Skip to content
Skill

llm-pipeline

by majiayu000

AI Summary

Pulse Radar uses LLM pipeline to transform raw Telegram messages into structured knowledge. Core philosophy: Messages individually are noise; batched extraction reveals patterns. 1. JSON-only output — explicitly state "respond with ONLY JSON"

Install

Copy this and paste it into Claude Code, Cursor, or any AI assistant:

I want to install the "llm-pipeline" skill in my project.

Please run this command in my terminal:
# Install skill into your project (2 files)
mkdir -p .claude/skills/llm-pipeline && curl --retry 3 --retry-delay 2 --retry-all-errors -o .claude/skills/llm-pipeline/SKILL.md "https://raw.githubusercontent.com/majiayu000/claude-skill-registry/main/skills/ai-llm/llm-pipeline/SKILL.md" && curl --retry 3 --retry-delay 2 --retry-all-errors -o .claude/skills/llm-pipeline/metadata.json "https://raw.githubusercontent.com/majiayu000/claude-skill-registry/main/skills/ai-llm/llm-pipeline/metadata.json"

Then restart Claude Code (or reload the window in Cursor) so the skill is picked up.

Description

Pydantic-AI agents, RAG, embeddings for Pulse Radar knowledge extraction.

LLM Pipeline Skill

<overview> Pulse Radar uses LLM pipeline to transform raw Telegram messages into structured knowledge. Core philosophy: Messages individually are noise; batched extraction reveals patterns. </overview> <entity-hierarchy> ` TOPICS (categories) └─ ATOMS (knowledge units: problem/solution/decision/insight...) └─ MESSAGES (raw data, hidden layer) ` </entity-hierarchy> <extraction-flow> `python

1. Message arrives via Telegram webhook

await save_telegram_message(message) # triggers TaskIQ

2. Scoring (AI Judge, not heuristics - ADR-003)

score = await importance_scorer.score(message)

3. Auto-trigger extraction when threshold met

if unprocessed_count >= 10: # ai_config.message_threshold await extract_knowledge_from_messages_task.kiq()

Discussion

0/2000
Loading comments...

Health Signals

MaintenanceCommitted Today
Active
Adoption100+ stars on GitHub
428 ★ · Growing
DocsMissing or thin
Undocumented

GitHub Signals

Stars428
Forks72
Issues0
UpdatedToday
View on GitHub
MIT License

My Fox Den

Community Rating

Sign in to rate this booster

Works With

Claude Code