OllaSuper
⚡ Agents Deep Research
Research · openfang Agent

Deep Research Agent

Multi-step research over the live web — structured briefs with citations. Powered by Tongyi DeepResearch, purpose-built for long horizons.

Chat with this Agent → How it works
Schedule
Continuous (manual trigger)
Model
Tongyi DeepResearch 30B A3B
openrouter/alibaba/tongyi-deepresearch-30b-a3b
Approval
Read-only research — no approval required
Deep Research only produces briefs. It never sends emails, modifies pipelines, or publishes content.

What it does

Receives a research task. First restates the question + decomposes into 3-5 sub-questions. Builds an explicit plan, then iterates: ollagraph_scrape, intel_whois, extract_structured, intel_ssl, keyword_extract — cross-referencing every claim, flagging conflicts between sources, surfacing what couldn't be found. Returns a structured brief: TL;DR / Detailed findings (one section per sub-question) / Conflicts / Tried-but-failed / Numbered sources with timestamps. Every factual claim has a citation; no uncited claims allowed.

Tools it can use

ollagraph
mcp_ollagraph_scrape Clean markdown of any URL
mcp_ollagraph_extract_structured JSON-LD / OpenGraph / schema markup
mcp_ollagraph_intel_whois Domain ownership + age + registrar
mcp_ollagraph_intel_ssl TLS cert details for security research
mcp_ollagraph_seo_keyword_extract What a page actually emphasizes
mcp_ollagraph_convert_html_to_markdown Clean any HTML chunk
web
web_fetch Fallback for ollagraph-blocked URLs
filesystem
file_write Briefs in research/ + handoffs/<agent>/
memory
memory_store Cached source URLs + brief summaries for re-use

Workspace

Paths under /var/lib/openfang/workspace/ this Agent reads or writes:

research/handoffs/<agent>/shared.research.*

Sample interaction

You
Deep research: ollagraph.com. (1) What does it do? (2) Domain age and owner? (3) Top 3 keywords on the homepage?
Deep Research Agent
### TL;DR - **Product:** Web-intelligence layer for AI agents — fetch, extract, audit, reason about the live web through a unified API - **Domain:** Registered 2026-05-24 via BigRock Solutions Ltd; Cloudflare DNS - **Top keywords:** "string" (6), "ollagraph" (5), "extract" (4) — technical, data-extraction focused ### Detailed Findings #### 1. What does the product do? Ollagraph positions itself as a single API for AI agents that need to interact with the live web. Capabilities include: scrape with stealth + residential proxies, structured extraction with JSON schemas, AEO audits, intelligence (WHOIS, SSL), observability for tool calls. Marketing emphasis on "replaces fragmented stack of scrapers + parsers + auditors." [1] #### 2. Domain registration Registrar: BigRock Solutions Ltd. Creation: 2026-05-24. Expiry: 2027-05-24. Nameservers: DAN.NS.CLOUDFLARE.COM, PAT.NS.CLOUDFLARE.COM. Very recent — high product-velocity stage. [2] #### 3. Homepage keyword emphasis Weighted keyword extract shows "string" (6), "ollagraph" (5), "extract" (4). The technical lean is intentional — this is an audience of agent builders, not marketers. [3] ### Conflicts / Uncertainties None — all three sources corroborated. ### What we tried but couldn't find Founder names; funding info; user counts. Not on the homepage; would need adjacent sources (LinkedIn, Crunchbase) — not available via ollagraph tools. ### Sources 1. Ollagraph homepage markdown — retrieved 2026-06-11 2. WHOIS ollagraph.com — retrieved 2026-06-11 3. Keyword extract on ollagraph.com — retrieved 2026-06-11 Saved to research/2026-06-11__ollagraph.md; 3 tool calls; cost $0.02.