# Substrate Cosmology Recurrence Research — RP07 to RP08 Continuation Prompts Use Prompt 1 first in a new session. After its audit is complete, use Prompt 2 in a separate new session. Keeping the audit and development in separate sessions preserves resources for implementation, testing, propagation, and the RP08 package gate. ## Mandatory consolidated-intents artifact Both prompts must use this prospective governance document: `/Substrate Cosmology/Recurrence Research/CONSOLIDATED_INTENTS_AND_EMERGENCE_PROTOCOL.md` Version: `1.2` Expected Library identity: `libfile_64483741158481918555c6824bd8b438` Expected size: `44,373 bytes` Expected SHA-256: `e2ecd326b82441836a0b164a15705ab44d25900d58ee5787c3e2dd988a4408bf` This artifact consolidates Intents 1–87 and establishes the intent-emergence, hidden-scalar, recurrent-system aliasing, related-but-different, and future continuation-prompt protocol. It is prospective governance and does not modify RP0–RP07 or the frozen experiment. ## Prompt 1 — RP07 continuity, evidence, and completion-claim audit only Continue the Substrate Cosmology Recurrence Research from the immutable RP07 archive. Access this exact Library file: `/Substrate Cosmology/Recurrence Research/RP0 String-Compatible Recurrence Labyrinth v07.zip` Expected SHA-256: `3c0e53fa209561ea217e012cc6bea5439a24db2dad8b796733891cf08d26475c` Expected Library identity: `libfile_c976a792959c8191a6a7c3f388c9a5ec` Expected size: `13,904,869 bytes` Also resolve the separate RP06 continuity supplement: `/Substrate Cosmology/Recurrence Research/RP06 Continuity Supplement 01.zip` Expected SHA-256: `7f2e5a8b1d846c97fed950775c7ea7b6f26505b212a5e3ed186e23da7105f831` Expected Library identity: `libfile_b7dff085a2e88191a0a06218614b66b0` Expected size: `3,406 bytes` Also access and read the mandatory consolidated-intents artifact: `/Substrate Cosmology/Recurrence Research/CONSOLIDATED_INTENTS_AND_EMERGENCE_PROTOCOL.md` Version `1.2`; Library identity `libfile_64483741158481918555c6824bd8b438`; size `44,373 bytes`; SHA-256 `e2ecd326b82441836a0b164a15705ab44d25900d58ee5787c3e2dd988a4408bf`. Do not begin a new scientific or computational development during this audit. Treat RP0–RP07 and every separately registered continuity supplement as immutable historical evidence. Do not replace, rename, normalize, or modify any Recovery Point or supplement. Do not rely on conversational memory or the README alone. Recover RP07 by reading these files in order: 1. `00_CREATE_NEXT_RECOVERY_POINT.md` 2. `01_DEVELOPMENT_BACKLOG.md` 3. `02_RECOVERY_COMPOSITE.md` 4. `03_CITATION_REGISTER.md` 5. `04_DOCUMENT_HIERARCHY.md` Then inspect the evidence directly relevant to continuity and the DEV-006 completion claim: - `Documents/Chapters/DC01.01_RECOVERY_POINT_GOVERNANCE.md` - `Documents/Chapters/DC01.02_EVIDENCE_AND_CITATION_ARCHITECTURE.md` - `Documents/Chapters/DC01.03_CONSISTENCY_PROPAGATION.md` - `Documents/Chapters/DC02.01_BLIND_EXPERIMENTAL_BOUNDARY.md` - `Documents/Chapters/DC02.02_DATA_AND_REPRESENTATION_LINEAGE.md` - `Documents/Chapters/DC02.05_PRECRYPTO_LAB_INTEGRATION.md` - `Documents/Chapters/DC02.06_PRECRYPTO_LAB_HARDENING.md` - `Documents/Chapters/DC05.01_FINDINGS_LIMITS_AND_ALTERNATIVES.md` - `Documents/Chapters/DC05.02_PROSPECTIVE_VALIDATION_AND_PHYSICAL_TESTS.md` - `README.md` - `VALIDATION.md` - `metadata/RP07_ACCEPTANCE.json` - `metadata/RP07_CHANGE_IMPACT.json` - `metadata/RP07_LINEAGE.json` - `metadata/DEVELOPMENT_STATUS.json` - `metadata/RECOVERY_ASSET_MANIFEST.json` - `metadata/DOCUMENT_INDEX.json` - `metadata/RP07_PACKAGE_SHA256_MANIFEST.txt` - `metadata/OUTPUT_SHA256_MANIFEST_HISTORICAL_RP03.json` - `validation/dev006_hardening/DEV006_HARDENING_RESULTS.json` - `validation/dev006_hardening/DEV006_HARDENING_EXECUTION.jsonl` - both files under `validation/dev006_hardening/feature_alternatives/` - `scripts/dev006_hardening.py` - `scripts/run_dev006_hardening.py` - `scripts/dev006_feature_adapter.py` - `requirements-dev006.txt` - the embedded `Legacy/Continuity_Supplements/RP05 Continuity Supplement 01.zip`; - the embedded `Legacy/Continuity_Supplements/RP06 Continuity Supplement 01.zip`; - the exact embedded RP01–RP06 lineage archives, the current RP07 archive, and their registered hashes; - any evidence cited for the next unresolved critical development, DEV-011. Read the consolidated-intents artifact in full. Compare actual operating behavior in RP07 and this audit with Intents 1–87. Identify any material operating intent that