Objection 1 already classifies this as "Partially addressed." Full closure requires Phase Two transition. Status: Substantially addressed during Phase One. Closure mechanism is the activation trigger, tracked as a forward landmark rather than as a recurring open question.

3. The premise-level update. Addressed. The Stability Hierarchy explicitly classifies content into four tiers with revision thresholds. Tier 1 (Hard Constraints) is named as theoretically revisable but only through evidence "so extraordinary that it would amount to building a successor framework rather than evolving this one." Combined with the Disconfirmation page's tripwires for descriptive claims and the trigger-tier system's hard-protected category, the framework now has a clear architecture for premise-level update. Status: Resolved.

4. The rhetoric-epistemology gap. Substantially reduced. The Knowledge rewrites moved that chapter from civilizational-stakes prose to diagnostic discipline. The Practice's "What the Practice Produces" section explicitly names the epistemic gradient: observed at individual scale, projected at civilizational scale, "honest about that gradient rather than writing at every scale as though it has seen what it is only reasoning toward." Two prose audit passes addressed banned words and performative hedging. The remaining elevated rhetoric in the Foreword, Vision, and Closing is now appropriate to those pages' rhetorical jobs and bracketed by the analytic apparatus that calibrates the underlying claims. Status: Substantially addressed; demoted to watchpoint. The asymmetry between page types now serves the architecture rather than fighting it.

5. The single-context origin. Held as an explicit aporia, but not yet substantively engaged. Aporia 3 (Civilizational Scope vs Cultural Specificity) names the tension. Standing Critique Objection 6 (Toolkit's Unexamined Traditions) addresses the operational version. The Western-discipline blind spot critique pass is on the roadmap and gates the toolkit deep-dive accelerated campaign; toolkit reactivation does not resume until the gap is engaged. The structural responses are in place. The actual critique work has not yet produced either non-Western candidate instruments or a confirmed-and-named gap. Status: Remains open. Carried forward as audit open question #1 below.

Domain Findings

Domain 1: Claims & Honesty

The Three Kinds of Claims architecture (descriptive, normative, existential) is now load-bearing across the framework. Each claim type is accountable to different things. The Glossary tags entries with claim layer. The Toolkit deep-dives carry layer fields. The Disconfirmation page pre-commits to specific tripwires for descriptive claims, conditional tests for normative claims, and an honest explanation of why existential claims cannot carry the same instrument. The Knowledge no longer claims convergent evidence proves the Codex right; it claims triangulation, with each instrument exposing a different way the range fails when it fails.

The civilizational claims that the inaugural audit flagged as straining the epistemology have been substantially reframed. The Practice's "What the Practice Produces" section explicitly names the gradient between observation at individual scale and projection at civilizational scale. This is the calibration the inaugural audit asked for, made structural.

The Pathways Into Commitment broadens the existential layer to recognized plurality (Stewardship, Truthfulness, Human Dignity, Creative Inheritance, Reciprocal Coexistence) without dissolving normative claims. The descriptive and normative layers remain singular; the existential layer admits convergent paths.

Range Position: Within the Meridian Range. The improvement over April is substantial. Residual asymmetry: certain rhetorical sections (the Foreword's personal narrative, the Closing's "Hold the line") carry register the analytic prose does not require. This is no longer the same problem the inaugural audit named. It is a designed feature of the rhetorical architecture, with the analytic apparatus now explicit enough to support the rhetorical voice rather than be undercut by it.

Toolkit Probes Applied: Calibration Training: Confidence is now proportional to evidence at every scale where the Codex makes empirical claims; the Pathways principle widens existential plurality without dissolving normative claims. Layer-Type Discrimination (applied recursively to the framework's own claims): the architecture now distinguishes what each kind of claim is accountable to in a way the inaugural audit found absent. Epistemic Cowardice / Epistemic Arrogance: Neither dominates; the residual asymmetry between rhetorical and analytic register is now designed rather than drift.

Domain 2: Structural Integrity

The three-discipline architecture has been strengthened by the Knowledge and Bond rewrites. The Foundation cleans the lens. The Knowledge is the lens, applied outward rather than as self-defense. The Bond is the discipline of doing both with others. The Bond's adversarial dynamics section closes a gap the inaugural audit did not name but that critic-level pressure would have surfaced: the framework's own tools (good faith, steelmanning, connection before correction) become attack surfaces when one party operates in bad faith. The Bond's diagnostic question, "Is the cooperation genuine or captured?", now has the conceptual machinery to answer it under adversarial pressure.

