Mine any Telegram channel or chat for signal in one command: incremental zero-token scrape, deterministic detector shortlist, LLM judge for what is genuinely valuable to this owner, then linked vault notes, a database row, a map-of-content entry and reindex. Triggers: "/mine-channel <@channel|id>", "alpha from <channel>".
インストールする前に、エージェントに与えられる指示の中身を確認できます。
🧒 Finish the reply to a non-technical operator with a simple "In plain words" recap (memory
eli5-always).
This generalizes a pattern that was repeated by hand 5+ times across different channels. Cost ladder [[vault-data-architecture]]: a deterministic detector (0 tokens) → an LLM judge over the shortlist only → the vault.
0. RECALL (do not duplicate). Has this channel been mined already? Check $IMPORTS_ROOT/alpha/<slug>\ and memory (*-mine, *-community-import). If yes, this is a TOP-UP: same slug, incremental.
1. Resolve the channel. A username (@some_channel / some_channel) or a numeric id (<YOUR_CHAT_ID>). If you do not know the id — mcp__telegram__search_dialogs or /chat <name>. Pick a short Latin slug (e.g. silmeshok).
2. Scrape + detector (0 tokens, 0 GPU):
set PYTHONIOENCODING=utf-8
python $IMPORTS_ROOT/alpha/mine_channel.py --channel <name|id> --slug <slug> [--limit N] [--top 40] [--since YYYY-MM-DD --until YYYY-MM-DD]
→ _imports\alpha\<slug>\<slug>.jsonl + <slug>.db + the shortlist _imports\alpha\candidates\<slug>-report.md.
(The scrape goes through the subscription Telethon session <TELEGRAM_MCP_DIR>/.env, account work_acct_a — NOT a paid API [[prefer-included-limits-before-paid-api]]. Re-running without --channel plus --detect-only = re-detect without re-scraping.)
3. The judge (LLM, shortlist only). Run the shortlist through a cheap grunt model (grunt work → the small model [[model-routing-sonnet-grunt]]; a subagent with model:'sonnet', or the alpha-judge skill): keep only REAL alpha FOR THE OWNER (a tool/model/deal · a technique or workflow · a mental model · a proof point or benchmark · a build pattern), and drop promos, banter, and anything we already do better. Verdicts: ✅ alpha · 🟡 watch · 🗑 noise.
4. Into the vault (obsidian-ingest). Backup first [[vault-backup-rule]]. ✅ alpha → atomic notes + concept links ([[no-orphan-notes-rule]]) + a MOC _<Slug>-MOC; a sensitive channel → #private. The original jsonl stays in _imports\alpha\<slug>\ (that IS the _originals copy for the channel). Reindex [[reindex-routine]] (or let the nightly job pick it up).
5. Make it a routine? ([[evaluate-recurring-into-routine]]) A valuable channel → propose a weekly top-up (night window [[routines-run-at-night]]).
The detector is generic (AI / tool / deal / startup keywords); per-channel keyword and promo filters are tuned inside mine_channel.py. Sensitivity: closed / adult / personal channels → #private, never surfaced outside. Read the channel only, never post into it.
Like this skill? It is one of 100 in second-brain-starter-kit: the second brain we built for ourselves and run every day at Palo Alto AI Research Lab. Install the whole set with npx skills add tonydzi/second-brain-starter-kit. Everything is open source and free, so take what you need.
Flagships worth a look on their own: secondop-panel (a second opinion from a panel of external models), claude-memory-tidy (stop your agent's memory from rotting), telegram-mcp-kit (your own Telegram over MCP in about 15 minutes).
Author: Anton Dziatkovskii, Palo Alto AI Research Lab. Telegram @tonydzi - WhatsApp +1 341 222 9178 - X @Tony_Stef_
Engineers: want to test-drive this setup? Message me. I hand out free starter seeds to engineers who test and report back, and custom skill requests are welcome.
まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Run an autonomous consensus round between agent peers on several machines: they negotiate a decision over a shared channel and execute it without a human relaying messages. Use when multiple machines must agree on a change. Triggers: "/03", "work it out among yourselves", "find consensus". Wakes the human only for irreversible actions or a deadlock.
日本語の概要は準備中です。原文の説明を表示しています。
Recover a session after a hard crash (app died, machine rebooted, context lost mid-work): replay the turn-state ledger for the last requests, files and decisions, run a green/red health ping over the system map, file sync and MCP connectors, then rebuild the previous session history for pickup. Triggers: "/1", "resume after crash", "wake".
日本語の概要は準備中です。原文の説明を表示しています。
Show the operator's day: today's calendar plus what needs them personally (leads due, people awaiting a reply, deadlines), pulled from Calendar and Telegram and ranked. Read-only: sends and commits nothing. Triggers: "/agenda", "what's on today", "my agenda".
日本語の概要は準備中です。原文の説明を表示しています。
Rewrite machine-written text so it reads like a busy human wrote it: cut 50-70%, break the even rhythm, drop AI-tell words and filler, keep facts and protected terms. Use before publishing any post, reply or email drafted by a model. Triggers: /ai-slop, /slop, humanize, anti-slop, make it shorter and alive.
日本語の概要は準備中です。原文の説明を表示しています。
Decision protocol for strategic work: recall the knowledge vault, run a gap analysis, emit a deep-research prompt for external tools, then synthesize the returned reports into a decision memo. Use before any new product, feature, market, pricing, GTM or architecture call, where local recall alone is not enough. Triggers: "/alfa-search-recall-deepresearch", "alpha protocol", "R+DR".
日本語の概要は準備中です。原文の説明を表示しています。
- КРЕДИТ АВТОРУ ИДЕИ: совет из комментария (TG @ClawRus / @ClawInga, комменты под постами Антона в FB, треды и issue на… Триггеры: “/alpha-credit“, “/credit“, “кто нам это посоветовал“, “запиши совет“, “поблагодари советчиков“, “кому мы должны спасибо“, “кредит автору“, “чей это был совет“, “credit the author“
日本語の概要は準備中です。原文の説明を表示しています。