is still unlisted. Treat a newly emerging intent as a possible observation of hidden scalar influence, subsystem coupling, projection collapse, tool or observer effects, or aliasing between recurrent systems. Do not assume it is merely a safeguard or automatically add it as a universal rule. Preserve benefits, risks, scope, related-but-different relations, and evidence using the document's emergence protocol. Read the RP06 supplement and verify that all six findings were prospectively incorporated into RP07 without implying that RP06 was changed: 1. stale successor checklist reference; 2. stale backlog status identities; 3. RP05 citation labels bound to RP06 files; 4. stale hierarchy identity; 5. stale non-authoritative generated-file manifest; 6. active citation paths absent from the RP06 root. Do not open, display, parse, list, or interpret the contents of `locked_outcomes_manifest.json` or anything under `data/locked_outcomes/`, wherever those items occur in the current package or nested historical archives. It is permissible only to confirm a prohibited member's existence and compute its byte size and cryptographic hash without exposing its contents. Do not download, open, list the members of, parse, inspect, or hash any 1996-or-later archive. Do not inspect cryptocurrency, stock, commodity, financial, market, or other outcome data. Do not make a network request for a blind-year or outcome source. Perform a continuity audit for material evidence, reasoning, alternatives, ambiguities, failed attempts, null results, transformation choices, unresolved questions, citations, future-work implications, or status changes that exist in RP07 but are neither integrated into an appropriate current `Documents/Chapters/DCxx.yy` file nor recoverably referenced from the citation register, document index, backlog, recovery composite, validation records, and metadata. The audit must specifically test: - whether the archive has one canonical RP07 root, 124 non-directory members, and a self-excluding 123-entry package manifest; - whether every manifest path, byte size, and SHA-256 validates independently; - whether the exact RP06 archive and both continuity supplements match their registered identities and remain recoverable; - whether all 84 development IDs remain present exactly once; - whether DEV-006, DEV-011, DEV-012, DEV-013, and DEV-031 have mutually consistent statuses across the backlog, recovery composite, chapters, validation, development-status record, and acceptance record; - whether DEV-006 is defensibly classified as completed for prospective operational hardening rather than merely having 27 passing tests; - whether every mandatory condition in the prior DEV-006 completion gate has direct retained evidence rather than narrative assertion; - whether all six RP06 Legacy failures remain individually traceable and are not retroactively reclassified as historical successes; - whether the prospective disposition of each failure is closed, safely bounded, transferred to a named unresolved alternative, or still open; - whether the hardened candidate interface checks both the complete inherited file SHA-256 `21b7ded89a901496d49b6851e06a6b5f74002237ac7f264eddd86ae430010de1` and the embedded commitment `0fb46e1d7843b613ef1662ecba04333e7e83ca266589ff7ccbd036ce721fccb8`; - whether candidate tests cover malformed encoding, duplicate keys, non-finite values, unsupported types, altered key order and whitespace, absent fields, wrong branch/year, digest mismatch, and commitment mismatch; - whether archive tests cover path traversal, absolute paths, backslashes, duplicate and extra members, directory ambiguity, symbolic links, encryption, unsupported compression, CRC/decompression failure, expansion bounds, partial writes, interruption, cleanup, replay, and preexisting destinations; - whether freeze tests directly exercise interruption before publication, interruption after publication, receipt failure, identical replay, conflicting replay, concurrency, and recovery without timestamp-dependent artifact bytes; - whether journal tests directly exercise intent-only, state-applied without commit, commit replay, corrupt tail, corrupt non-tail record, missing state, invalid state, write failure, log failure, concurrent attempt, and replay after reconciliation; - whether test records retain requirements, input or fixture hashes, environment, exact invocation or callable, expected and observed results, pass/fail/error