Before 100xprompt changes a line, it learns your codebase the way a strong engineer would: it searches for the right places, follows references between them, and reads only what matters. That’s why its edits land in the correct file, respect your existing patterns, and hold up in large, unfamiliar projects. This page covers:

How context builds

You describe an outcome. 100xprompt turns that outcome into targeted code reading - a small number of precise searches and reads instead of loading your whole repository. It widens when it needs more context and narrows the moment it has enough to act.
You never run these searches yourself. You state the goal in plain language, and 100xprompt chooses the right combination of searching, navigating, and reading to get there.

What you can do

Codebase understanding is four complementary skills - finding code, navigating meaning, reading with judgment, and reaching beyond the repo.

Search by content

Find every place a behavior lives using full regular expressions across a repo of any size.

Search by file pattern

Locate files by name or path shape with glob patterns, honoring your ignore rules.

Follow symbols

Jump to a symbol’s real definition, references, implementations, and callers - semantically, not by guessing.

Read intelligently

Open large files in sections, and read images, PDFs, and notebooks - not just plain text.

Search by content

When you describe what the code does rather than where it is, 100xprompt searches file contents directly. It uses full regular-expression matching, so it can find flexible patterns - a function shape, an error string, a config key - not just exact words. Searches can be scoped to a directory or filtered to specific file types, and they run fast whether your project has a hundred files or a hundred thousand.
Content search is how 100xprompt answers questions like “where do we handle failed logins?” or “which files still use the old date format?” - it finds all the relevant sites at once, so nothing gets missed in a rename or refactor.

Search by file pattern

When the target is better described by its name or location - “the checkout page,” “every test file,” “the config for this service” - 100xprompt matches files by pattern instead. Glob patterns like **/*.ts or src/api/** find files by path shape, and results respect your project’s ignore rules so generated output and dependency folders never clutter the picture.

Follow references and symbols

Text search finds strings; real code navigation follows meaning. For supported languages, 100xprompt resolves symbols semantically - it goes to the actual definition of a function, finds every genuine reference to it, locates the implementations of an interface, and traces who calls what. That precision means it distinguishes a real usage from a same-named variable, a comment, or a string, which is exactly what makes cross-file changes safe.

Read files intelligently

Once 100xprompt knows where to look, it reads with judgment rather than brute force:
  • Large files are read in focused sections, so it can pull the exact region it needs without wading through everything.
  • Images and PDFs are read as visual content - useful for design mockups, screenshots, diagrams, and reference documents.
  • Jupyter notebooks are understood as cells, so data and ML work is first-class.
  • Binary files are recognized and skipped, keeping attention on what’s actually readable.

Bring in external context

Your repository is not the only source of truth. When a change depends on a library, framework, or SDK you use, 100xprompt can pull focused, up-to-date usage context for that dependency - so it applies APIs correctly instead of guessing from memory.

Capability reference

CapabilityWhat it does for youBest for
Content searchFull-regex search across file contents, scopable by folder and file typeFinding every place a behavior or pattern lives
File-pattern searchGlob matching on file names and paths, honoring ignore rulesLocating files by name, type, or location
Directory explorationSurveying structure and neighboring filesOrienting in an unfamiliar area of the repo
Go to definitionResolves a symbol to where it’s actually definedUnderstanding what a function or type really is
Find referencesLists every genuine use of a symbolSafe renames and impact analysis
Find implementationsLocates concrete implementations of an interfaceTracing polymorphic or abstract code
Call hierarchyShows callers and callees of a functionFollowing how logic flows through the system
Sectioned readingReads large files in targeted rangesWorking with huge files efficiently
Rich file readingUnderstands images, PDFs, and notebooksDesign assets, docs, and data/ML work
Library contextPulls focused usage context for dependenciesGetting external APIs right

Work in a large or monorepo project

Suppose you’ve dropped 100xprompt into a large, multi-package repository it has never seen. The winning move is to read less, not more - but read the right things. 100xprompt starts from your request, casts a targeted search, follows the threads that matter, and stops as soon as it has a confident picture of the change.
1

Anchor on the request

Your goal points to a starting search - by behavior (content) or by location (file pattern).
2

Locate the candidates

A focused search returns the handful of files that actually relate to the task.
3

Follow the connections

From those files, 100xprompt jumps to definitions, references, and callers to see how the pieces fit together across packages.
4

Read what matters

It reads the exact regions it needs - full small files, targeted sections of large ones - and confirms the surrounding patterns.
5

Act with confidence

With enough context assembled, it makes a change that fits your conventions and touches every place it should.
This is why 100xprompt can be dropped into a repository it has never seen and still make a correct, consistent change: it derives the context on demand rather than assuming it.

Tips for large codebases

“The billing webhook handler” or “the mobile login screen” gives 100xprompt a precise anchor to search from. Broad asks like “fix the backend” force it to explore more before it can act.
If you already know roughly where the change lives, say so. Scoping to a directory or a file keeps searches tight and answers fast.
For anything that spans many files - a renamed field, a deprecated call - describe the behavior or string. Content search will surface every site so the change is complete.
File and content searches respect your project’s ignore rules. Keeping generated output and vendored code ignored means searches stay focused on code you actually own.
Facts that hold across a project - where key areas live, which conventions to follow - are better captured once in memory and rules than re-explained each session.
Semantic symbol navigation depends on language support for the file in question. Where it isn’t available, 100xprompt falls back to content and pattern search - still finding the right code, just by text rather than by resolved meaning.

Code Editing

From context to change: how understanding turns into precise, multi-file edits.

Memory & Context

Skip the re-explaining: carry project knowledge across sessions.

Rules

Keep context on target: teach 100xprompt your conventions once.