Source identity
| Field | Recorded value |
|---|---|
| Source type | project Library file |
| Original record | GITHUB_LIBRARY_AUDIT.md |
| Date | Oct 15, 2026 |
| Path or archive member | Conciousness research/GITHUB_LIBRARY_AUDIT.md |
| Original SHA-256 | a7bcaca4ae30f37a1b7398e432994ca0b87b1f9cf0230dcc0c5db0028215a746 |
| Reading-copy SHA-256 | a7bcaca4ae30f37a1b7398e432994ca0b87b1f9cf0230dcc0c5db0028215a746 |
Every recorded occurrence
These 2 records contain identical original source bytes. Their paths and revisions remain part of the archive.
| Original record | Branch or archive | Revision / identifier | Original location |
|---|---|---|---|
| GITHUB_LIBRARY_AUDIT.md | Archive occurrence | libsrc-c5a31c49e6b7 | Conciousness research/GITHUB_LIBRARY_AUDIT.md |
| GITHUB_LIBRARY_AUDIT.md | {"sourceId": "libsrc-678bab94146b", "member": "GITHUB_LIBRARY_AUDIT_2026-10-14_v1/GITHUB_LIBRARY_AUDIT.md"} | libsrc-0870bf782256 | archives/libsrc-678bab94146b/GITHUB_LIBRARY_AUDIT_2026-10-14_v1/GITHUB_LIBRARY_AUDIT.md |
Recovered source transcript
Download complete source text ↓161 lines · 14.9 KB
# GitHub Library Recovery Audit — Joseph / Janady13
**Version 1 · October 14, 2026 · Read-only connected inspection**
## Executive finding
GitHub contains substantially more source and tracked recovery material than the earlier Netlify deployment archive: native C/C++ and Verilog source in AI_Brain, the three-lineage complete-agi-core tree, Seferim database entries and learned-items data, Rhiannon and G16 C++ implementations, a separate JavaScript mathematical reference, infrastructure backups and a MEGAMIND Go integration inside seo-sidecar.
This is an **inventory and source-inspection audit, not a repository backup or restored AI runtime**. No repository was cloned onto Joseph's Mac, no database payload was downloaded/validated, and no recovered program was executed. Existing Netlify archives were read without modification.
## Scope and method
| Measure | Result |
|---|---:|
| Connected repository records | 61 |
| Public / private | 44 / 17 |
| Confirmed forks / non-forks | 16 / 45 |
| Code-search indexed / unindexed | 19 / 42 |
| Non-fork root trees inspected | 44 |
| Additional non-fork inspected by branch-specific README | 1 |
| Repositories with selected file-content inspection | 7 |
| Full Git history/release/LFS audits completed | 0 |
| Repository backup archives created | 0 |
| Recovered application programs executed | 0 |
| Netlify source files independently matched to GitHub blobs | 7 of 7 |
The connected repository inventory used page 1 with limit 100 and include-archived enabled, followed by page 2, which returned no further records. Every returned owner was Janady13. No archived records appeared in the returned set. The separate fork-only search identified the 16 fork records. Absence of other owners is not proof that no other accounts/organizations exist; access is limited to the connector's grants.
Every non-fork received a tree or README inspection, but this does **not** mean every nested file, every branch, or every line was read. Large recursive responses were sometimes truncated; selected subtrees were inspected separately. The catalogue labels the depth and limitations for each repository. Forks received metadata/fork classification only; their modifications were not compared against upstream.
Keyword code search was not used as the completeness mechanism because 42 records reported no code-search index. The inventory is a normalized transcription of connected tool results, not a raw API-response archive. Returned Git tree identifiers are not silently relabeled as commit IDs.
## 1. Highest-value engine and state leads
### complete-agi-core — preserve three system lineages
The repository has distinct `thataiguy/`, `seferim/`, and `unified/` trees. Its README calls these the two original systems and a unified third system. Source paths include:
```
unified/agents/pantheon-daemon.sh
unified/core/unified-orchestrator.sh
unified/federation/federation-mesh.sh
unified/knowledge/autonomous-learner.sh
```
These are tracked implementation files, not merely filenames quoted in an old chat. The Seferim listing also contains the following blobs:
| Tracked path | Bytes | Git blob SHA |
|---|---:|---|
| seferim/federation/remote_memory.db | 7,000,064 | f80be423f86e28e1bb01f6952af1da0037173561 |
| seferim/federation/shared_memory.db | 6,897,664 | 0c4fe843d08b11514d9572d83d04a6d9bcc90711 |
| seferim/data/learned_items.json | 1,384,902 | 407c5347d96678184a3b1e66c5d1ba15615ffc4f |
Those byte counts and hashes are **listing evidence**. The databases were not opened; their schema, integrity, learned content, date, application consistency and relationship to the latest running system remain unverified. Tiny state files and JSON status records must be preserved, but their names do not establish complete trained-state recovery.
The README's scale and consciousness assertions are historical project descriptions, not independently reproduced measurements. Its 24-agent Unified roster must not overwrite separately recorded 17- or 16-member architectures.
