ADVERSARIAL_REVIEW_CORPUS_MANIFEST.md#

Status: MANIFEST v1.2 — 2026-08-15 — inventory and identity record; authorizes nothing. Scope: Every file in the adversarial-review corpus classified in ADVERSARIAL_REVIEW_CORPUS_CLASSIFICATION.md (v1.5.1), plus the classification round's own deliverables. This manifest records what each file is — identity, instrument, subject, version role, and content digest. Analysis and rationale live in the classification report; this file is the lookup, not the argument. Versioning rule: This manifest follows the canonical-plus-instance model. ADVERSARIAL_REVIEW_CORPUS_MANIFEST.md is the canonical rendering; each cut pins an immutable instance ADVERSARIAL_REVIEW_CORPUS_MANIFEST_v<N>.md before the canonical advances. The integrity check is: canonical digest equals newest instance digest. Digest law: Every file row carries a SHA-256 digest. The digest identifies; the filename locates. If a filename and digest disagree with this manifest, the digest governs — a matching digest under a different name is the same artifact; a matching name with a different digest is a different (or tampered) artifact. Tables show the first 12 hex characters; §5 is the full-digest register.

1. Classification-round deliverables#

Integrity check at manifest time: canonical digest f51d8af1… = instance digest f51d8af1… — PASS.

2. Corpus artifacts — Set A: Republic whitepaper, paper-only#

Set B: Republic repository-grounded#

Set C: Companion subset (ACP + VSB)#

Set D: Magistracy / emergence#

Set E: Resolvable/Resolved-Reference Register (RRR)#

3. Roll-up#

22 corpus artifacts (21 files in the upload set + the preserved Copilot transcript) · 22 evidenced review passes — 20 consumed, 2 preserved-but-excluded (both Copilot) · 6 instruments, every fingerprint family operator-anchored: F1/OpenAI (rc8 review), F2/Grok (repo-informed feedback), F3/Gemini (repo audit), F4/OpenAI (RC9 review), F5/DeepSeek (re-supply match), F6/Copilot (paste), CL/Claude (self-declared + corpus-internal corroboration).

Standing corrections this manifest carries: (1) Review_ADV_COMP_VAL_SCOPE_BINDING.md is DeepSeek regardless of its internal Claude frontmatter — the digest d847e1729f50… binds that correction to the exact bytes; if a wrapped copy with a corrected header is ever cut, it enrolls as a new digest with lineage to this one. (2) The two Copilot passes are excluded from all consumption counts. (3) The absorption round's instrument-class ledger predates the F5 reclassification and awaits re-annotation.

3a. Source-session register (operator-supplied 2026-08-15)#

The originating chat sessions for each external instrument, recorded at the vendor level. Per the digest/anchor law these URLs locate — they are account-gated and resolvable only by the operator, so they strengthen each anchor's recoverability (the origin context can be re-opened, re-exported, or re-verified by the accountable party) without changing its attestational class. Per-artifact session mapping is not asserted: each link is the vendor's review session; individual artifacts attribute to vendors via the anchors in §2.

Claude-family sessions are indexed separately in SESSION_INDEX v0.2 (Architect / Author / Archivist / Researcher UUIDs).

4. Verification procedure#

To verify any file against this manifest: compute sha256sum <file> and compare against §5. To verify the classification deliverables' version integrity: the canonical's digest must equal the highest-numbered instance's digest. A canonical that diverges from its newest instance means an unpinned edit occurred — pin or revert before consuming.

