Conversation
The plugin already injects a display:none hint, but WebFetch-style tools summarize a page and drop hidden DOM, so a neutral fetch misses it (which is exactly why an LLM reading the site never learns /llms.txt exists). Move the hint into visible, server-rendered HTML, plus redundancy: - index.md: a visible (non-JS, SSG-baked) footer line pointing to /llms.txt and /llms-full.txt -- the load-bearing signal a WebFetch will surface. - config.mts: link rel=alternate type=text/plain to both files for agents that parse the head. - public/robots.txt: a visible "# LLM index: /llms.txt" line. Not done: the matching HTTP Link response header (rel=alternate to /llms.txt). GitHub Pages is a static host and cannot set custom headers. Add it if the site moves to a host that can (Cloudflare Pages / Netlify). Acceptance gate (run after deploy): a neutral-prompt WebFetch of the landing page returns the /llms.txt line. Verified locally that the line is baked into dist/index.html as visible text. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…esign Guidelines, Ethical Design Policy Restructure the handbook around OSBR's own spine — a 3-lens quality gate (Reliability / Security / Sustainability) as the definition of done, the AI-native operating model, and the pre-work planning phase — while keeping the existing policies as the standards the gate references rather than restates. - doc/quality-gate.md (new): the three checks every change clears before Done; each lens carries code-level standards and references the Infrastructure Planning / Security policies and NFR for posture. - doc/ai-usage-guideline.md (new): one engineer owning the whole with AI beside them — data boundaries, provider resilience, day/night rhythm, voice, policies-as-plugins, building for AI users. - doc/design-guidelines.md (new): OSBR's UX/interaction standards (accessibility floor, self-explanatory UI, every state designed, modeless and reachable) — the counterpart to the code Style Guide. - doc/ethical-design-policy.md (new): no deceptive/dark patterns; honest, symmetric, freely-given consent. - development-guide.md: 1-10 Quality Gate and 1-11 AI Usage Guideline as short pointers to the new pages; add 2-2 "Repositories Are Not a Support Channel" and 2-4 "Planning & Shaping". - sheq-policy.md: add policy-conformance auditing under Quality Assurance. - config.mts: wire the four new pages into the sidebar. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add a light value framing where the new pages talk about standards, using the full value names: - quality-gate: the three-check bar as Be Nice (care) / Be Kind (ownable) / Be Strong (find our own gaps). - sheq-policy: policy-conformance auditing framed by the three values. - ethical-design-policy: honest design as Be Nice made concrete. - design-guidelines: getting the experience right as Be Nice in the interface. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Address over-consolidation: the hubs added earlier (Quality Gate, AI Usage Guideline, Planning & Shaping, Design Guidelines) summarised ~50 source drafts into thin sections. Restore workaholic-level depth by publishing each topic as its own full standard page, with the hubs kept as concise overviews that reference them. New pages (31): - Engineering Standards (Quality Gate references): testing-standards, observability-resilience, incident-management, code-review, ci-cd-pipeline, application-security, access-control, data-protection, supply-chain-risk, architecture-standards, repository-documentation-standards, api-design. - AI operating model (AI Usage Guideline references): ai-data-handling, multiple-ai-agents, overnight-ai, weekly-ai-quota, voice-input, policies-as-plugins, building-for-ai-users. - Planning (Planning & Shaping references): market-research, requirements-modeling, verify-before-building, cost-estimation, it-investment-evaluation, legal-compliance, domain-terminology, capability-over-track-record. - Design (Design Guidelines references): accessibility, self-explanatory-ui, modeless-design, interaction-design. Each page is OSBR's own voice, RFC-2119 MUST/SHOULD, tied to Be Nice / Be Kind / Be Strong, standard industry terms, no competitor named. Wiring: - quality-gate: each lens now lists the standards it holds work to. - ai-usage-guideline / design-guidelines / development-guide (Planning & Shaping): concise overviews with in-depth links to the full pages. - config.mts: pages nested under Development Guide (Design Guidelines, Planning & Shaping, Quality Gate, AI Usage Guideline) in the sidebar. - verified: no dangling internal links across doc/. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- meeting-recording (new): record/transcribe only with prior all-party consent, scope-limited use, retention/deletion; nested under AI Usage Guideline. - privacy-policy (new): OSBR's organisation-level privacy notice (個人情報保護方針), distinct from the /data-protection engineering standard; added to Policies. Addresses #64. - NFR: reconcile WCAG version 2.1 → 2.2 AA to match the HTML/CSS guide and the new Accessibility standard, and link to it. - config.mts: sidebar entries for both new pages. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
