How to let Claude Code or Cursor search the PDFs and Word documents in a repository Leave feedback

Coding agents search a repository with grep, and grep cannot read inside a PDF datasheet or a Word specification — the text is compressed. Register the GroupDocs.Search MCP server locally in Claude Code or Cursor, index the repository’s docs/ folder, and the agent can search and quote those documents like code:

claude mcp add groupdocs-search -- docker run --rm -i \
  -v "$(pwd)":/data ghcr.io/groupdocs-search/search-net-mcp:latest

Index the docs folder, then tell me what the sensor datasheet says about the operating temperature and which ADR decided the telemetry storage.

Note

The commands and config snippets on this page are for the .NET build of the server, which ships as a Docker image only. Installation and client setup: MCP server for .NET. Other platforms will expose the same tools with their own launch command; everything else on this page applies unchanged.

• Docker; run the command above from the repository root, so the repository is /data. On Linux add --user "$(id -u):$(id -g)" after -i.
• Cursor and VS Code: the same docker run arguments in .cursor/mcp.json or .vscode/mcp.json — Register in AI clients.

In a test repository with a PDF datasheet and two Word documents, grep -ri "TimescaleDB\|operating temperature\|I2C" docs/ finds nothing — PDF and DOCX store their text compressed. The only hit is a code comment pointing at the datasheet. The same questions through the index, captured from server 26.10.1:

-> add_to_index {"documentsPath":"docs","indexName":"gateway"}
<- Indexed 'docs' into index 'gateway': 3 document(s) added,
   3 in the index.

-> search {"query":"TimescaleDB","indexName":"gateway"}
<- documentsFound 1: docs/adr-007-telemetry-storage.docx

-> get_text_fragments {"query":"I2C","indexName":"gateway"}
<- 2 document(s) match.
   === docs/requirements-v2.docx (1 occurrence(s)) ===
     [content] …REQ-15: the I2C bus runs at 400 kHz.…
   === docs/tmp117-sensor-datasheet.pdf (1 occurrence(s)) ===
     [content] …I2C address selection: ADD0 pin to GND gives 0x48…

(Paths shown relative to the repository root, as they are when the repository is mounted at /data.)

user              agent                         Search MCP server
 |  "Which contracts mention a termination penalty?"           |
 |---------------->|                                           |
 |                 |-- add_to_index ------------------------->|  reads the
 |                 |   {"documentsPath":"contracts",           |  files once,
 |                 |    "indexName":"contracts"}               |  stores text
 |                 |<- "5 document(s) added, 5 in the index" --|
 |                 |-- search ------------------------------->|
 |                 |   {"query":"termination AND penalty"}     |
 |                 |<- {"documentsFound":3, "documents":[ ... |  which files
 |                 |     "contracts/acme-...-agreement.pdf",   |
 |                 |     "contracts/globex-...-agreement.docx",|
 |                 |     "contracts/initech-nda.docx" ]}       |
 |                 |-- get_text_fragments ------------------->|
 |                 |   {"query":"\"termination penalty\""}     |
 |                 |<- === contracts/acme-...-agreement.pdf ===|  the passages
 |                 |   ...a termination penalty of three       |
 |                 |   monthly fees applies...                 |
 |<- "Two contracts carry one: Acme (3 monthly fees, 12.3),    |
 |    Globex (50% of the remaining fees). Initech excludes it."|

The loop is the same as for any folder; what changes is that the agent now has a retrieval tool for the documents next to the code — the datasheet behind a constant, the ADR behind a design choice, the requirement behind a test.

The index is written to .groupdocs-search-index/ at the root of the mounted folder. Add it to .gitignore:

.groupdocs-search-index/

After pulling changes, ask for an update_index — it re-reads only what changed (keep an index current).

Answers that cite docs/…pdf and docs/…docx passages alongside the code the agent already reads — “the datasheet gives an operating range of -55 C to 150 C; ADR-007 chose TimescaleDB with 13-month retention.”

Why not let the agent convert the PDFs to text first? It can, file by file, every time. The index does it once, keeps it current with one call, and answers which file before reading any.

Does it index the source code too? It indexes what you point it at. Code is better searched with the agent’s own tools; point add_to_index at docs/ (or wherever the documents are).

Will it slow the agent down? The first indexing takes a moment. On our small test index, searches returned in 80–240 ms, and the first call of a session in about a second.

Agents follow project instructions about when to use a tool. Paste this into the instruction file your agent reads (CLAUDE.md, AGENTS.md, .github/copilot-instructions.md, .cursorrules):

PDF, DOCX and XLSX files under docs/ are indexed in the groupdocs-search MCP index
'docs'. grep cannot read them: use search (quote phrases) and get_text_fragments
for anything in specs, datasheets or ADRs, and cite the file.
Run update_index (indexName: docs) after pulling changes.

• Evaluation mode: an index holds at most 100 documents and a search returns at most 10 — the response says so in its evaluation field. Licensing.
• Full-text, not semantic: it matches words, not meaning. Ask the agent to try variants (OR, fuzzy: true).
• No OCR: scans without a text layer are indexed by their file properties only.
• Docker only: there is no dnx channel for this product (why).

• Search documents with AI agents
• Search a folder of contracts
• Quote matching passages
• Keep an index up to date
• Search email archives
• Find documents by date and amount
• On-premise deployment
• Tools reference · Licensing · Troubleshooting & FAQ
• Questions: Search forum

• grep can’t read PDF: give your coding agent a document index — the same scenario on the GroupDocs blog, with worked examples and the trade-offs.