Static GitHub Pages research hub at https://ddugi.github.io/.
- /ai-signal/ — distinct editorial publication, JSON-driven demonstration stories, Signal Map and Weekly Signal. See ai-signal/README.md for authoring and build commands.
- / — research hub and roadmap
- /ai-readiness/ — five-dimension, equal-weight evidence self-assessment
- /gitops-maturity/ — cumulative Level 0–5 evidence gates
- /ai-agent-security/ — deterministic trust-boundary simulation
- /ai-infrastructure-report/ — proposed study, with no empirical findings
- /data/measurement-template.csv — header-only measurement schema
Run python3 -m http.server 8000 from this directory to preview locally. Deploy main at the repository root using GitHub Pages' branch source. .nojekyll keeps this plain static content. Pull requests and pushes run the quality workflow in .github/workflows/quality.yml.
For production governance, protect main, require pull requests, require the validate job, prevent force pushes, and enable private vulnerability reporting in repository settings. These server-side controls cannot be established by files alone.
No backend, external fonts, advertising trackers, or model calls. Answers in the interactive tools live in page memory and are cleared on reload. Users can export JSON results. AI Signal has optional, disabled-by-default Cloudflare Web Analytics support for aggregate audience measurement; its public privacy notice and project README describe the active state and collected metrics. GitHub Pages may retain normal hosting request logs. Do not add confidential organizational data to exported files intended for public sharing.
AI readiness: round(100 × sum(scores) / 15). All five scores must be integers from 0 through 3. Unknown scores zero; it is not inferred evidence of failure. GitOps: highest consecutive gate scoring 3, starting at gate 1. A gap prevents higher levels even if subsequent controls are present. These models are author-defined drafts, not validated indexes or certifications.
Three fixed scenarios: injection, excessive privilege, and identity scope. Task-boundary enforcement blocks first; capability policy blocks second; approval pauses third; with none enabled the UI marks the action as simulated execution. Audit changes visibility, not prevention. No actions are executed.
Add each future publication under its own slug/index.html and link it from the root. Preserve visible status labels, source attribution and assumptions. Never turn hypothetical benchmark values into reported findings. The initial site and content were developed with AI assistance.
Source code is MIT-licensed. Editorial text, research, diagrams, datasets, and branding remain all rights reserved unless explicitly stated otherwise; see LICENSE and NOTICE.md. Repository governance, security reporting, and contribution terms are documented in GOVERNANCE.md, SECURITY.md, and CONTRIBUTING.md. AI Signal publishes editorial, correction, and privacy policies under /ai-signal/policies/.
Publication information architecture was inspired by https://harnesstax.github.io/ without copying its design, text or results. Primary references (NIST, Google SRE, OpenTelemetry, OpenGitOps and OWASP) are linked on the relevant pages.