5. Full-digest register#

53f9a08ba0351bd4529c218dc8ea4357781b1b6870f0c7e7101c8e440f6acb31  Adversarial_Review_Companion_Subset.pdf
1d0569e1bf04b68c61d728a011c6b17d4933d3da75563b592e45d5fab574a549  Companion_Subset_Adversarial_Review.docx
ad6adb9a649fbe932514327dc81f6f81e348c870997b53691bd39eee76530430  MAGISTRACY_ADVERSARIAL_REVIEW_COMPREHENSIVE.md
77dcc408c5cefd2bb9598d8f05d02a2899a79b6ef57ed8760f9b9e1c1b8cd9e0  MAGISTRACY_REVIEW_FINDINGS.md
717eb959ea3be8e59f262512a13ff2f5a03c07864d0f6c2e5dd97ad362d51907  REPUBLIC_RC9_REPOSITORY_GROUNDED_ADVERSARIAL_REVIEW.docx
83bbcc3da2b375beceb51108a70120c787f79c1d617f8fea445ac1dd01877852  REPUBLIC_v1_0-rc7_critical_review.md
4d177616964217a08d6eff2c1c22b5cf85a12c16ee5a7b5de994e4d854b042cc  REPUBLIC_v1_0-rc8_ADVERSARIAL_REVIEW.md
3fe96d873f81beafed8c85acbd76995c1936bee1da6faf220cac248dda2af99c  RESOLVED_REFERENCE_REGISTER_v0_1_adversarial_review.md
11fc55e922ddd045646cc28d7ecaceed80419ec2ccd5037b67531ed358c9d07c  RRR_COVER_NOTE.md
30a41327eb24b1ffdb25fcad934e0f1d80de7c0445bec5e64d910bc655caa7cd  RRR_MULTI_INSTRUMENT_SUMMATION.md
2cc0c041d2cd40bd1cdbdf604301bcc6498f3fd246bf417da0edbf4b9e514764  RRR_v0_1_INDEPENDENT_GRADE.md
4b5cf2c4e3280875da24e42d90f09f1f1a09e81890d36c1d1b079975ce9a20f2  Republic_Repo_Informed_Feedback.docx
8eed26239c2e851eab937851c7dc7ae5360c8ff66666fb828224f7dca20e43c1  Republic_Whitepaper_Critical_Analysis.docx
c538ecf1d4b6eedb4b74807280394250502a5fcb1ec6bf95fc1c8e60a9c15d22  Republic_Whitepaper_rc8_Adversarial_Review.docx
de03c079a50bbb956137abd37fceea4e745593447a7d84ed37d897d6cfb519c3  Resolved_Reference_Register_Adversarial_Review.docx
d847e1729f509c33f635085412ee5ba0165957ea71e98414866cd143ef47321d  Review_ADV_COMP_VAL_SCOPE_BINDING.md
fee506c693a8fe9faae005db02d8ff7891876681a9371abcad7c5323e8d76a7a  VALUATION_ACP_COMPANION_ADVERSARIAL_REVIEW.md
013f4aa6b48ca12afbc47e662bba40a11af9a6a477aac63c2407dc6cff8ecedf  VSB_ACP_ABSORPTION_REVIEW_FINDINGS.md
85d55e4b84b31738f5eb4edfbe5dd28d5098c3e1ae38b0df0cd61421817543e2  adversarial_review_republic.pdf
302a01c28cefcaf013bf5dc45a627d2485a93ff4971c509600a0e30584565480  repository_audit_republic.pdf
27737ec52ea562b8ffbd6bca484b87d22ae60b8b76b32fd148289ae9ed31f1ad  republic_critical_analysis.pdf
88c3a4a594403ef6d5d0fb111004fbcbf9619e0f2698eb453ffc239f566e118e  COPILOT_REVIEW_TRANSCRIPT_2026-08-13_14.md
f51d8af1a1d1b681888478642a4e6180c9071b8a4c855d6ddb9e95b608d70a54  ADVERSARIAL_REVIEW_CORPUS_CLASSIFICATION.md
f51d8af1a1d1b681888478642a4e6180c9071b8a4c855d6ddb9e95b608d70a54  ADVERSARIAL_REVIEW_CORPUS_CLASSIFICATION_v1.5.1.md
72bf04ccdd051b171cc6915acded18d8ef7434fba21f82debb3a04e9ba80f1df  DIGEST_ANCHOR_VERIFICATION.md
72bf04ccdd051b171cc6915acded18d8ef7434fba21f82debb3a04e9ba80f1df  DIGEST_ANCHOR_VERIFICATION_v0.1.md

Change log#

• 2026-08-15 v1.2 — source-session register added (§3a): operator-supplied origin links for all five external instruments, recorded as vendor-level anchor-recoverability data. DeepSeek and Copilot origins now bind their respective correction records to named sessions. Instance _v1.2 pinned at cut.
• 2026-08-15 v1.1 — DIGEST_ANCHOR_VERIFICATION.md v0.1 enrolled (canonical + pinned instance) as the pattern formalizing the integrity law this manifest runs. Instance _v1.1 pinned at cut.
• 2026-08-15 v1.0 — initial manifest. 22 corpus artifacts + 3 deliverables inventoried with SHA-256 identity; all six instrument families operator-anchored; standing corrections (F5 false self-label, Copilot exclusions, absorption-ledger re-annotation) recorded as manifest-carried facts. Instance _v1.0 pinned at cut.