AI SummaryA skill-format reference for any agent (Claude Code, Codex, Cursor, Hermes, plain-MCP) calling this server. For human-readable installation instructions, see README.md. For harness-agnostic routing logic and the pasteable instruction block (drop into your harness's global / or a standalone agent's
Install
Copy this and paste it into Claude Code, Cursor, or any AI assistant:
I want to install the "gigaxity-deep-research" skill in my project. Please run this command in my terminal: # Install skill into your project mkdir -p .claude/skills/gigaxity-deep-research && curl --retry 3 --retry-delay 2 --retry-all-errors -o .claude/skills/gigaxity-deep-research/SKILL.md "https://raw.githubusercontent.com/yoloshii/gigaxity-deep-research/main/SKILL.md" Then restart Claude Code (or reload the window in Cursor) so the skill is picked up.
Description
Deep research MCP server wrapping Qwen3-30B-A3B-Thinking via OpenRouter. Use when an agent needs cross-source synthesis with citations, exploratory expansion of an unfamiliar topic, chain-of-thought reasoning over evidence, or fast conversational lookups grounded in live web search. Exposes seven MCP tools — three primitives (search, vertical_search, research) plus four deep-research tools (discover, synthesize, reason, ask) — with matching REST endpoints for each.
Deep-research tools — discrete steps, drive each independently
ask(query) # Fast conversational answer (direct LLM, no search) discover(query) # Exploratory expansion + gap detection synthesize(query, sources) # Citation-aware fusion of pre-gathered content reason(query, sources) # Synthesize + explicit chain-of-thought depth control `
MCP tools (stdio)
All seven tools return markdown strings, not JSON. Every tool also accepts an optional openrouter_api_key: str | None = None for per-request key override (omitted from the signatures below for brevity). ` mcp__gigaxity-deep-research__search( query: str, top_k: int = 10, focus_mode: Literal["general", "academic", "documentation", "comparison", "debugging", "tutorial", "news"] | None = None ) -> str # markdown ranked results, no LLM call; focus_mode routes the SearXNG category lane mcp__gigaxity-deep-research__vertical_search( query: str, vertical: Literal["videos", "science", "it", "docs", "packages", "general"] = "videos", top_k: int = 10 ) -> str # markdown results from one SearXNG category + failing engines, no LLM call mcp__gigaxity-deep-research__research( query: str, top_k: int = 10, reasoning_effort: Literal["low", "medium", "high"] = "medium" ) -> str # markdown: search + synthesis + citations mcp__gigaxity-deep-research__ask( query: str, context: str = "" ) -> str # the LLM's response text, direct call (no search) mcp__gigaxity-deep-research__discover( query: str, top_k: int = 10, identify_gaps: bool = True, focus_mode: Literal["general","academic","documentation","comparison","debugging","tutorial","news"] = "general" ) -> str # markdown: knowledge landscape + gaps + sources mcp__gigaxity-deep-research__synthesize( query: str, sources: list[dict], # {title, content, url?, origin?, source_type?} style: Literal["comprehensive","concise","comparative","academic","tutorial"] = "comprehensive", preset: Literal["comprehensive","fast","contracrow","academic","tutorial"] | None = None ) -> str # markdown: synthesis + citations (+ contradictions if preset enables) mcp__gigaxity-deep-research__reason( query: str, context: str = "", sources: list[dict] | None = None, style: Literal["comprehensive","concise","comparative","academic","tutorial"] = "comprehensive", reasoning_depth: Literal["shallow", "moderate", "deep"] = "moderate" ) -> str # markdown: CoT response. If sources is provided, # synthesizes over them; otherwise depth-controlled CoT # over the model's own knowledge plus optional context. ` For the JSON / typed shape, call the matching REST endpoints (/api/v1/<tool>) — see docs/reference/rest-api.md.
Gigaxity Deep Research
A skill-format reference for any agent (Claude Code, Codex, Cursor, Hermes, plain-MCP) calling this server. For human-readable installation instructions, see README.md. For harness-agnostic routing logic and the pasteable instruction block (drop into your harness's global CLAUDE.md / AGENTS.md or a standalone agent's system prompt), see CLAUDE.md.
Quick Start
The MCP server exposes seven tools — three primitives plus four deep-research tools. Pick one based on query type, call it, return the result to the user verbatim. `
Discussion
Health Signals
My Fox Den
Community Rating
Sign in to rate this booster