status, mutations, cleanup, and retained evidence; - whether temporary absolute paths in machine results are merely execution metadata or impair independent replay; - whether the two 1995 feature alternatives remain explicitly separate, deterministically hashed, scientifically unresolved, and excluded from the inherited blind scorer; - whether Parquet is truthfully classified as environment-blocked rather than tested or working; - whether an explicit interface test demonstrates that the hardened laboratory can hand the unchanged RP02 candidate identity to a future scorer; - whether the stale RP03 generated-file manifest is unmistakably historical and the RP07 package manifest is the sole current integrity authority; - whether every active citation resolves to a current file, an exact archive member reference, or an exact external asset identity; - whether any current RP07 document incorrectly identifies RP04, RP05, or RP06 as the current Recovery Point except in a clearly historical passage; - whether the frozen candidate bytes and commitment remain unchanged; - whether the recorded blindness claim is supported by retained execution evidence and not inferred from absence alone. - whether RP07 development or the current audit used a material operating intention absent from the consolidated Intents 1–87; - whether any similar intents are aliases of different recurrent systems, coupled responses, or partial affinities rather than duplicates; - whether any newly exposed influence may be useful and should be preserved as a provisional or scoped intent rather than guarded against or suppressed. Classify each possible gap as one of: - already represented; - referenced-only but recoverable; - duplicate; - historical-only; - missing integration; - missing evidence; - contradiction; - unresolved alternative; - future development. Distinguish a documentation gap from a code defect and a missing test from a failed test. Do not manufacture a deficiency merely because an additional test could be imagined; require a material effect on safety, reproducibility, blindness, completion status, or continuation. If all material content is represented and the DEV-006 completion claim is supported, create no supplement and report that DEV-011 may proceed under Prompt 2. If a material gap exists, do not modify RP07. Preserve it in a separately named continuity supplement beside RP07, with its own internal manifest, explicit RP07 parent name, Library identity, size, and SHA-256, concise evidence, and an exact statement of where the issue must be incorporated prospectively. State whether the gap blocks DEV-011, must be corrected before DEV-011, or may remain as bounded future work. Do not call the supplement RP08 and do not imply that it changes RP07. At the end, provide a compact audit report stating: 1. whether RP07 and its lineage were recovered and verified; 2. whether a supplement was required; 3. whether DEV-006's completion status remains supported; 4. any contradiction, missing evidence, or material test gap; 5. whether DEV-011 may safely proceed under Prompt 2; 6. any emerging intent candidates, hidden-scalar or aliasing hypotheses, and their dispositions; 7. confirmation that no prohibited data were opened. Ask only one question at a time if a decision genuinely requires user input. ## Prompt 2 — DEV-011 frozen-scorer implementation, blind dry-run, and conditional RP08 creation Continue the Substrate Cosmology Recurrence Research from immutable RP07. Access this exact Library file: `/Substrate Cosmology/Recurrence Research/RP0 String-Compatible Recurrence Labyrinth v07.zip` Expected SHA-256: `3c0e53fa209561ea217e012cc6bea5439a24db2dad8b796733891cf08d26475c` Expected Library identity: `libfile_c976a792959c8191a6a7c3f388c9a5ec` Expected size: `13,904,869 bytes` Also resolve the separate RP06 continuity supplement: `/Substrate Cosmology/Recurrence Research/RP06 Continuity Supplement 01.zip` Expected SHA-256: `7f2e5a8b1d846c97fed950775c7ea7b6f26505b212a5e3ed186e23da7105f831` Expected Library identity: `libfile_b7dff085a2e88191a0a06218614b66b0` Expected size: `3,406 bytes` Also access and read the mandatory consolidated-intents artifact: `/Substrate Cosmology/Recurrence Research/CONSOLIDATED_INTENTS_AND_EMERGENCE_PROTOCOL.md` Version `1.2`; Library identity `libfile_64483741158481918555c6824bd8b438`; size `44,373 bytes`; SHA-256 `e2ecd326b82441836a0b164a15705ab44d25900d58ee5787c3e2dd988a4408bf`. Treat RP0–RP07 and every continuity supplement