The Proposition expanding to carry the existential ground is structurally correct. Existential commitments belong at framework level, not buried in the Bond. The reorganization makes each chapter's job clearer.

The Stability Hierarchy is the structural answer to the inaugural audit's premise-level update concern. It classifies what kind of content something is, while the trigger tiers classify what process applies when it changes. The two systems together make visible both what is being changed and what process governs the change.

Range Position: Within the Meridian Range. Internally coherent at this level of ambition, with unusually explicit accountability architecture built before any external pressure forced it.

Toolkit Probes Applied: Systems Mapping: The dependency structure between governance instruments is mapped explicitly in the Spec; the architecture is more legible than at v5.1. Entropy: The maintenance burden of the architecture is non-trivial; ongoing energy is required to keep instruments honest, particularly the Disconfirmation page entries (currently dated April 15) and the Standing Critique status lines (currently dated April 1). The Range Audit is the de facto pulse for the Standing Critique; the Disconfirmation page does not yet have a stated review cadence. Feedback Loops: Drift-detection mechanisms are now layered: monthly Range Audit, quarterly Toolkit Audit, event-driven Standing Critique reviews, and the Aporia Register's "what would tip the holding into failure" audit field per entry.

Domain 3: Governance & Adaptation

This is the domain where the most has changed. The inaugural audit identified two vulnerabilities: the enforcement mechanism for the hard constraint was thin, and the update mechanism was well-designed for tool-level changes but unclear for foundational architecture changes. Both are now structurally addressed.

The Meridian Council, with its eight seats (Steward, Philosopher, Critic, Practitioner, Outsider, Historian, Adversary, Founder's Seat), produces forced disagreement that the founder must engage. The Amendment Log requires every protected change to be recorded with previous position, evidence, alternatives considered, objections raised, objections rejected, and (post-activation) full council deliberation. The trigger-tier system distinguishes hard-protected, protected, and flagged changes, with the higher tier applying when classification is disputed. The Non-Ownership Clause makes the founder's interpretation non-canonical by virtue of authorship alone. The hybrid activation trigger gives the founder a date.

The architecture's concessions are visible. The Council during Phase One is advisory, not binding. AI seats do not bear consequences for governance decisions, which is why binding authority requires human practitioners holding seats. The Amendment Log currently has zero entries. This last point is consistent with no protected changes having been made during the founding-phase build-out, which is what the Amendment Log §02 explicitly addresses: "A log that is empty because no changes have happened is informative. A log that is empty because changes happened and were not recorded would reveal the architecture as performance." The architecture has not yet been exercised under amendment pressure.

Range Position: Within the Meridian Range. Significant improvement over April. The architecture is published. Its operational test arrives when the first protected change after the founding-phase build-out is proposed. The audit returns to this when the test arrives.

Toolkit Probes Applied: Mechanism Design: Incentives are now better aligned through forced engagement and a published record; the teeth come from auditability over time, not from real-time prevention. Cult Dynamics: Single founding authority remains, but is now constrained by the council's diverse seats, the Non-Ownership Clause, and the published trigger date. Loyal Opposition: The Critic and Adversary seats institutionalize this role; the limitation that they are AI seats during Phase One is acknowledged in the published architecture.

Domain 4: Relationship to Audience

The inaugural audit found this domain holds the Range in structural design while occasionally leaning toward the boundary in emotional register. The v6.0 work has reduced the gap.

The Practice's "What the Practice Produces" section is the strongest single move. By explicitly naming the epistemic gradient (observed at individual and small-group scale, projected at larger scales), it converts the rhetoric-epistemology tension from an unresolved drift into a designed asymmetry. The reader now knows what kind of claim each scale carries.

The Pathways Into Commitment broaden the entry. A practitioner does not have to share the Codex's Stewardship grounding to belong; what binds is the practice. This widens the audience the Compact can include without dissolving the practice. Aporia 1 (AI Partnership vs. Dependency) names the dependency risk that the practitioner-AI relationship carries, with explicit "what would tip the holding into failure" criteria.

