Agent-Skills.md

Agent Skills: glhf — Conversation History Search

Search past Claude Code sessions to recall prior solutions, commands, fixes, and decisions. Use when the user references earlier work ("how did I fix", "last time we", "pick up where we left off", "a few months ago", "what was that … I set up", "didn't I once").

UncategorizedID: trevors/dot-claude/searching-history

Repository

trevors/dot-claude
TrevorS
10

Install this agent skill to your local

pnpm dlx add-skill https://github.com/TrevorS/dot-claude/tree/HEAD/skills/searching-history

Skill Files

Browse the full folder contents for searching-history.

Download Skill

Loading file tree…

skills/searching-history/SKILL.md

Skill Metadata

Name
searching-history
Description
Search past Claude Code sessions to recall prior solutions, commands, fixes, and decisions. Use when the user references earlier work ("how did I fix", "last time we", "pick up where we left off", "a few months ago", "what was that … I set up", "didn't I once").

glhf — Conversation History Search

Hybrid search (text + semantic), so natural-language queries work.

Full command list, every flag, and per-task patterns: REFERENCE.md.

How to run it

Derive the search query yourself, then run glhf search.

  • Explicit /searching-history <query> → use $ARGUMENTS as the query: glhf search "$ARGUMENTS" --compact
  • Auto-triggered from the conversation → $ARGUMENTS is empty. Derive concise search terms from what the user is trying to recall (the topic, error text, command, or project they referenced) and run the search with those.

Core usage

glhf search "authentication" --compact              # find past solutions
glhf search "docker" -t Bash --compact              # find commands you've run
glhf search "bug" -p myapp --since 1w --compact     # scope by project and time
glhf search "failed" --errors --compact             # only errors
glhf recent -l 10                                   # browse recent sessions
glhf session abc123 --summary                       # overview, then dive deeper

Always pass --compact — it significantly reduces output tokens. Chain search → session ID → glhf session <id> --summary when you need fuller context than the hits give.