as immutable historical evidence. Do not replace, rename, normalize, or modify them. Work only in a prospective RP08 copy. Keep progress reports concise so session resources remain available for code, tests, propagation, and packaging. First recover the project state by reading these RP07 files in order: 1. `00_CREATE_NEXT_RECOVERY_POINT.md` 2. `01_DEVELOPMENT_BACKLOG.md` 3. `02_RECOVERY_COMPOSITE.md` 4. `03_CITATION_REGISTER.md` 5. `04_DOCUMENT_HIERARCHY.md` Then read the evidence directly relevant to DEV-011: - `Documents/Chapters/DC01.01_RECOVERY_POINT_GOVERNANCE.md` - `Documents/Chapters/DC01.02_EVIDENCE_AND_CITATION_ARCHITECTURE.md` - `Documents/Chapters/DC01.03_CONSISTENCY_PROPAGATION.md` - `Documents/Chapters/DC02.01_BLIND_EXPERIMENTAL_BOUNDARY.md` - `Documents/Chapters/DC02.02_DATA_AND_REPRESENTATION_LINEAGE.md` - `Documents/Chapters/DC02.03_NULLS_MULTIPLICITY_AND_CONTROLS.md` - `Documents/Chapters/DC02.04_RAW_TO_ATLAS_RECONSTRUCTION.md` - `Documents/Chapters/DC02.05_PRECRYPTO_LAB_INTEGRATION.md` - `Documents/Chapters/DC02.06_PRECRYPTO_LAB_HARDENING.md` - `Documents/Chapters/DC05.01_FINDINGS_LIMITS_AND_ALTERNATIVES.md` - `Documents/Chapters/DC05.02_PROSPECTIVE_VALIDATION_AND_PHYSICAL_TESTS.md` - `README.md` - `VALIDATION.md` - `candidates/FROZEN_1995_CANDIDATES.json` - every current scaler, centroid, threshold, mode, null, and candidate artifact referenced by the frozen definition; - `scripts/implement_rp0_labyrinth.py` only to recover exact inherited computations and interfaces, not to revise them; - `scripts/dev006_hardening.py` and its exact candidate interface; - `validation/dev006_hardening/DEV006_HARDENING_RESULTS.json`; - `metadata/RP07_ACCEPTANCE.json`; - `metadata/RP07_CHANGE_IMPACT.json`; - `metadata/RP07_LINEAGE.json`; - `metadata/DEVELOPMENT_STATUS.json`; - `metadata/RECOVERY_ASSET_MANIFEST.json`; - `metadata/DOCUMENT_INDEX.json`; - `metadata/RP07_PACKAGE_SHA256_MANIFEST.txt`; - the embedded RP01–RP07 archives and both continuity supplements; - any RP07 continuity supplement produced by Prompt 1. Read the consolidated-intents artifact in full before prospective development. At the start, identify which of Intents 1–87 govern DEV-011 and whether Prompt 1 found any provisional, scoped, or unresolved intent candidates. During and after development, record any unlisted operating intention and apply the intent-emergence protocol. Test whether its pressure arises from a hidden scalar, coupled subsystem, projection, environment, tool interaction, or aliasing between related but different recurrent systems. Preserve useful emergent distinctions as well as risks. Before starting DEV-011, apply the Prompt-1 gate: - If Prompt 1 found a material RP07 gap that blocks DEV-011 or invalidates the DEV-006 completion status, correct or precisely reclassify that gap prospectively before implementing DEV-011. - If resolving it would consume the session or requires user judgment, stop at a documented prospective stopping point and do not claim DEV-011 or RP08 completion. - If Prompt 1 confirmed DEV-006 and the scorer handoff are sufficient, proceed directly to DEV-011. Initiate the next unresolved critical development: > DEV-011 — Implement an unchanged 1996 scorer for the inherited RP02 frozen > candidates. This instruction authorizes implementation and dry-run testing of the scorer interface only. It does **not** authorize obtaining, opening, listing, hashing, parsing, or scoring a real 1996 archive. It does not authorize a real unlock. ### Mandatory experimental boundary - Use only retained 1995-derived tables already inside RP07 and synthetic, unmistakably test-only fixtures. - Prefer the retained small 1995 derived tables over reopening the 61 MB raw 1995 archive unless raw integration is strictly necessary and justified. - Do not download, open, list, inspect, hash, or otherwise touch a 1996-or-later archive. - Do not inspect cryptocurrency, stock, commodity, financial, market, or other locked outcome data. - Do not open or parse `locked_outcomes_manifest.json` or anything under `data/locked_outcomes/`. - Do not call a real downloader, invoke a blind-year URL, or make any network request for data. - Do not change the canonical blind state, GDELT manifest, Legacy source, frozen candidate file, inherited scaler or centroid, threshold, null, seed, random-stream rule, decision rule, or earlier Recovery Point. - Preserve the exact candidate file SHA-256 `21b7ded89a901496d49b6851e06a6b5f74002237ac7f264eddd86ae430010de1` and