Darwin RAG

Install

Works with Claude Code, Codex, VS Code, Cursor, Windsurf, Claude Desktop. Read the repository before you give any tool access to your data.

npx agentmarket install connector/brightdotdev-darwin-rag

About

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this server can do and what it asks for, from its tool list, install config and code. Checked Oct 25, 2026.

Tools

8 read · 2 write · 1 delete

Secrets

3 (API keys, tokens or passwords)

Install scripts

None found

Runs as

Python package

Tools (11)

Read from the source code; the running server may differ. Descriptions are the server’s own words; hidden characters are shown as [U+…].

• create_storeCreate Store

  Create a new DarwinStore at the given path for isolating data. Usage tips: - Use this to create project-specific stores before running pipeline. - Pass the returned store ID to other tools via the 'store' parameter. - name is optional; if omitted, the store is identified by path only.

• file_statusFile Statusread-only

  Get detailed status for a single file across all pipeline stages. Usage tips: - Use this to check what artifacts exist for a specific file. - filename must match the original filename used during ingestion. - Pass store (name/path/ID) to check a file in a specific store.

• get_logsGet Logsread-only

  Query Darwin session logs (oldest first for LLM readability). Parameters: session: Filter by session name. Exact values: "PIPELINE" — full pipeline orchestrator run "RETRIEVAL" — search/retrieval orchestrator run "Ingestion" — parsing + chunking stage "Indexing" — BM25 + embedding stage Omit for all sessions. level: Filter verbosity. "verbose" — SUCCESS + WARNING + ERROR (default) "quiet" — SUCCESS + ERROR only Omit for verbose. limit: Max log entries to return (1-1000, default 100). Notes: - lo…

• get_schemaGet Structured Schemaread-only

  Get schema information for structured files (JSON arrays, CSVs). Three modes of operation: 1. Single file: file="tickets.csv" Returns schema for one file (keys, types, record_count). 2. Multiple files: files=["tickets.csv", "orders.json"] Returns schemas mapped to each file. Also detects overlapping keys (keys that appear in multiple files with potentially different meanings). 3. List all: list_all=true Returns schemas for ALL indexed structured files in the store. 4. Validate filters: validate_filters={"status": "open", "priority": "high"} Validates filter keys/types again…

• get_search_result_by_idGet Search Result By IDread-only

  Load a saved search result by its filename. Pass the filename from get_search_results (e.g. "machine-learning_20260718_143022.json"). Returns the full RetrievalContext with passages, metadata, and scores. If not found, returns an error suggesting you run search_darwin with save=true. Pass filter=true to strip bloat (clean, audit, candidates_found, raw_results, etc.).

• get_search_resultsGet Search Resultsread-only

  List all saved search results. Returns a list of saved result files with metadata (query, method, passage count, max score, timestamp). This listing is already minimal — no filter param needed. Use get_search_result_by_id to load a specific result (pass filter=true for a stripped-down view). Tip: The darwin://search/results resource provides the same data without a tool call — use it for quick context lookups.

• list_filesList Filesread-only

  List all files tracked by a store with their pipeline status. This is the primary tool for checking pipeline status. Usage tips: - Returns the pipeline report — each file shows completed stages. - Pass filename to filter to a single file. - Pass store (name/path/ID) to list files from a specific store. - Use this to find exact filenames before running purge_artifacts. - Call without arguments to see all files from the default store.

• purge_artifactsPurge Artifactscan change or delete data

  Delete pipeline artifacts for specific files. Parameters with enums (pass exact string): stages: "ingestion" (or: ingest, extract, parse, chunk, ocr, clean) "indexing" (or: index, embed) Usage tips: - Always pass dry_run=true first to preview what will be deleted. - Stages: 'ingestion' removes markdown/records; 'indexing' removes embeddings/vector/bm25. - filenames must be exact matches to the original filenames used during ingestion. - Use list_files to see available filenames and their pipeline status.

• run_pipelineRun Pipeline

  Run ingestion/indexing pipeline stages. Parameters with enums (pass exact string): stages: "ingestion" (or: ingest, extract, parse, chunk, ocr, clean) "indexing" (or: index, embed) Defaults (applied by PipelineParameters if not provided): target: "" (empty — use files param to target specific files) ocr: false, recursive: false, precise: false, rerun: false, threshold: 0.5, idf_smoothing: 1.5, only_kw: false, only_semantic: false, strict: true Usage tips: - target: a file path, directory path, or URL to ingest. - Directory: ingests all files within it (use recursive=t…

• search_darwinSearch Darwinread-only

  Search the DARWIN knowledge base for relevant passages. Parameters with enums (pass exact string): method: "hybrid" (default) | "semantic" | "keyword" merge_strategy: "adjacent" (default) | "none" | "aggressive" fallback_strategy: "both" (default) | "expand" | "relax" Defaults (applied by RetrievalParameters if not provided): top_k: 5, max_passages: 5, max_chars: 6000, min_passage_score: 0.15, fill_gaps: false, max_per_source: 2, filter_structural: true, structural_penalty_factor: 0.5, diversity_rerank: true, include_audit: false, rerank: false Usage tips: - Check 'summary' in…

• search_listsSearch Structured Listsread-only

  Search structured data (JSON arrays, CSVs) by field values with hybrid text fallback. This tool routes through RetrievalOrchestrator.search_lists(), following the same orchestration pattern as search_darwin. It handles store resolution, logging, session management, and result formatting. Parameters: query: Text query for hybrid fallback (optional if filters provided). filters: Dict of field-value pairs to filter by (e.g. {"status": "failed"}). mode: "auto" (try structured, fallback to text), "structured" (filter only), or "text" (hybrid only). store: Store name/path/ID (optio…

Signals

Stars

2

Forks

0

Open issues

0

Weekly downloads

49

Last commit

Oct 14, 2026 (14 days ago)

Last release

Aug 26, 2026 (2 months ago)

Last package publish

Aug 26, 2026 (2 months ago)

From GitHub (fetched Oct 24, 2026) and the package registry.

Details

Registry name

io.github.Brightdotdev/darwin-rag

Version

0.2.18

Auth

API key

Packages

pypi darwin-rag

License

MIT

Repository

github.com/Brightdotdev/DARWIN

First seen

Oct 24, 2026

Last checked

Oct 25, 2026

Source

Official MCP Registrygithub-graphql