interdeep
# interdeep
Deep research plugin for [Claude Code](https://docs.anthropic.com/en/docs/claude-code). Extracts clean content from web pages and orchestrates multi-phase research sessions that produce structured markdown reports.
interdeep provides the content extraction backbone for research workflows. It pairs with companion plugins like [interject](https://github.com/mistakeknot/interject) (search) and [interknow](https://github.com/mistakeknot/interknow) (knowledge storage) for a complete research pipeline, but works standalone for extraction and report compilation.
## Installation
```bash
# Install from the Interverse marketplace
claude plugin install interdeep
# Install with browser extraction support (optional)
cd ~/.claude/plugins/interdeep && uv pip install -e ".[browser]" && playwright install chromium
```
## Usage
### Slash Command
```
/interdeep:research what are the best practices for MCP server design
/interdeep:research quick trafilatura vs readability comparison
/interdeep:research deep autonomous agent architectures and failure modes
```
### MCP Tools (programmatic)
The plugin exposes 4 MCP tools that any agent can call:
- **`extract_content`** — Extract clean text from a single URL
- **`extract_batch`** — Extract from multiple URLs concurrently
- **`compile_report`** — Compile findings into a structured markdown report
- **`research_status`** — Check extraction capabilities (trafilatura, Playwright availability)
## Architecture
```
src/interdeep/
server.py # FastMCP server — 4 tools
extraction/
trafilatura_ext.py # Fast extraction via trafilatura
playwright_ext.py # Browser fallback for JS pages
hybrid.py # Router: trafilatura-first strategy
reports/
markdown.py # Structured report compiler
skills/deep-research/ # 5-phase research orchestration protocol
agents/ # research-planner, source-evaluator, report-compiler
commands/research.md # /research slash command
```
## Design Decisions
- **Extraction + orchestration only** — interdeep does not own search. Search providers come from companion plugins (interject, interflux/exa).
- **trafilatura-first** — Fast path handles most pages. Playwright is an optional fallback for JavaScript-heavy sites, not a requirement.
- **Host-agent-as-brain** — MCP tools are stateless utilities. Research intelligence lives in the skill definition and agent prompts, executed by Claude as the host agent.
- **Graceful degradation** — Every component works standalone. Missing companion plugins reduce capability, they do not break the pipeline.
## License
MIT
TDQS
Scored across 4 tools
Each tool has a clearly distinct purpose with no ambiguity: compile_report creates reports from research, extract_batch handles multiple URLs concurrently, extract_content extracts from single URLs with fallback options, and research_status provides system status. The descriptions clearly differentiate their functions, making misselection unlikely.
All tools follow a consistent verb_noun pattern (compile_report, extract_batch, extract_content, research_status) with clear, descriptive names. There are no deviations in naming conventions, making the set predictable and easy to understand.
With 4 tools, the count is reasonable for a research/content extraction server, though slightly lean. Each tool earns its place by covering distinct aspects of the workflow, but the scope might benefit from additional tools for tasks like managing research sources or advanced report formatting.
The tool surface covers core research and extraction workflows well, including content extraction (single and batch), report compilation, and system status. Minor gaps exist, such as tools for editing reports, managing extracted data, or integrating with external research databases, but agents can work around these with the provided tools.