What does grep mean for a PDF?

Mempipe derives searchable text from the PDF while keeping the original file canonical. Grep runs exact literal or regular-expression matching over that derived view, then the workspace uses the source map to open the corresponding passage in the PDF.

Can grep search more than one PDF at a time?

Yes. Grep searches across a whole Space and can be narrowed with a glob such as raw/**/*.pdf, an exact path, or a file type. You do not have to open each PDF separately.

Can I search for identifiers, numbers, and exact clauses?

Yes. Exact search is useful when the characters matter: contract clauses, error codes, part numbers, quoted phrases, dates, and figures. Use literal mode for plain text or regular expressions for patterns.

Can AI agents grep the same PDFs?

Yes. Every Space exposes grep over MCP. Agents search the same index humans use, not a separate agent-only copy. Each result carries the file path, line number, passage text, and a citation link.

Does Mempipe replace the original PDF with Markdown?

No. The PDF remains the logical source of truth. Searchable Markdown, search units, and source anchors are managed projections that Mempipe keeps attached to that resource, not a shadow folder you have to reconcile.

Grep PDFs like source code. Then open the exact passage.

PDFs are easy to read one at a time and awkward to search as a corpus. Mempipe gives them grep semantics without turning them into a shadow folder: exact matches across your PDFs, filesystem-style scope, and a path back to the page and highlight in the original. Humans use it in the workspace; agents call the same capability over MCP.

Try Mempipe See how it works

The developer search instinct, applied to PDFs

You already know how to hunt through a codebase when the characters matter. Mempipe carries that workflow into document collections: give grep a pattern, narrow the file tree, and get the matching passage with its source path.

agent · MCPexact content search

grep({

pattern: "Section 7.2",

glob: "contracts/**/*.pdf",

regex: false

})

{

path: "contracts/vendor-msa.pdf",

line_no: 482, page_no: 31,

text: "Section 7.2: Supplier will retain all audit records for…",

citation_url: "https://app.mempipe.ai/@acme/legal/files/contracts/vendor-msa.pdf?cite=n_…"

}

Exact search is a different promise from relevance

Ranked search is useful when you know some of the words but not the exact phrase. Grep is for the opposite case: the wording, identifier, number, or pattern is the evidence. Search for a quoted clause, an invoice prefix, a regulation number, or every date matching a regular expression without asking a relevance model to interpret your intent.

Literal or regex
Match plain text exactly, or use a pattern when the shape matters more than one known value.
Filesystem scope
Search one PDF, one folder, a file glob, or every eligible document in the Space.
Corpus-wide results
Sweep the collection once instead of repeating Ctrl+F in every open document.

The PDF stays canonical

Most PDF-to-text workflows leave you with two collections to manage: the originals and a detached folder of converted text. Mempipe treats the searchable representation as a managed view of the same logical resource. Rename or update the source and the searchable view follows it once Mempipe processes the change, so search results still lead back to the original evidence.

Find the match, then inspect it in context

Grep returns the source path, matching line, and surrounding text. Open the result in the workspace, then follow its source anchor to the PDF page and highlighted passage. The searchable view helps you find the match; the original PDF is where you verify it.

One grep surface for humans and agents

Agents call grep over MCP against the same Space index people search in the workspace. MCP returns the matching path, line, and text; people can inspect the corresponding evidence through the workspace. There is no separate agent-only corpus to reconcile.

Learn how

Frequently asked questions

What does grep mean for a PDF?: Mempipe derives searchable text from the PDF while keeping the original file canonical. Grep runs exact literal or regular-expression matching over that derived view, then the workspace uses the source map to open the corresponding passage in the PDF.
Can grep search more than one PDF at a time?: Yes. Grep searches across a whole Space and can be narrowed with a glob such as raw/**/*.pdf, an exact path, or a file type. You do not have to open each PDF separately.
Can I search for identifiers, numbers, and exact clauses?: Yes. Exact search is useful when the characters matter: contract clauses, error codes, part numbers, quoted phrases, dates, and figures. Use literal mode for plain text or regular expressions for patterns.
Can AI agents grep the same PDFs?: Yes. Every Space exposes grep over MCP. Agents search the same index humans use, not a separate agent-only copy. Each result carries the file path, line number, passage text, and a citation link.
Does Mempipe replace the original PDF with Markdown?: No. The PDF remains the logical source of truth. Searchable Markdown, search units, and source anchors are managed projections that Mempipe keeps attached to that resource, not a shadow folder you have to reconcile.

Explore more

Build knowledge that comes with receipts.

Bring your sources. Find the evidence. Leave behind something worth keeping.

Try Mempipe Build your first wiki

Public beta. Start on Free and upgrade to Starter for more capacity. Bring your own API key for the in-app agent. Reading, searching, and editing files needs no key.