Skip to content

History / ai os 1.stub

Revisions

  • chore(wiki): weekday curator — purge 7 archives, ingest LangGraph_STRIDE Org listing Public-archive purge: sttbench, qwenbench-mlx, sst-autoresearch, ace-step-1-5-mlx, longcat-video-avatar-1-5-mlx, plus TSLIT archived stubs (tslit, tslit-dspy-ar) while keeping the collapsed-lineage compound page. Ingest langgraph-stride under ai-security. Home count 38 live / 25 archived. Date bumps for cleanroom and loopscope. Sitemaps regenerated. No push.

    @ai-agents-cybersecurity ai-agents-cybersecurity committed Aug 28, 2026
  • curator: ingest loopscope + STRIDE-Lite + singularity-atlas; purge 11 newly-archived repos New stubs (3): - loopscope (agentic-frameworks) - in-process live dashboard for LangGraph graphs and Ralph loops; non-invasive hook, stall as a first-class terminator - stride-lite (ai-security) - STRIDE/DREAD threat models, ATT&CK kill-chain scenarios, Vault + Campaign Score; v2.1 agent-runtime attack templates - singularity-atlas (agentic-frameworks) - feed fusion over Neo4j into a globe, eight vectors, a 0-100 index and a locally-written daily brief Purged 11 repos archived on GitHub 2026-08-18, each confirmed by both the org listing pill and the repository's own archive banner. Home 52->44 live repos, 18 archived. MLX category took the brunt: 18->12. Also fixes a Home.md orphan row for deepseekvl2-PDF-OCR-private left behind by the 2026-08-18 purge, and corrects that run's archive finding: the org listing it trusted was a stale cached snapshot, so stable-audio-3 and animated-GIF-Creator were genuinely archived, not false positives. See Log.md for the revised method. 5 contested repos deliberately left in place pending better evidence.

    @ai-agents-cybersecurity ai-agents-cybersecurity committed Aug 22, 2026
  • curator: ingest AI-OS-1 + cleanroom; refresh tars-ai, screenlens, skillspector, venture-pathfinder New repos: - AI-OS-1 (local-inference-dgx) - research corpus for a model-aware distributed inference OS on N>=2 DGX Sparks; no runtime by design. - cleanroom (developer-tools) - four-agent LangGraph file-hygiene pipeline with an explicit verifiable/best-effort honesty contract. Architectural changes captured: - tars-ai dropped LangGraph entirely (dependency gone, no StateGraph). - screen-lens-mlx re-converged to dual-platform (one client over vLLM/CUDA or oMLX/MPS), partially answering the July open question about mlx/dgx fork convergence - now recorded under Resolved questions. - skillspector gained a loopback oMLX AI Analyst (triage/adjudicate/verdict). - venture-pathfinder gained a loopback command deck with background jobs. Lint: repointed 11 stub cross-refs from the unresolvable [[ai-security]] to [[ai-security-pentesting]].

    @ai-agents-cybersecurity ai-agents-cybersecurity committed Aug 18, 2026