Evidence: connected tree and README outputs `turn9file0`, `turn10file0`, `turn11file0`, `turn12file0`, `turn23file0`; repository `https://github.com/Janady13/complete-agi-core`.
### AI_Brain — native source under the_tired_agi
The `the_tired_agi/core` subtree contains CMake configuration, C/C++ headers and sources, Verilog files and `knowledge/agi_knowledge_base.txt` (597,002 bytes). Particularly relevant paths include:
```
the_tired_agi/core/src/dna_core_real.cpp
the_tired_agi/core/src/dna_core_rt_main.cpp
the_tired_agi/core/src/dna_api.cpp
the_tired_agi/core/src/fgtp_engine.cpp
the_tired_agi/core/src/helix.cpp
the_tired_agi/core/src/agent_arrays.cpp
the_tired_agi/core/src/ingest_pipeline.cpp
the_tired_agi/core/src/agi_migrate.c
the_tired_agi/core/src/fpga_pipeline.v
the_tired_agi/core/src/fpga_memory_bus.v
```
The inspected DNA source excerpt instantiates HelixEngine, FGTPEngine, IngestPipeline and AgentArrays, defines scheduling/growth parameters and starts a runtime loop. Its 1 kHz value is a configured target, not a measured capability. No CMake build, dependency closure, FPGA synthesis, timing test or application restoration was attempted. Source existence is now established; equivalence to Joseph's last canonical MEGAMIND binary remains open.
Evidence: `turn15file0`, `turn20file0`, `turn41file0`, `turn42file0`; repository `https://github.com/Janady13/AI_Brain`.
### ai-brain-core — preserve backups before cleanup
The root listing includes `manager.py`, adapter/orchestrator/sub_brains directories, `db/`, `backups/`, and `.git_backup_1755669199`. Environment backup directories, launch/sync scripts and an `env` symlink are also tracked. These are high-priority preservation leads. The symlink must not be followed into unrelated local paths during extraction. An embedded Git backup can hold separate historical evidence, so it must not be discarded as generated clutter.
No embedded history/object audit or database validation has yet been performed. Credential-looking filenames were recorded without reading or printing values. Evidence: `turn21file0`; repository `https://github.com/Janady13/ai-brain-core`.
## 2. Implementation versus scaffold: preserve the distinction
**Rhiannon.** `src/agents/two_way_engine.cpp` implements deterministic initialization, sigmoid behavior, squared-error gradient steps, correction toward a target and error measurement. The same source explicitly defines the SIMD path as a no-op and the Metal path as a placeholder. This is implemented CPU optimization logic with unfinished acceleration, not a verified finished GPU engine. Evidence: `turn13file0`, `turn14file0`, `turn22file0`.
**SEFERIM G16 C++.** The README describes a research scaffold. `src/common/g16_core.cpp` implements named family transforms, normalization, inertia/stability gating and an Omega objective; its comments label the family definitions placeholders with interpretable mathematics. Several mathematical labels are proxy operations, not implementations of the full named scientific quantities. Language-binding stubs are not a recovered Julia numerical engine. Evidence: `turn17file0`, `turn18file0`, `turn39file0`, `turn40file0`.
**seferim-core JavaScript.** `src/seferim-core.js` contains a separate reference implementation, including 16 family transforms, `MetaState.step` and `DNAMemory.learn`. Foundations/mathematics documents and a test runner are also present. The header's claim of faithful implementation of 69 equations was not independently audited; no tests were run. Keep this specification-linked JavaScript version separate from the C++ implementation, whose operation ordering is not assumed identical. Evidence: `turn16file0`, `turn44file0`.
None of these findings requires dismissing the work. They identify exactly what can be recovered and which implementation gaps remain to be tested.
## 3. Verified links to the Netlify recovery
### Previously blocked ThatAIGuy repository is readable
Authenticated connector access successfully read `Janady13/ThatAIGuy`. The earlier anonymous clone failure is therefore not evidence that the repository was deleted. Its default tree contains seven website files under `thataiguy-website 2`.
`Janady13/thataiguy-website` has the same seven-file subtree under `thataiguy-website`. Both return subtree SHA `f8e97caa2b5afe0c8e55b90d77a9aa9749cb408b`. This verifies content duplication within those subtrees; the repositories' full histories were not compared. Keep both identities.
Evidence: `turn8file0`, `turn38file0`.
### Route Builder: all seven source-map files match
The prior Netlify audit ZIP was opened read-only. For each of its seven embedded first-party Route Builder source files, this audit computed Git's blob SHA-1 over `blob <byte-length> + NUL + file bytes` and compared it with the corresponding GitHub tree entry. **Seven of seven matched.**
The files are App.tsx, AuthModal.tsx, Dashboard.tsx, LandingPage.tsx, index.css, index.tsx and reportWebVitals.ts. `ROUTE_BUILDER_SOURCE_COMPARISON.json` records every local path, length, calculated Git blob hash, GitHub blob hash and SHA-256.
GitHub additionally lists the client dependencies/configuration, server tree, database schema and Netlify function wrapper. This materially expands the source-recovery target, but does not establish the deployed commit, all historical versions or a recoverable production database.
Evidence: `turn37file0` and the independently generated local comparison JSON.
