메뉴
Search X (Twitter) archives for insights, conversations, and knowledge. Use when mining tweets, DMs, Grok chats, likes, or extracting information from X data exports.
기본적으로 소스를 먼저 확인하는 Prompt가 선택됩니다. 직접 명령으로 전환하거나 로컬 사본을 다운로드할 수도 있습니다.
설치 여부를 결정하기 전에 SKILL.md와 SkillsMP에 표시된 보조 파일을 읽어 보세요.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
직접 명령은 검토 Prompt를 거치지 않습니다. 실행하기 전에 소스를 확인하세요.
npx skills add https://github.com/oysteinkrog/dotfiles --skill xf명령은 한 줄로 유지됩니다. 복사하기 전에 가로로 스크롤해 전체 내용을 확인하세요.
로컬 사본을 원하시나요? SkillsMP에서 현재 제공할 수 있는 파일을 다운로드하세요.
Consult Fable (primary oracle) for expert second opinions; escalate to GPT-5.6 Sol (via Codex CLI) or GPT-5.5-Pro (via PAL) only for extremely important or complex tasks (always paired with Fable). Use for complex decisions, architecture choices, debugging hard problems, or when user says "consult oracles", "ask the experts", or wants a second opinion.
Run 2x oracle sessions (FOR + AGAINST stances) to validate design decisions, plans, or bead readiness. Default = two Fable subagents; escalate to GPT-5.6 Sol via Codex CLI or PAL 2x GPT-Pro (always paired with Fable) for extremely important or complex validations. Use after design rounds, before implementation, or to challenge architecture decisions.
Use the local Codex CLI as an independent second agent. Two branches — (1) proactively run `codex review` for a second opinion after completing a substantive change, before presenting it as done or committing; (2) delegate a well-defined implementation task via `codex exec`, ONLY when the user explicitly asks for Codex to do it. Also covers how to prompt Codex.
SKILL.md 표시 중
SOC 직업 분류 기준
| name | xf |
| description | Search X (Twitter) archives for insights, conversations, and knowledge. Use when mining tweets, DMs, Grok chats, likes, or extracting information from X data exports. |
Core Insight: Your X archive is a goldmine. Sub-millisecond search makes iterative exploration practical. Mine your history before rebuilding context.
Your X archive contains:
The insight: Mining your X history surfaces forgotten context faster than rebuilding it.
1. Bootstrap: Check health, get archive overview
xf doctor && xf stats --format json | jq '{tweets, likes, dms, grok_messages}'
2. Search: Find content by topic (hybrid mode default = best)
xf search "KEYWORD" --format json --limit 50
3. Filter by type: Narrow to specific content
xf search "KEYWORD" --types dm --context --format json # DMs w/ conversation
xf search "KEYWORD" --types grok --format json # AI Q&A
4. Follow hits: Get full context
xf tweet <ID> --thread --format json # Tweet thread
xf search "KEYWORD" --types dm --context --format json # Full DM thread
5. Extract: Export for analysis
xf search "topic" --format json --limit 1000 | jq '[.[] | {text, date: .created_at}]'
--format json — Machine-readable output for piping to jq--context for DMs — Shows full conversation threads around matches# Health + overview (ALWAYS first)
xf doctor && xf stats --format json
# Search modes (hybrid is default = best)
xf search "query" --format json # Hybrid (keyword + semantic)
xf search "exact phrase" --mode lexical # Keyword only (BM25)
xf search "feeling stressed" --mode semantic # Meaning-based
# Filter by content type
xf search "query" --types tweet --format json # Your tweets
xf search "query" --types dm --context # DMs with full conversation
xf search "query" --types like --format json # Liked tweets
xf search "query" --types grok --format json # Grok AI chats
# Date filtering
xf search "query" --since "2024-01" --until "2024-06" --format json
xf search "query" --since "last week" --format json
# Output control
xf search "query" --limit 100 --offset 50 --format json
| You Want | Use | Why |
|---|---|---|
| Find content by topic | --mode hybrid (default) | Best of keyword + semantic |
| Find exact phrase | --mode lexical + "quotes" | Literal string match |
| Find conceptually related | --mode semantic | "feeling overwhelmed" → burnout tweets |
| DM conversations | --types dm --context | Shows full thread around matches |
| What you engaged with | --types like | Your curation over time |
| AI Q&A history | --types grok | What you asked Grok |
| High engagement content | --sort engagement | Your best tweets |
| Rule | Why | Consequence |
|---|---|---|
--format json | Machine-readable output | Required for jq pipelines |
| Index first | Archive must be indexed | xf index before search |
--context requires --types dm | Only works for DMs | Ignored otherwise |
| All data local | Zero network calls | Privacy-first, offline-capable |
| Signal | Meaning | Action |
|---|---|---|
| Many likes on topic | Topic resonated with you | Search likes for curated content |
| DM mentions project | Detailed private discussion | Use --context to get full thread |
| Grok chat on topic | You asked AI about it | May have refined explanations |
| High engagement tweet | Your audience resonated | Check thread for discussion |
| Semantic finds more | Wording varies | Use semantic for broad exploration |
# Extract just text
| jq '.[].text'
# Text with dates
| jq '.[] | {text, date: .created_at}'
# Filter by engagement
| jq '[.[] | select(.favorite_count > 10)]'
# Count by year
| jq -r '.[].created_at[:4]' | sort | uniq -c
# DM conversation IDs
| jq '[.[].metadata.conversation_id] | unique'
# Find hashtags
| jq -r '.[].text' | grep -oE '#\w+' | sort | uniq -c | sort -rn
| Need | Reference |
|---|---|
| Full command reference | COMMANDS.md |
| Workflow recipes | RECIPES.md |
| jq patterns | PATTERNS.md |
| Pitfalls & fixes | PITFALLS.md |
| X archive data structures | DATA_STRUCTURES.md |
| Script | Usage |
|---|---|
./scripts/quick_analysis.sh | One-command archive overview |
./scripts/topic_miner.py "topic" | Deep dive on a topic |
./scripts/validate.sh | Validate xf is working |
# Quick health check
xf doctor
# Should show: all checks pass
If issues: run xf index --force to rebuild.