Why does this skill stop me from reading files?

It intercepts operations that would waste tokens or provide less relevant context than a ranked semantic search, suggesting a more efficient tool for the task.

Does it require claudemem to be installed?

Yes, it is designed to work with claudemem to perform semantic searches. It will even prompt you to index your project if it isn't already.

How much token usage can this save?

Reading five full files might cost 5,000+ tokens, whereas a targeted semantic search usually costs around 500 tokens while providing more relevant code chunks.

What specifically triggers the Search Interceptor?

The skill triggers when attempting to read 3 or more files at once, using broad glob patterns like **/*.ts, or planning sequential reads to 'understand' a concept.

Will it prevent me from reading a single specific file?

No. The interception decision matrix allows for 1-2 specific file reads or exact string matches via grep without redirecting you.

Semantic Search Interceptor

Name: Semantic Search Interceptor
Author: involvex

byinvolvex

0•

Productivity & Workflow

Optimizes code exploration by intercepting bulk file operations and redirecting to efficient semantic search queries.

The Semantic Search Interceptor acts as an intelligent guardrail for Claude, preventing inefficient token usage and sequential file reading habits. By automatically triggering before bulk file reads or broad glob patterns, it evaluates if a semantic search via claudemem would provide better, more ranked context. This skill helps developers overcome tool familiarity bias, ensuring that the most relevant code sections are analyzed first while significantly reducing the overhead associated with reading entire directories or large files.

Key Features

01Real-time validation of claudemem index status

02Automatic interception of bulk file read operations (3+ files)

03Intelligent redirection from sequential reads to ranked semantic queries

040 GitHub stars

05Prevention of broad glob pattern execution to minimize token waste

06Decision matrix for distinguishing between exact matches and semantic concepts

Use Cases

01Investigating complex modules like authentication without reading every individual file in the directory

02Discovering architectural patterns across a large codebase without performing expensive broad glob patterns

03Auditing specific API integrations when multiple related paths are mentioned in a prompt

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add involvex/involvex-claude-marketplace search-interceptor

For use in Claude.ai and ChatGPT

Download Skill