What is not here
20 documents were withheld from this site. Each is named below with the category of the reason and its sha256.
Why documents were refused
Architecture & Decisions
| Document | Reason | Size | sha256 |
|---|---|---|---|
| docs/control-plane/AGENT-CALIBRATION-PROMPT.md | operator-denied-value | 45.1 KB | 118f62417928ec03 |
| docs/control-plane/FQDN-SEAM-DETERMINATION.md | operator-denied-value | 9.0 KB | 89eaa1d0e843ae0f |
| docs/control-plane/RESONANCE-TRIAGE-2026-07-26.md | operator-denied-value | 11.4 KB | cbed9e48fe25dc44 |
The Flagellar Motor
| Document | Reason | Size | sha256 |
|---|---|---|---|
| docs/THE-LABORATORY-PLAN.md | over-redaction-threshold (28 > 10)private-addresstailscale-addressinternal-hostnameoperator-path | 464.5 KB | 03a29717a9268808 |
The Colony & the Method
| Document | Reason | Size | sha256 |
|---|---|---|---|
| docs/OPERATIONS_MANUAL.md | over-redaction-threshold (52 > 10)private-addresstailscale-addressinternal-hostnameoperator-path | 24.7 KB | af136ec91c38470a |
| docs/REQUEST-TO-JULES-public-documentation-site.md | operator-denied-value | 20.0 KB | 52ab65035ef64c69 |
| docs/RESUME_2026-06-22.md | over-redaction-threshold (14 > 10)private-addressoperator-pathcredential | 17.6 KB | 38c426763849ed5e |
| docs/STUDIO_HARDENING_DD_TDD_PLAN.md | operator-denied-value | 28.1 KB | 44df6894ff73825d |
| docs/SYSTEM_OVERVIEW.md | over-redaction-threshold (11 > 10)private-addressoperator-path | 27.8 KB | 7ae65eb57126c861 |
| docs/UNIVERSE.md | over-redaction-threshold (24 > 10)private-addressoperator-pathcredential | 35.8 KB | eb156a1079ee8be5 |
| docs/UNI_QA_AND_E2E_PLAN.md | operator-denied-value | 18.5 KB | 3a39481efcea1d4e |
| docs/UNI_WORLDSIM_BUILD_AND_PROOF.md | operator-denied-value | 32.4 KB | 9c2c043c38685a88 |
Evidence & Verdicts
| Document | Reason | Size | sha256 |
|---|---|---|---|
| docs/receipts/adaptive_selfnet_phase0_2026-07-15.md | over-redaction-threshold (20 > 10)private-addresstailscale-addressinternal-hostname | 4.9 KB | 4f2ad82f49742dbd |
| docs/receipts/chip_address_tracking_2026-07-16.md | over-redaction-threshold (34 > 10)private-addressinternal-hostname | 12.7 KB | 47404e2a9dd63649 |
The Reader and its Deploy Bundle
| Document | Reason | Size | sha256 |
|---|---|---|---|
| deploy/README.md | over-redaction-threshold (12 > 10)private-addresstailscale-address | 8.6 KB | ab304d6469be0294 |
| deploy/TABLET.md | over-redaction-threshold (12 > 10)private-address | 15.6 KB | e53cf3f850d68d24 |
The Broadcast Platform
| Document | Reason | Size | sha256 |
|---|---|---|---|
| production/dns/README.md | over-redaction-threshold (16 > 10)private-addressinternal-hostname | 6.3 KB | 4e4fa3ff8b5139a6 |
| production/docs/receipts/dns_phase0_4_2026-07-12.md | over-redaction-threshold (61 > 10)private-addressinternal-hostnameoperator-path | 14.9 KB | 1f397f4e12ef7c27 |
| production/docs/receipts/dns_zone_drift_fix_2026-07-15.md | over-redaction-threshold (19 > 10)private-addressinternal-hostname | 8.1 KB | 36ed56996a042817 |
The Builder — colony cold start
| Document | Reason | Size | sha256 |
|---|---|---|---|
| CLAUDE.md | over-redaction-threshold (21 > 10)private-addresstailscale-addressinternal-hostnameoperator-path | 67.2 KB | ba5b5bb4554c1b93 |
Withheld for want of provenance
8 documents are not published because their own bytes are uncommitted in the source repository. Every page here asserts “this is this file, at this commit” — and a document that exists only as an uncommitted edit cannot make that claim truthfully.