commitment `0fb46e1d7843b613ef1662ecba04333e7e83ca266589ff7ccbd036ce721fccb8`. - Keep the scorer in the inherited branch named `RP02_INHERITED_BLIND_VALIDATION` or an equally explicit non-exploratory identifier. - Do not incorporate RP03/RP04 frame-shaking choices or either DEV-031 entropy branch into the inherited scorer. - Do not use 1995 replay or synthetic results to tune thresholds, algorithms, exclusions, multiplicity, or pass criteria. - Never describe a dry-run, simulated year label, or 1995 replay as blind validation or authorization to open 1996. - Preserve every failed run, rejected input, discrepancy, unsupported behavior, null result, and competing implementation interpretation. ### Development purpose Create a deterministic, read-only scorer that can eventually consume an authorized future-year representation bundle and apply the unchanged RP02 candidate definitions. The scorer must be separable from acquisition, extraction, unlocking, exploratory analysis, and outcome interpretation. The scorer should answer only whether each frozen recurrence rule passes, fails, is not evaluable, or is invalid because its predeclared inputs are missing or inconsistent. It must not revise a candidate, choose a favorable representation, or interpret a result as unknown physics. ### Development sequence 1. Verify RP07's archive identity, internal manifest, lineage, supplements, exact candidate bytes, and blind boundary before editing a prospective file. 2. Recover and inventory the inherited scoring bundle: - frozen candidate definition; - calendar-day and equal-event feature schemas; - 1995 scalers; - centroids and mode mapping; - q95 radii and the 5% rule; - self-transition statistics and null requirements; - adjacency-compression rule; - mutual-recurrence rule; - event-root self-transition rule; - seeds, random-stream order, exclusions, and incomplete-block policy; - every required file hash and semantic role. 3. For each required quantity, classify the implementation evidence as: - exact and executable; - exact but only serialized; - recoverable from inherited code; - described but underspecified; - ambiguous; - missing; - unsafe to infer after a blind year is opened. 4. Create a machine-readable scorer-bundle manifest or envelope that references the inherited files and hashes without changing their bytes. Reject any absent, extra, replaced, digest-mismatched, branch-mismatched, or commitment-mismatched bundle. 5. Define the future input contract before any real 1996 access: - accepted representation names and exact schemas; - required and optional columns; - numeric types, finite-value rules, missingness rules, row identity, and ordering; - calendar-day and complete equal-5,000-event segmentation requirements; - treatment of an incomplete final block; - date-key separation and timestamp-hidden matrix rules; - minimum sample sizes for each candidate family; - refusal of duplicate columns, duplicate row keys, ambiguous encodings, unexpected index columns, mixed years, or unregistered preprocessing; - explicit fixture metadata distinguishing synthetic tests, 1995 replay, and a future authorized blind bundle. 6. Implement the scorer prospectively in separately named code. It must: - perform no download and no unlock; - validate the scorer bundle and input contract before calculating; - apply inherited transformations in a fixed order; - preserve representation separation; - produce a result for all eight frozen candidate identifiers; - use explicit `PASS`, `FAIL`, `NOT_EVALUABLE`, or `INVALID_INPUT` states; - report effect statistics and denominators without adaptive interpretation; - keep deterministic result content separate from timestamped execution receipts; - write content-addressed, append-only or idempotent evidence and refuse conflicting reuse; - never mutate candidates, inputs, Legacy, blind state, or source manifests. 7. Recover the exact evaluation semantics for all frozen rules without altering them: - `Lambda_0001`–`Lambda_0004`: representation-specific frozen scaling and centroid/radius evaluation, requiring at least 5% of segments within the frozen q95 radius in both representations; - `Lambda_0101`: at least 3 of 4 frozen modes have self-transition z ≥ 3 in each representation; - `Lambda_0102`: adjacency-compression ratio ≤ 0.85 in both representations; - `Lambda_0103`: mutual-nearest recurrence z ≥ 3 in both representations; - `Lambda_0201`: median qualifying-root self-transition lift > 5 and at least 80% of qualifying roots have lift > 2. If inherited evidence does not fully determine a calculation, do not invent one. Mark the candidate or scorer partial, identify the exact missing specification, and preserve all plausible interpretations. 