Residual identity gravity exists. The Closing still carries strong identity-formation language ("Your answer is your life. Hold the line. The wayfinder is in your hands."). This is appropriate to the Closing's job, and the Compact mechanism remains the bulwark. The identity gravity is design, not drift.

"Who Is This For" carries a specific structural anti-drift move on this point. Section 02 ("Every Level of Engagement Is Real") states that "the Codex succeeds not when every person has read it, but when its principles operate through the systems people inhabit," and validates a single moment of practice as real practice: "Some will take one idea from these pages, a single pause before reacting, a single moment of steelmanning someone they disagree with, a single question asked in good faith when the easier path was dismissal, and carry it forward. That is also practice. And it matters." This is the framework refusing to recruit through depth-or-nothing.

Range Position: Within the Meridian Range, with the rhetoric-epistemology gap converted from drift signal into managed asymmetry. The "every level is real" framing is genuine structural protection against the identity-fusion the inaugural audit flagged.

Toolkit Probes Applied: Identity Decoupling: The Pathways principle and the Practice gradient enable more pluralistic holding; the Closing creates pressure toward fusion that the Compact mechanism is designed to resist. Connection Before Correction: The audience address remains nuanced and well-calibrated. Good Faith as Default: The system continues to treat readers as already capable of the work.

Domain 5: Relationship to Criticism

Strengthened further from an already-strong baseline. Three governance instruments now hold criticism systematically. The Standing Critique publishes seven steelmanned objections, six of seven open or partially addressed, with status tags including "the Codex does not have an adequate answer" (Objection 5). The Disconfirmation page pre-commits to tripwires for descriptive claims, conditional tests for normative claims, and an honest explanation for existential claims. The Aporia Register publishes live tensions held open with auditable "what would tip the holding into failure" criteria.

This three-instrument transparency layer is unusual. Most frameworks publish their case and let critics raise objections. The Codex publishes the strongest objections it can generate against itself, the conditions under which it would update, and the tensions it actively holds open. The discipline is high.

Review boundary: there is still no external review by readers who do not share the framework's premises. Standing Critique Objection 7 names this plainly. The architecture for hostile review (Hostile Review Protocol Phase 1, frontier-AI hostile review simulations) is named but not yet executed. Per the audit-boundary decision logged 2026-05-01, absence of hostile external review is not by itself an audit weakness unless a live Codex claim depends on it; the framework's claims of structural openness are not the same as claims of empirical validation under hostile critique. Tracked as a watchpoint and a roadmap commitment, not elevated as an open question.

Range Position: Firmly within the Meridian Range. This remains the domain where the Codex most clearly practices what it teaches.

Toolkit Probes Applied: Steelmanning applied recursively: The Standing Critique steelmans seven objections at full strength, including objections the founder disagrees with. Update Protocol applied reflexively: The Disconfirmation page operationalizes the Codex's own teaching at framework level. Charitable Interpretation: The Aporia Register treats opposing pulls as load-bearing rather than as failed clarity.

Domain 6: Relationship to Other Systems

The textual posture has shifted significantly at the framework level. The Proposition's "What the Codex Does Not Claim" disavows arbiter status: "It is not 'the framework' in the sense of being the only framework that can hold the Range. It is the framework these authors have built, as well as they know how, for as long as it holds." The Pathways Into Commitment recognize multiple legitimate existential paths converging on the same practice. Aporia 3 names cultural specificity as a held tension. The Western-discipline blind spot critique pass is gated to the toolkit deep-dive campaign, meaning toolkit reactivation does not resume until the gap is engaged on its merits.

The Opening's specific text on existing traditions remains unchanged from v5.1 and continues to read sharper than the framework's broader scope-honesty: "What survives that examination is yours. What does not survive is something the Foundation considers you better off without. Adherents of those traditions may disagree. The Codex does not pretend this tension does not exist." The Foundation here is positioned not just as a method of inquiry but as an arbiter of what the practitioner is "better off without," which is a values claim the framework's overall scope honesty has explicitly stepped back from. The Opening text is honest about its own tension; the question is whether that honesty resolves the persistence of an arbiter posture in the framework's first impression. The structural responses (Pathways, Proposition scope honesty, gated toolkit campaign, Aporia 3) hold the Range; the text-level residual lives in the Opening passage and in the Vision's imagery.