… standard Point §1 at the /code-review standard (always-on AI-automated review as the default gate, security one of its three concerns, human accountable in the loop) instead of mandating a human security reviewer — completing item A of epic #92. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A subagent left stray closing tags at the end of the file, which broke the VitePress (Vue) build with "Invalid end tag". Removed them; docs:build now passes cleanly (55 files, llms-full.txt generated). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The sidebar entry linked to the whole /development-guide page; point it at the #_2-4-planning-shaping section anchor so it lands on the Planning & Shaping section directly. Verified: docs:build passes. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Integrity review of the sidebar vs. page content found two gaps: - Design Guidelines existed as a page and in the sidebar but was not mentioned in the Development Guide body; added §1-9 Design Guidelines (Database Guidelines, Quality Gate, AI Usage Guideline renumbered 1-10..1-12). - Security Policy was an orphan — linked from many pages but absent from the sidebar; added it to the Policies group. Verified: docs:build passes; no dead sidebar links, no orphan pages, no dangling internal links, Planning & Shaping anchor resolves. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The privacy and legal pages were merged from Japan-context source drafts, so they led with Japanese law (APPI, 特定商取引法) and omitted Malaysia. OSBR is registered in Malaysia (home jurisdiction), has a Japanese parent studio, and aims to follow international standards — so every jurisdiction reference is now ordered Malaysia -> Japan -> EU/US, adding Malaysia as the lead without removing the Japanese or GDPR content. - privacy-policy: rewritten on Malaysia's PDPA 2010 (Act 709) + 2024 amendment — the seven principles, the Personal Data Protection Commissioner, mandatory breach notification, DPO, data portability, cross-border transfer — with APPI and GDPR/OECD/ISO as the Japan and international layers. - legal-compliance: Malaysia leads (PDPA; Consumer Protection (Electronic Trade Transactions) Regulations 2024 + Electronic Commerce Act 2006 for seller disclosure), then Japan (APPI, 特定商取引法), then EU/US. - incident-management: breach notification leads with the PDPA Commissioner, then APPI/PPC, then GDPR Art. 33/34. - data-protection: PDPA added as the lead legal anchor throughout. - ai-data-handling, meeting-recording: jurisdiction wording reordered MY->JP->EU. Facts verified against pdp.gov.my and authoritative sources, not memory. Verified: docs:build passes; no dangling links, no orphan pages. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
An adversarial content review across the hubs, standards, and legal pages found requirement-level mismatches and board/wording contradictions: - quality-gate: the gate holds at Impl Review (before merge) — the previous "between Impl Review and Done" wording contradicted the board, where Shipping/release sits in between; same fix in development-guide §1-11. - quality-gate: masking personal data before a log or AI context is a MUST, matching incident-management's inviolable rule (was SHOULD). - testing-standards §3-7: critical paths MUST run against real dependencies (the Quality Gate's bar); boundary code more broadly stays SHOULD. - development-guide: Impl Review row now reflects the decided review stance — AI review is the default gate, a human interprets findings and owns the merge; Shipping/Done rows no longer disagree with the flowchart on when the issue closes; legacy typos fixed (piplines, short-descripted, belongs our); bookmark-bar rule aligned to the Security Policy's actual scope (project or client names). - ai-usage-guideline: add Meeting Recording to the in-depth list (it is nested there in the sidebar). Verified: docs:build passes; cross-link consistency and dangling-link checks clean. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The review mechanism was written as an "always-on" gate, which is not how it works: the AI agent lives on each member's own machine, carrying OSBR's review tooling — the /code-review plugin and, for security-sensitive surfaces, the asvs-audit plugin (osbrjp/DevTool) — and the review is performed when a change is declared Impl Review, before merge. - code-review: mechanism, trigger, and tooling stated; "always-on" framing replaced with run-at-Impl-Review throughout. - quality-gate, development-guide (Impl Review row), security-policy §1: definitional wording aligned. - 13 other pages: referential "the always-on review" rephrased to "the AI code review". Verified: docs:build passes; only remaining "always-on" is infra's always-on servers. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
LLMs don't discover /llms.txt — page-reading agents miss the LLM index
github-actions
Bot
changed the title
github-actions
Bot
added
the
release-candidate
sn0wm1ku
approved these changes
Jul 16, 2026
sn0wm1ku
left a comment
sn0wm1ku
left a comment
Member
There was a problem hiding this comment.
Release candidate approved — prepare-release and run-tests checks both green; contents previously reviewed on the source PRs.
🤖 Posted by Claude Code (on the user's request)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Uh oh!
There was an error while loading. Please reload this page.