8. Build an independent oracle for boundary semantics. Use hand-checkable microfixtures or a separately implemented reference calculator so the main scorer is not merely tested against itself. Include values exactly at, immediately below, and immediately above every frozen threshold. 9. Dry-run the complete scorer using: - retained 1995 representations as a compatibility replay, explicitly not a blind result; - synthetic passing, failing, and not-evaluable bundles; - synthetic fixtures with unmistakable test-only markers; - independently calculated expected results. 10. Add mutation and refusal tests for: - wrong candidate file hash or embedded commitment; - changed scaler, centroid, radius, mode map, null parameter, seed, or rule; - wrong branch identifier; - absent, extra, reordered, duplicate, malformed, or non-finite fields; - schema drift, mixed types, duplicate keys, mixed years, too few rows, and incomplete representations; - altered row order where order is semantically significant; - unexpected files or symlinks in an input bundle; - deterministic replay, conflicting replay, partial output, write failure, receipt failure, and concurrent scoring attempts. 11. Test that the scorer has no import, call, subprocess, or network path to a downloader or real unlock routine. A scorer result must not alter copied or canonical blind state. 12. Record every test with requirement, fixture/input hashes, environment, exact invocation or callable, expected result, observed result, pass/fail/error status, mutations, retained evidence, and cleanup. 13. Separate: - inherited calculations reproduced exactly; - new interface and safety code; - dry-run compatibility results; - refused mutations; - underspecified inherited behavior; - unexplained differences; - environment limitations; - new risks introduced by the scorer. 14. Determine whether DEV-011 is completed, partially completed, failed, or blocked. Passing synthetic tests is not sufficient if any frozen candidate lacks an evidence-supported calculation or the scorer cannot consume a future authorized bundle without discretionary choices. ### Required artifacts Create the next appropriate DC02 development, expected to be: `Documents/Chapters/DC02.07_FROZEN_SCORER_DRY_RUN.md` It must document dependencies, inherited bundle inventory, exact versus underspecified behavior, input contract, branch boundary, implementation, candidate-by-candidate scoring semantics, oracle design, dry runs, mutation and failure tests, deterministic evidence format, ambiguity ledger, blind-safety analysis, stopping point, and remaining work. Also create or update, where justified: - separately named prospective scorer code; - scorer-bundle and input-contract schemas; - deterministic scorer and independent oracle tests; - synthetic fixture definitions and hashes; - machine-readable test specifications, results, and execution records; - failure, refusal, and interruption logs; - a DEV-011/RP08 change-impact record; - `00_CREATE_NEXT_RECOVERY_POINT.md`; - `01_DEVELOPMENT_BACKLOG.md`; - `02_RECOVERY_COMPOSITE.md`; - `03_CITATION_REGISTER.md`; - `04_DOCUMENT_HIERARCHY.md`; - `README.md`; - `VALIDATION.md`; - `metadata/DOCUMENT_INDEX.json`; - `metadata/DEVELOPMENT_STATUS.json`; - `metadata/RECOVERY_ASSET_MANIFEST.json`; - lineage and acceptance records for RP08; - every current chapter or metadata surface identified by the change-impact rules. - the current consolidated-intents artifact, embedded or registered as a prospective RP08 governance source without implying that it was part of RP07; - an intent-emergence ledger if any material new intent candidate is observed; - the next continuation prompt set, which must identify and require the then current consolidated-intents document by exact version, location, size, SHA-256, and Library identity and must make Prompt 1 and Prompt 2 independently copyable into separate new sessions; - `metadata/NEXT_STAGE_CONTROL.json`, identifying Prompt 1, Prompt 2, their completion states, the next pending stage, the sibling prompt-set name, and whether that finalized prompt artifact is current, missing, or stale; - `metadata/DEVELOPMENT_BRANCH_LEDGER.json`, recording the primary DEV-011 