TF TrustFortress Resource Library

Cloudflare R2 backed docs library

TrustFortress Resource Library

Browse repo-linked evidence packs, recovery bundles, reference datasets, and manifests kept outside git to keep the codebase lean.

Open Watch Browse datasets Agent map Ask the library Sign in for IP materials
D1 now backs the machine catalog while R2 remains the source of downloadable bytes. Repo manifests stay the provenance layer.
11 visible collectionsVisible repo manifests mapped into browsable resource families.
D1 catalog liveD1-backed /api/catalog with R2-listing fallback
R2 bytes livetrustfortress-resource-library
Cron sync liveProtected manual reindex follows the same R2-to-D1 path.
11
Collections
0
Files Here
0 B
Visible Bytes
All collections
Active View

Start here

One map for people, agents, and CUB lookups

Run the detector, open public datasets, verify theorem status, or connect an agent without hunting through the archive.

Browser tool

Watch a log

Choose a local CSV, text, JSONL, or gzip log and run the shared Rust/WASM detector entirely in your browser.

Open Watch
Dataset collection

Public agent datasets

Open the WWT demo archives for FinBot CTF, SkillGuard, MITRE ATLAS, and AgentDojo Inspect.

Browse datasets
Dataset collection

TEP reproduction data

Open the public Tennessee Eastman Process raw and detector-ready reproduction dataset.

Browse TEP data
Status lookup

CUB theorem registry

Look up the authoritative public theorem status before citing or implementing a CUB claim.

Open registry
Machine API

Public CUB API

Query the public-safe CUB projection by ID, repository, status, wiring, crate, or use case.

Open CUB API
Evidence guide

Workload certification evidence

Interpret ontology/compiler replay, CUB declarations, embedded receipts, runtime observations, exact-head certification, and owner authorization without collapsing them into one claim.

Read evidence guide
Agent map

Agent index

Give coding agents one compact map of entry points, collections, object URLs, and search routes.

Open agent map
Machine endpoint POST

Resource Library MCP

Connect an MCP client for library search, grounded answers, collection discovery, and CUB lookup.

Open MCP help

Find answers

Ask the library

Chat with the library: answers are written by Gemma (Cloudflare Workers AI) over collection manifests, catalog metadata, and AI Search semantic recall of the document content in scope.

All collections
Ask a question — Gemma answers from the library index and AI Search.

Cloudflare path

Knowledgebase architecture

R2 is the artifact store, D1 is now the live exact catalog, Workers AI answers source-grounded questions, and Vectorize/Queues/AI Gateway remain planned retrieval upgrades.

Updated 2026-08-01
Live Source library

R2 stays the binary and dataset backing store. Git keeps small locator manifests, source scripts, and regeneration notes.

trustfortress-resource-library
Live Catalog

D1 is the exact object catalog for deterministic provenance. The browser still falls back to R2 listing if D1 is unavailable.

trustfortress-library-index
Planned Retrieval

Search starts with collection metadata and manifests. Vectorize will add semantic recall after exact lookup.

trustfortress-library-vectors
AI live / gateway opt-in Answering

Workers AI answers from retrieved excerpts, routed through AI Gateway for response caching, analytics, and fallback when AI_GATEWAY_ID is configured.

trustfortress-library
  1. GitHub workflow or local agent publishes locator manifest and R2 objects.
  2. The Worker lists R2 prefixes and refreshes D1 through scheduled sync or protected manual reindex.
  3. D1 serves /api/catalog as the exact machine catalog, with R2 listing as a fallback if D1 is unavailable.
  4. Search and ask routes run deterministic collection/manifest lookup first.
  5. Workers AI generates answers from retrieved source excerpts and falls back to ranked search when inference is unavailable.
  6. Vectorize, Queues, and AI Gateway are planned upgrades for semantic recall, async indexing, and answer audit logs.
Capabilities JSON OpenAPI schema AI Search docs Cloudflare RAG guide

Library archive

Collections

2 folders
Q&A knowledge Knowledge base

Knowledge base

Public Q&A knowledge base for coding agents, onboarding admins, and customers: product overview, FAQ, honest learnings/scope, integrator and admin guides, workload certification evidence, plus claims / MCP / AI Gateway / AI Search guides. Ask via /api/ai-search or MCP ask_library.

knowledge-base/2026-07-31
Browse Manifest
docs/resource-library/knowledge-base/README.md
Current path

Resource browser

root
2 folders in this view
JSON
DIR
cubie-research
Directory cubie-research/
Open
DIR
cubie-tf
Directory cubie-tf/
Open

Developer access

API console

Live routes exposed by this Worker. Object and listing routes accept GET/HEAD; the answer route accepts POST.

Request URL
Open in browser
cURL 

        

        

          Response preview
          

          
Send a request to preview status, headers, and text or JSON bodies.