### Two autonomous-music repositories share one HTML file
The original repository's `autonomous-ai-music.html` and the 2.0 repository's `index.html` are each 35,928 bytes with blob SHA `927667d828f00cf8d5ba8c1ced7706e02a7af293`. The 2.0 repository has additional README/workflow material. Do not count the shared HTML as two independent engine implementations or delete either history. Evidence: `turn45file0`, `turn46file0`.
## 4. Additional systems and knowledge material
The complete 61-entry catalogue records each individually. Notable groups include the AI_Brain_Control orchestration/job record tree, AI_Brain_backend Python service, AI_Body build scaffold, healer-pro-pack's shell modules, Jennys-Drumste application source and the Photoshop UXP integration in design-studio-mcp.
`ai-music-system` contains worker/scheduler/status/trainer files and checkpoint directories. A specifically inspected 128-byte checkpoint contains only step, loss, time and datasetDigest. Preserve it as telemetry/history rather than presenting it as a trained model. Other payloads still require inspection. Evidence: `turn28file0`, `turn31file0`, `turn43file0`.
`seo-sidecar` is important despite its non-AGI name: `megamind/seo` contains Go files for schema feedback, experiments, learning logs, queues, workers, GA signal sources, validation and tests. This is a concrete example of AI-related work that a narrow repository-name filter would miss. Evidence: `turn51file0`, `turn52file0`.
The private `seo-aio-26-discipline-library` default snapshot contains a 1,767,268-byte HTML file. Its contents were not deeply audited and its filename does not establish the later requested 38-discipline completion. The current ThatDeveloperGuy, ThatStupidComputer and FeedTheJoe repositories also preserve application/project structures that must remain separate from older Netlify page versions. Evidence: `turn50file0`, `turn53file0`, `turn58file0`, `turn67file0`.
Several default snapshots are minimal: Healer-AI has a 30-byte README, skynet-backend has only a funding file, groovesteps has a README, and clarksboxing exposes only the funding subtree at root. This is a snapshot finding, not proof that their entire histories never contained source. Preserve and inspect their histories rather than deleting them.
## 5. Open recovery issues and boundaries
1. **Actual backups remain to be made.** This audit saves metadata, findings and comparisons—not full Git mirrors, private database bytes, LFS payloads or releases.
2. **History and alternate branches.** No exhaustive commit/tag/branch or deleted-file investigation was performed. The default branch may not contain the newest or fullest build. Missionmatch's inspected branch is `feature/opportunity-to-outcome`.
3. **External state.** Live databases, model stores, local `/Volumes/MEGAMIND` material and off-repository state are not made recoverable by source inspection. Exact native binary/build/state correspondence remains to be established.
4. **Missionmatch durability warning.** Its README documents a pilot PostgreSQL restart-durability and backup/restore blocker. This is a source warning, not a current test of the M4. Validate it before relying on the application operationally. Evidence: `turn33file0`.
5. **Sensitive files.** Some trees track environment/backup filenames. Presence does not prove live secrets are exposed; avoid publishing or displaying payloads, preserve original bytes privately and review credentials separately.
6. **Scope outside connector grants.** The 61-record count covers the connected accessible set, not every possible account, organization or deleted repository.
7. **Scientific provenance.** Source-defined Phi/Psi/Omega fields, small state records and README capability assertions are not independently validated consciousness measurements. The original manuscript numerical engine/data still needs exact version matching.
8. **Netlify account discrepancy remains.** This GitHub inspection does not resolve the earlier 142-versus-141 Netlify discovery/inventory discrepancy.
## 6. Desktop backup handoff
The included `Claude_Code_GitHub_Desktop_Backup.txt` tells Claude Code running on Joseph's Mac to create a fresh timestamped Desktop recovery directory, reconcile the complete inventory, mirror all accessible histories, fetch LFS objects, preserve default-commit source snapshots and tracked data, verify archives, and record missing hosted metadata separately. It explicitly prohibits executing recovered programs or modifying production.
GitHub's current backup documentation describes mirror clones, separate LFS retrieval, archive compression and separate wiki handling. It also warns that migration archives do not include all service data. Documentation checked for this handoff: `https://docs.github.com/en/repositories/archiving-a-github-repository/backing-up-a-repository`.
## Deliverables and integrity
- `ALL_61_REPOSITORIES.md`: readable catalogue, one section per repository, with limitations.
- `GITHUB_LIBRARY_INVENTORY.json` and `.tsv`: complete normalized inventory and coverage/priority fields.
- `SOURCE_PROVENANCE.json`: repository tree URLs and connector result references.
- `ROUTE_BUILDER_SOURCE_COMPARISON.json`: seven independent blob comparisons.
- `CONTENT_RELATIONSHIPS.json`: matched website subtree and music HTML relationships.
- `Claude_Code_GitHub_Desktop_Backup.txt`: executable-workflow instructions for a local Claude Code session.
- `SHA256SUMS.txt`: integrity hashes for the generated audit files.
The supplemental ZIP contains **audit material only**, not the 61 repositories. Checksums verify this audit package's local integrity; they do not prove remote recovery completeness.