# Blerbz AI Agent Identity, Integration, And Access Plan Status: implementation-backed plan Last updated: 2026-05-19 ## Objective Make Blerbz legible, trustworthy, and directly usable by AI agents through stronger identity files, machine-readable pricing, structured data, well-known discovery, markdown fallbacks, NLWeb/MCP-style access, and a clear manual path for external authority surfaces. ## Phase 1 - Immediate Discovery And Identity Fixes Autonomous actions: - Publish root and well-known agent files: `/agents.md`, `/.well-known/agents.md`, `/.well-known/agent-card.json`, and `/.well-known/agent-skills/index.json`. - Publish `/home` as the crawlable home overview and `/index.md` as the markdown site index. - Publish `/pricing`, `/pricing.md`, and `/.well-known/pricing.md`. - Publish `/schemamap.xml` and reference it from robots.txt with `Schemamap:`. - Add HTTP Link headers for sitemap, llms, markdown alternate, OpenAPI, agent card, and MCP card. - Block training-only crawlers that are not needed for Blerbz user-facing retrieval while keeping user-facing AI retrieval crawlers allowed. Acceptance checks: - `curl -i https://www.blerbz.com/agents.md` returns markdown. - `curl -i https://www.blerbz.com/pricing` returns the crawlable HTML pricing page. - `curl -i https://www.blerbz.com/.well-known/agent-skills/index.json` returns v0.2.0 index JSON with SHA-256 digests. - `curl -i https://www.blerbz.com/home` returns the crawlable HTML home overview. - `curl -i https://www.blerbz.com/index.md` returns the markdown site index. - `curl -I https://www.blerbz.com/` redirects to `/window`. - `curl https://www.blerbz.com/robots.txt` includes `Schemamap:`, `Sitemap:`, allowed retrieval agents, and disallowed training-only crawlers. ## Phase 2 - Agent Integration Autonomous actions: - Add NLWeb-style `POST /ask` with JSON and SSE responses. - Add a minimal MCP JSON-RPC endpoint at `/mcp` with `initialize`, `tools/list`, `tools/call`, `resources/list`, and `resources/read`. - Add `/.well-known/mcp/server-card.json` and `/.well-known/mcp.json`. - Add JSON error responses for unmatched API routes. - Add rate-limit headers on API responses. - Add read-only `/api/v2/batch` for bounded agent smoke tests. - Add `/.well-known/sandbox.json` and `/api/v2/sandbox/status` to make the public test environment explicit. Acceptance checks: - `curl -s -X POST https://www.blerbz.com/ask -H 'content-type: application/json' -d '{"query":{"text":"What is Blerbz?"}}'` returns `_meta.response_type=answer`. - `curl -N -X POST https://www.blerbz.com/ask -H 'content-type: application/json' -H 'accept: text/event-stream' -d '{"query":{"text":"What is BLERBZ100?"},"prefer":{"streaming":true}}'` returns `start`, `result`, and `complete` SSE events. - `curl -s -X POST https://www.blerbz.com/mcp -H 'content-type: application/json' -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'` returns tool schemas. - `curl -s https://www.blerbz.com/api/v2/sandbox/status` returns `type=sandbox-status`. - `curl -s -X POST https://www.blerbz.com/api/v2/batch -H 'content-type: application/json' -d '{"requests":[{"id":"pricing","path":"/pricing.md"}]}'` returns a read-only batch result. ## Phase 3 - Structured Data And Trust Autonomous actions: - Add root JSON-LD breadth for Organization, NewsMediaOrganization, WebSite, SoftwareApplication, Product, WebAPI, DataFeed, OfferCatalog, and SpeakableSpecification. - Link pricing, API, llms, and agent resources from schema graph and HTTP Link headers. - Add operation IDs and typed schemas for public agent, batch, and sandbox operations in OpenAPI. - Keep sameAs limited to verified official profiles until external authority pages exist. Manual actions: - Create or verify Wikidata only after an externally verifiable public source exists. - Create Wikipedia only if Blerbz has independent reliable coverage. Do not submit a promotional page or self-sourced article. ## Wikipedia Draft Content And Instructions Do not publish this as-is unless Blerbz has independent reliable sources. Wikipedia requires notability demonstrated by significant coverage in independent secondary sources. Suggested neutral draft: ```text Blerbz is a news aggregation and news intelligence platform that uses artificial intelligence to rank breaking news and trending stories. The platform includes a real-time news feed called Window, a daily ranked list called BLERBZ100, reading and search surfaces called Paper and Focuz, and developer access through the NUCLEUZ API. The service describes its ranking approach as using freshness, velocity, source, and impact-oriented signals to surface stories that are gaining importance. Blerbz also includes AI-assisted story analysis and media workflows through ARTEMIZ and STREAM. ``` Publication instructions: 1. Collect at least 3 independent, reliable secondary sources that discuss Blerbz in detail. Product pages, GitHub, press releases, social posts, and founder-authored content are not enough. 2. Create a Wikipedia user account and draft in Draft namespace, not mainspace. 3. Use a neutral title such as `Draft:Blerbz`. 4. Cite every factual claim with independent sources. 5. Avoid claims such as "best", "first", "leading", "revolutionary", or unsupported market comparisons. 6. Disclose conflict of interest on the draft talk page if you are affiliated with Blerbz. 7. Submit through Articles for Creation only after sources establish notability. ## Phase 4 - Ecosystem Distribution Manual or semi-autonomous actions: - Publish `@blerbz/sdk` to npm when the package is ready and versioned. - Add a CLI package only if it wraps real developer workflows such as `blerbz feeds latest`, `blerbz search`, and `blerbz sync`. - Register agent skills in external directories only after the well-known index is live. - Submit MCP metadata to relevant registries only after `/mcp` passes initialize, tool-list, tool-call, and resource-read smoke tests. ## Phase 5 - Ongoing Monitoring Autonomous recurring checks: - Probe `/llms.txt`, `/llms-full.txt`, `/agents.md`, `/pricing.md`, `/.well-known/agent-card.json`, `/.well-known/agent-skills/index.json`, `/.well-known/mcp/server-card.json`, `/openapi.json`, `/ask`, and `/mcp`. - Verify content type, cache headers, Link headers, digest integrity, and non-HTML API error responses. - Re-run after deploy and after any domain, redirect, locale, or robots change.