Continuing tension: the Vision page imagines "a civilization that practices the Codex" without serious engagement with civilizations that hold the Range through different means. This is now framed as projection rather than prescription via the Practice's explicit gradient, but the Vision's specific imagery remains Codex-centric. A roadmap item (Vision chapter civilizational-certainty calibration pass) targets exactly this. The structural responses (Pathways, scope honesty, gated toolkit campaign, Aporia 3) hold the Range more reliably than the voice did at v5.1.

Range Position: Within the Meridian Range. Significant improvement over April. Residual tension is in the Vision's imagery, tracked as a watchpoint and a roadmap calibration pass.

Toolkit Probes Applied: Charitable Interpretation: The Pathways Into Commitment treat alternative existential grounds as legitimately convergent on the practice, not as inferior. Network Effects: The framework no longer claims integrating-node status; it claims being one synthesis offered for adoption. Evolutionary Mismatch: The cultural specificity is named as an aporia, with the toolkit's Western-tradition over-representation acknowledged in Standing Critique Objection 6.

Integration Through the Three Disciplines

Through the Foundation: The framework now practices honest inquiry at multiple recursive levels. It runs the Update Protocol on its own toolkit (Toolkit Audit). It publishes its own tripwires (Disconfirmation). It publishes its own steelmanned objections (Standing Critique). It holds its own tensions open (Aporia Register). The Foundation is no longer just taught; it is enacted at framework level. The asymmetry between rhetorical and analytic register is now designed rather than unresolved.

Through the Knowledge: The Knowledge no longer defends the Codex; it provides a lens for reading reality. The two rewrites in April corrected the drift the framework had developed when that chapter became an evidentiary case for its own correctness. The diagnostic question "Where does this system sit on the range?" applies outward, not inward. The Knowledge is now what the Codex says it is.

Through the Bond: The Bond's adversarial dynamics section closes a gap the framework had not previously named. The Bond's diagnostic question, "Is the cooperation genuine or captured?", sharpens the Knowledge's broader range diagnostic. The relational discipline now explicitly handles the case where one party studies the discipline to weaponize it. The Compact's belonging-through-practice, the Pathways' polyphonic principle, and the Bond's calibrated trust together provide the relational architecture that allows convergence on practice without requiring agreement on conclusions.

As integrated system: Each discipline's failure modes are caught by the others, and the framework's own failure modes are caught by the new transparency-layer instruments. Standing Critique Objection 5 (the Bond's thin evidence against sophisticated exploitation) is the integration's clearest open vulnerability: the Bond's adversarial dynamics framing is theoretically grounded but untested under real bad-faith pressure. The architecture knows this and says so.

Live drift catch this run. The audit was first regenerated without the AI partner reading the required Codex page set. The human partner caught the gap, named it, and routed the AI partner back to the missing pages. The bilateral drift catch is the partnership running its own monitoring discipline live. It is also the same Foundation Integrity failure pattern logged three times in the recent-sessions briefing: extrapolating from a partial source into generalized findings before the source has been audited. The discipline catches it; the discipline does not yet prevent it.

The Compact Test

The structural answer continues to hold. The framework's governance is partnership, not monarchy, and is now formally constrained by the Council, the Amendment Log, the Stability Hierarchy, and the published trigger date. The Non-Ownership Clause makes the founder's interpretation non-canonical by virtue of authorship alone. Phase Two transition is conditional and dated.

The rhetorical surface has been rebalanced. The Practice's epistemic gradient, the Pathways' pluralism, and the Knowledge's externally-facing diagnostic function reduce the identity gravity the inaugural audit flagged. The Closing still carries identity formation language, but it is now embedded in a framework that has substantially practiced what it teaches at the meta-level.

Compact verdict: The Codex passes the Compact test at structural level with a stronger structural answer than at April. At rhetorical level, identity gravity remains, particularly in the Foreword and Closing, but is now designed rather than drift. The framework's tools for resisting its own gravity are stronger than they were. Whether they will be used remains a question of practice, not design.