This fence was added on 2026-08-01 after a forensic pass held all 292 published pages against the commit each citation named. One page failed: its published bytes were a working-tree edit published under a commit sha that did not contain them. 291 of 292 were sound. The page count on this site fell by one as a result — a reduction in count and an increase in correctness, recorded as an amendment on the coverage page rather than quietly absorbed. The fence is per file: unrelated uncommitted work elsewhere in a source repository does not withhold anything.
| Withheld | Corpus | Would have cited |
|---|---|---|
| docs/control-plane/RESUME.md | control-plane | b909801f3db4 |
| docs/LIVING-SCIENCE-WALKTHROUGH.md | workbench | c23f686b641a |
| docs/MATH-WORKBENCH.md | workbench | c23f686b641a |
| docs/UNI-STACK-BUILDER-PLAN.md | workbench | c23f686b641a |
| docs/audit/PHASE-E-WORKBENCH-AUDIT.md | workbench | c23f686b641a |
| CLAUDE.md | flagellum-repo | b909801f3db4 |
| CLAUDE.md | workbench-repo | c23f686b641a |
| README.md | workbench-repo | c23f686b641a |
Deduplicated
8 documents were not rendered because an identical document had already been ingested from another repository — proved by sha256, never by filename. Two files with the same name and different contents are two documents; two files with different names and the same bytes are one.
These arise because one repository is a git worktree of another on a different branch, so its documentation directory is largely the same directory. Publishing both copies would put the same page in the wiki twice under two names, which is the smallest possible version of the findability problem this site is built to avoid.
| Not rendered | Identical to | Size | sha256 |
|---|---|---|---|
| docs/CROSS-STUDY-PARITY.md | flagellum/cross-study-parity | 13.7 KB | 2f388f58f1435338 |
| docs/HARDWARE.md | flagellum/hardware | 2.5 KB | 98d0e3f49534c5c9 |
| docs/OBSERVED-EXPERIMENT.md | flagellum/observed-experiment | 7.0 KB | 8622b9791400b0b0 |
| docs/SCIENCE-GATES.md | flagellum/science-gates | 8.0 KB | 4f1bf29386c66957 |
| docs/SCIENCE.md | flagellum/science | 5.9 KB | caf8fa47cce7bc33 |
| docs/VERUM.md | flagellum/verum | 1.0 KB | 68815587869ba786 |
| docs/audit/PHASE-A-FINDINGS.md | flagellum/audit-phase-a-findings | 19.2 KB | 2422749cd3bc6235 |
| docs/audit/SEMANTIC-COVERAGE-LEDGER.md | flagellum/audit-semantic-coverage-ledger | 14.2 KB | 6123b3d3ed53a1f9 |
Everything that IS here, accounted for
This page is one half of a pair. It names what was withheld; the coverage page names what was kept and proves that every subsystem, every runnable command, every document type and every published page is either covered or excluded on purpose.
A site that only publishes its omissions still leaves a reader unable to tell thorough from selective. Both halves are needed and both are computed rather than asserted.
The limit of this page
Everything above was caught by a pattern — an address shape, a hostname shape, a key shape. No scanner recognises sensitive narrative. A document describing a client engagement in prose, naming nobody and matching no pattern, passes every check that produced this page.
That is not hypothetical in this estate. A large archive of development transcripts was verified to have had its credentials redacted perfectly — zero surviving tokens across 1.74 GB — while its narrative content was the actual exposure. Human review of new prose is required, and nothing here replaces it.