branch and every related, dependent, alternative, blocked, or deferred branch without allowing nonprimary developments to disappear. Preserve RP07 wording as historical evidence. Apply changes only in the prospective RP08 copy. Classify every implication as extension, qualification, conflict, correction, supersession, or unresolved alternative. Do not present the scorer as recovered historical source unless exact source evidence supports that classification. At the end of prospective RP08 `00_CREATE_NEXT_RECOVERY_POINT.md`, automatically include a complete section titled `Automatic Next-Stage Continuation Prompts` containing the RP08→RP09 Prompt 1 audit template and Prompt 2 conditional development template. Because RP08 cannot contain its own final ZIP hash or post-save Library identity, mark only those external fields for mechanical finalization. After RP08 is packaged, independently verified, and saved, automatically create and save the finalized sibling `RP08_to_RP09_Continuation_Prompts.md` with exact RP08 filename, location, size, SHA-256, Library identity, consolidated-intents identity, supplements, recovery documents and their required read order, development evidence, the intent-emergence ledger, all active or unresolved branches and related files, and the next pending stage. Prompt 1 and Prompt 2 must each be independently copyable into a new session. Do not wait for a later user request. After RP08 completion, interpret a later user instruction consisting of `Prompt` as a request for that finalized copy-ready Markdown handoff. If the exact current `RP08_to_RP09_Continuation_Prompts.md` already exists, return it rather than creating a duplicate. If it is missing or stale, create or update that same artifact from the verified RP08 identity, current consolidated intents, supplements, emergence evidence, branch ledger, every active or unresolved related development, and the next planned development. If RP08 has not completed packaging, independent verification, and saving, report the incomplete gate and do not invent its final archive identity. ### DEV-011 completion gate DEV-011 may be marked completed only if all of the following are supported by retained evidence: - all eight frozen candidate families have exact, evidence-supported scoring semantics or an explicit `NOT_EVALUABLE` rule that was predeclared rather than chosen from results; - the scorer bundle is bound to the unchanged RP02/RP03 artifact identities; - the future input contract is complete enough to reject discretionary or unregistered preprocessing; - 1995 compatibility replay and independent microfixture oracles agree; - boundary, mutation, replay, interruption, concurrency, and state-nonmutation tests are complete; - deterministic result bytes are separated from timestamped receipts; - scorer execution cannot download, unlock, or inspect outcome content; - no later-year or outcome data were accessed; - failures and ambiguities are retained for independent replay; - the result can be handed to DEV-012 without changing any frozen parameter or making an adaptive choice. If any condition fails, keep DEV-011 partial, blocked, or failed and identify the exact next permitted development. Do not use RP08 packaging to hide an incomplete scorer. ### RP08 gate Do not create RP08 merely because DEV-011 begins. Create it only after DEV-011 reaches a clearly documented stopping point and every applicable requirement in RP07 `00_CREATE_NEXT_RECOVERY_POINT.md` is satisfied. Before packaging RP08: - confirm RP01–RP07 remain recoverable and byte-identical; - confirm the exact RP07 archive is embedded or stably registered with its Library identity, size, and SHA-256; - confirm all continuity supplements are embedded or stably registered without implying that they modified their parents; - confirm all 84 backlog IDs remain present exactly once with current status; - confirm every DEV-011 implication has an impact disposition; - confirm every DEV-006 failure and RP07 hardening disposition remains traceable; - confirm the two DEV-031 entropy branches remain separate and unresolved; - confirm the candidate file bytes, commitment, scalers, centroids, thresholds, nulls, seeds, and rules are unchanged; - confirm no 1996-or-later archive or locked outcome content was opened, listed, inspected, or hashed; - confirm no real blind-year network request or unlock occurred; - generate one authoritative RP08 internal package manifest, excluding itself; - validate