Format: Migrated the canonical Markdown bundle to OKF v0.2. Concepts now use structured generated, verified, status and sources frontmatter; reserved indexes and this log use their v0.2 structures.
Compatibility: Kept v0.1 consumption through timestamp and body # Citations fallbacks while giving v0.2 fields precedence.
Explorer: Added explicit trust-tier, lifecycle, freshness, provenance and passive Attested Computation contract presentation. Loading a bundle never executes declared computation, executor or attester resources.
2026-07-07
Update: Applied the Claude Fable 5 code review: XSS and URL-scheme hardening across the Svelte, static, and legacy viewers; harvested-URL sanitisation (credential redaction, scheme validation) and data-hygiene warnings in the UK Government API generator; CI now exercises the generator against fixtures. Full findings in docs/code-review-2026-07-07.md.
Update: Restructured this log to OKF §7 date-grouped headings; intentional OKF v0.1 deviations documented in docs/okf-conformance.md.
2026-06-27
Initialization: Bundle created from From API-Calling LLMs to Agent-Ready Digital Infrastructure (Published DRAFT), the Federated AI Research Execution Report, and the Engineering Agent-Ready Infrastructure deck. 137 concepts authored across nine sections; technical claims grounded in primary standards, RFCs, arXiv papers and UK guidance (URLs from the source paper's own references where available).
Update: Added Context Hub (Andrew Ng; andrewyng/context-hub, MIT) as the now-verified primary source for the content/skills registry layer, resolving the open item from the source paper.
Convention: Point-of-use glossary linking + reverse index; see Linking conventions. Open items (Context Hub; UK federated-AI deployment evidence) recorded in the evaluation.
Format: Open Knowledge Format v0.1 (Google Cloud, 2026).