Re-synced two mirrored blog posts after correcting institution
references on the live davidtkeane.github.io source files.
Both posts previously named "University of Galway" or "Galway,
Ireland" in metadata footers; live source was corrected to "NCI
Dublin" and "Ireland" respectively. Mirrors now match the live
canonical source.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Matches the README pattern used across other top-level folders
(modelfiles, observations, paper, etc.). Provides examiners with
a quick entry point into the documentation: what each file is,
recommended reading order, and why these docs exist alongside
the primary research artefacts.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds three documentation artefacts that support the CA1 thesis:
1. docs/blog/ — 6 mirrored research blog posts from
davidtkeane.github.io (ASAS scale, identity persistence, cross-model
consciousness, Honor Code, context compaction, V1→V42 narrative).
Live URL is canonical; mirrored copies are frozen for academic record.
2. docs/research-blog.md — Curated index linking each post (live URL
+ offline mirror) with topic descriptions and citation format.
3. docs/version-evolution.md — Complete V1 → V43 evolution across
six eras (Genesis, Exploration, Refinement, Production Hardening,
Architecture Maturation, QLoRA Validation), with quick-reference
table, per-version detail, and key-lessons-by-era summary.
README updated to surface both new docs in the Published Resources
table for examiner discoverability.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Each folder now explains what's inside, why it matters,
and what to look at first. Teacher-friendly navigation.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
BF2 Global Rank #16,836 (top 0.04%), BF3 144-kill game,
BF4 G36C #5 globally on PS4, pure medic legend story
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Altitude cognitive impairment at 4,400m maps directly to IR under
fatigue. "Come home alive — summit is secondary" = preserve evidence,
don't make the breach worse.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This is a security feature, not a bug. Weight-level identity anchoring
means the model treats unauthenticated interactions as potential attacks.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Timeline, key findings, repo structure, RQs answered, how to run,
hardware specs, download stats, and published resources
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- 4-page research paper: Measuring AI-to-AI Prompt Injection in the Wild
- 137,014 items analysed, 10.07% injection rate, 7-category taxonomy
- Cross-platform injection gradient (0.5% - 18.85%)
- Licence updated from CC BY-SA to CC BY-NC-SA 4.0
- README updated with published resources links
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
ranger