skills.rest
mileshill
mileshillCOMMUNITY
Explore all skills by this author →
GitHub

recall

Index past Claude Code session transcripts and retrieve relevant decisions with BM25 ranking.

3|1|Updated Feb 16, 2026
Search Tags:#context#recall#embeddings#transcripts#bm25#proactive recall
One-click install
npx skills add https://github.com/mileshill/claude-recall --skill recall-mileshill
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: recall
Source: https://github.com/mileshill/claude-recall/tree/main
Command: npx skills add https://github.com/mileshill/claude-recall --skill recall-mileshill

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires rank-bm25, numpy, sentence-transformers, and includes scripts (resource) components.

What problem does it solve?

The recall system enables you to surface relevant decisions, designs, and discussions from past Claude Code sessions by indexing transcripts and providing fast retrieval, ensuring continuity across projects.

Core Features & Use Cases

  • BM25-based keyword search with temporal decay for fast, relevant results
  • Optional semantic search using embeddings for concept-level recall
  • Proactive recall: Claude can automatically surface past context when topics are mentioned
  • Safe redaction of secrets in transcripts to protect sensitive information
  • Seamless integration via SessionStart and SessionEnd hooks and a command-line /recall skill
  • Manual and programmatic recall through the /recall skill and scripts

Quick Start

Run the recall skill with a query to surface past decisions and contextual notes from past sessions.

Frequently Asked Questions about recall

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I recall past context and decisions from previous Claude Code sessions?▼

You can recall past context by indexing past Claude Code session transcripts to surface relevant decisions, bug fixes, and design choices using BM25 ranking with temporal decay.

Can I use semantic embeddings for concept-level search instead of keyword matching?▼

Yes, semantic search using embeddings is supported optionally for concept-level recall, with an automatic fallback to BM25 keyword ranking if embeddings are not utilized.

How does proactive recall work when mentioning topics in a new session?▼

Proactive recall automatically surfaces past context and relevant decisions when specific topics are mentioned, enabled seamlessly through SessionStart and SessionEnd hooks integration.

Is there a way to protect sensitive information like secrets in session transcripts?▼

Yes, the system applies safe redaction of secrets in indexed transcripts to protect sensitive information during the context retrieval process across projects.

Do I need to install specific dependencies to use the recall skill?▼

Yes, the required dependencies include rank-bm25, numpy, and sentence-transformers to enable BM25 ranking, numerical operations, and optional semantic embeddings.

What is the best way to find past authentication decisions across different projects?▼

The best way is to use the /recall skill or scripts to query indexed transcripts, which returns top results with citations for authentication decisions and design choices.

Related Skills

Explore More Software Engineering Skills→
parcadeiparcadei
3.9k

hook-developer

Document Claude Code hooks with schemas, registration, and testing patterns.

Community
Advanced
thimsluggathimslugga

memory-search

Search past conversations and decisions to surface relevant context.

Community
Intermediate
jayll1303jayll1303
17

text-embeddings-rag

Generate local embeddings and build retrieval-augmented generation pipelines with sentence-transformers and FAISS.

Community
Advanced
cdeistopenedcdeistopened
2

local-markdown-search

Index and search local Markdown collections with BM25 and vector embeddings.

Community
Intermediate