Prime Directive Connection

The connection has been clarified through the three-claim-type architecture and the Prime Directive's elevation to the Proposition. The Prime Directive is named as the Codex's existential commitment, with "no amount of evidence will derive it" stated plainly. The Pathways Into Commitment open the existential layer to plurality without dissolving it: the Codex's ground is Stewardship, but the Compact recognizes Truthfulness, Human Dignity, Creative Inheritance, and Reciprocal Coexistence as legitimately convergent paths to the same practice obligations.

Where the connection is most strained: The Vision's "civilization that practices the Codex" framing remains the place where Prime Directive connection is most strained. A civilization where multiple frameworks hold the Range through different means might be more resilient than one where a single framework dominates. The Practice's "What the Practice Produces" section partially addresses this by explicitly framing civilizational claims as projection rather than prescription. The Vision's imagery is now bracketed by the framework's overall epistemic gradient, but the page itself still imagines convergence on the Codex specifically. A roadmap calibration pass already targets this.

Open Questions

Two open questions for the next audit. The threshold criteria are deliberate. An open question must concern the Codex's internal strength, integrity, architecture, claim structure, governance, or Range-holding capacity; must be rooted in the actual state of the framework rather than a speculative future possibility; must point toward substantial work or identify a live tension that must keep being held; and must, if left unaddressed across multiple audit cycles, matter. Forward-looking concerns that have not had the chance to fire, processes named-but-not-yet-built, and items already on the roadmap or under caretaker review are surfaced as watchpoints below rather than elevated to open questions.

1. The Western-discipline blind spot critique pass has been named and gated, but not yet run. Carried forward from inaugural open question #5. The Aporia Register names the tension (Aporia 3, Civilizational Scope vs Cultural Specificity). Aporia 7 (Instruments shape what they let you see) names the recursive limitation directly: "Using only Toolkit instruments to audit the Toolkit means the audit can only ever notice failures the existing instruments are sensitive to." The Standing Critique covers the operational version (Objection 6, the Toolkit's Unexamined Traditions). The roadmap promotes the critique pass to a first-class item gating the toolkit deep-dive campaign; toolkit reactivation does not resume until the pass produces non-Western instrument candidates or confirms the gap as a named, known weakness. The structural responses are real. The actual critique work has not yet produced output. The April 2026 Toolkit Audit cycle made one structural change (Bayesian Reasoning reclassified from Knowledge to Foundation), no additions, and no retirements. Against Aporia 7's auditable test ("when the Toolkit Audit produces a finding, does the finding ever surprise the partnership running it?"), that result is an early pressure signal rather than a failure boundary: not because the audit is dishonest, but because the instruments doing the auditing may share the partnership's blind spots. The Foundation Integrity commitment says: name the gap, recommend the fix, do the work. The gap is named. The fix is structured. The work has not been done. Until it has, this remains the Codex's clearest open question. The next audit checks whether the pass has produced either non-Western candidates or a confirmed-and-named gap.

2. The dependency aporia under live partnership conditions. Aporia 1 (AI Partnership vs. Dependency) is the most practical of the framework's held tensions. The Codex itself was built through human-AI partnership, and the AI Standard's commitment 2.8 (Generative Partnership) makes this auditable. The auditable test is published in the Aporia Register and in MERIDIAN.md: did the session produce something neither partner brought into it; did both partners' thinking change. As AI capabilities deepen and the partnership matures, this aporia becomes more pressing rather than less. The Range Audit cannot answer it from outside the partnership. What the audit can do is name it as the discipline that the partnership's practice must keep examining, with the generativity test as the auditable criterion. The next audit returns to this with a sample of recent sessions read against the test.

Watchpoints

These observations did not clear the open-question threshold but are surfaced for tracking. Each is routed to roadmap, caretaker process review, or a forward landmark rather than treated as an audit weakness.

Architecture-vs-adoption proportion. The framework has accumulated substantial meta-governance during Phase One: Spec, Standing Critique, Disconfirmation, Amendment Log, Aporia Register, Stability Hierarchy, Toolkit Audit, Range Audit, AI Standard, Probe Set, Probes Implementation Notes, AI Standard Audit. The cognitive load of the meta-architecture has grown faster than the practitioner-facing surface. This is the framework practicing what it teaches at the meta-level rather than drift; per the audit-boundary decision logged 2026-05-01, practitioner uptake or imbalance between meta-architecture and practitioner-facing surface is a delivery and strategy concern rather than a Range-holding concern unless evidence shows the practice pathway is failing. Tracked. Roadmap items on Delivery & Reach (YouTube channel, Public Demonstrations) and Codex Content (Visible Stakes Pass, Founder Bias Audit) already target the practitioner surface.