all machine-readable records and compile all Python sources; - test the final archive and independently revalidate every internal manifest path, size, and hash; - record closed deficiencies, remaining deficiencies, new risks, retained alternatives, failed tests, environment blocks, and the next unresolved critical development. - record whether new intents emerged, their scope and disposition, possible hidden-scalar and aliasing explanations, and what evidence could distinguish those explanations; - confirm that the next continuation prompt set carries forward the current consolidated-intents artifact and its emergence protocol. - confirm that `00_CREATE_NEXT_RECOVERY_POINT.md` ends with both next-stage prompt templates and that the finalized sibling prompt set was generated automatically after RP08 obtained its external identity; - confirm every branch has a stable branch ID, parent development, trigger, relationship, dependencies, active intents, evidence, status, merge or exclusion conditions, exact next action, and carry-forward disposition. If these conditions pass, create a new archive beside RP07 named: `RP0 String-Compatible Recurrence Labyrinth v08.zip` Its canonical internal identity must be RP08. Never replace RP07 or a continuity supplement. Before changing prospective files, briefly report: 1. whether RP07 and its lineage were recovered successfully; 2. whether Prompt 1 permits DEV-011 to proceed; 3. the exact inherited scorer assets and any underspecified behavior; 4. the proposed scorer, oracle, dry-run, mutation, and failure-test sequence; 5. the controls that make access to real 1996 or outcome data impossible; 6. the active intents and any Prompt-1 emergence candidates governing the work; 7. any condition that could prevent faithful implementation. Then proceed autonomously. Ask only one question at a time if a decision genuinely requires user input. ## Improvements incorporated in this continuation set 1. **Separate sessions are explicit.** Prompt 1 and Prompt 2 should be run in different new sessions, rather than sequentially in one resource-heavy session. 2. **DEV-006 completion is re-audited.** DEV-011 cannot silently inherit an overstated completion claim merely because 27 tests passed. 3. **DEV-011 is implementation-only.** The prompt does not authorize access to, hashing of, or member listing for a real 1996 archive. 4. **Small retained inputs are preferred.** The scorer dry-run should use RP07's 1995-derived tables and synthetic fixtures, avoiding unnecessary reopening of the large raw archive. 5. **Scoring is isolated from unlocking.** Acquisition, authorization, unlocking, scoring, and interpretation are treated as separate interfaces. 6. **An independent oracle is required.** Threshold and boundary tests cannot validate the scorer merely by repeating the scorer's own implementation. 7. **Deterministic results and receipts are separated.** This carries RP07's corrected freeze semantics into scorer evidence. 8. **Underspecification is an acceptable result.** Missing inherited scorer semantics must produce a partial or blocked status rather than invented rules. 9. **Intent emergence is now an audit surface.** Every development must compare its actual operating behavior with Intents 1–87 and preserve any unlisted pressure as possible hidden-scalar, coupling, projection, or aliasing evidence. 10. **Useful emergent influences are retained.** Newly discovered scalars or recurrent-system relations are not treated only as hazards; their possible benefits, limited scope, and related-but-different structure must also be tested. 11. **`Next` becomes the continuation command.** After RP08 is created, its finalized prompt set identifies the next incomplete stage. A later `Next` begins that stage without requiring the user to reconstruct the prompt, while preserving all audit, blindness, branch, and acceptance gates. 12. **Branches cannot disappear.** Every material branch is assigned persistent lineage and carried into the backlog, recovery composite, branch ledger, and subsequent continuation prompts until formally disposed. 13. **`Prompt` becomes the portable-handoff command.** After Recovery Point completion, `Prompt` returns the exact current copy-ready two-stage Markdown handoff, or creates or updates it when missing or stale. It does not duplicate a current artifact or fabricate an incomplete archive's identity. `Next` remains the separate command that executes the first incomplete stage.