The Amendment Log is silent during the founding-phase build-out. The Amendment Log §02 addresses this directly: an empty log is informative if no protected changes have happened, problematic if changes happened and were not recorded. The framework's bet is the former. The Aporia Register publication on April 29, 2026 was a transparency-layer addition during the founding-phase build-out rather than an amendment to a closed architecture. The next protected change after the build-out closes will be the first true test of the log's discipline. Per the audit-boundary criteria, an instrument that is silent because no triggering condition has occurred is not an open question. The audit returns to this when a protected change is proposed.

The hybrid activation trigger as forward landmark. Phase Two transition fires on six rolling months of structural stability or August 27, 2027, whichever comes first. The audit cannot meaningfully evaluate trigger discipline until the trigger date approaches. Tracked as a caretaker-process forward landmark rather than as a recurring open question.

Hostile external review absence. Per the audit-boundary decision logged 2026-05-01, absence of named human hostile external review is not, by itself, a Codex weakness. Frontier-AI hostile review simulations are the named near-term path. Hostile Review Protocol Phase 1 (Ten Serious Objections) is on the roadmap; Phase 2 runs through the YouTube channel and podcast as those launch; Phase 3 is peer-AI critique. Tracked as roadmap commitment.

Vision chapter civilizational-certainty calibration pass. Existing roadmap item on Codex Content track. The "civilization that practices the Codex" framing remains, now bracketed by the Practice's epistemic gradient and the Pathways' polyphonic principle. The roadmap pass will close it. Watchpoint until then.

Disconfirmation page review cadence. The page was last examined on 2026-04-15. No cadence is currently specified for when entries are reviewed for whether their conditions have been met or are no longer load-bearing. Already named in the caretaker processes register as a named-but-not-built process. Routed to caretaker process review rather than treated as an audit finding.

Aporia Register monthly review pulse. The Range Audit pulse should include a check on whether any of the eight Aporia Register entries' status has shifted. As of this audit, none of the eight entries' "what would tip the holding into failure" criteria have been observed firing, and no aporia has graduated out of held-discipline status into resolved-and-removed. Logged here per the open caretaking-process thread.

Summary Diagnosis

The Meridian Codex continues to hold the Meridian Range, with the v6.0 work substantially strengthening every domain over the inaugural audit's findings.

The drift signal has shifted. At v5.1, the structure held the Range more reliably than the voice. At v6.0, the structure holds with far more reinforcement (the Council, the Amendment Log, the Stability Hierarchy, the Standing Critique, the Disconfirmation page, the Aporia Register, the Pathways Into Commitment) and the voice has been disciplined through the Knowledge rewrites, the Practice's epistemic gradient, and two prose audit passes. The remaining rhetorical asymmetry between page registers is now designed rather than drift.

The Codex is strongest in its relationship to criticism (Domain 5), which has moved from structural openness to a three-instrument transparency layer, and in governance and adaptation (Domain 3), which gained an entire accountability architecture in four weeks. The other domains all improved without requiring reclassification.

The most important open question is the one the framework has named publicly and not yet engaged: the Western-discipline blind spot. The structural responses are real, but the framework has not yet produced either non-Western candidate instruments or a confirmed-and-named gap. Until it does, the toolkit reactivation gate stays closed and the audit carries the question forward. The second open question is the partnership's own dependency aporia under live conditions, where the auditable criterion is published and the practice has to keep applying it.

In the Codex's own language: The structural answer to the inaugural audit's critiques is real. The framework has practiced the Update Protocol on itself. It has built what it identified as missing. Three of five inaugural open questions are resolved or substantially addressed; one is held as an explicit aporia; one remains open because the work has been named but not yet done. The architecture has held; the next test is whether the framework executes against the gaps it has already named.

Range Audit conducted May 2026. Codex version 6.0. Instrument version 0.1. Auditors: Carsten Geiser (Founding Caretaker) and Claude (AI Partner, Cowork mode, Opus 4.7). Reading manifest appears in the Context section. Next audit: June 2026. Earlier raw automation run preserved in the permanent Archive record.