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".
The English summary is being prepared. Showing the original description.
See how to installReview the instructions the agent will receive before you install.
The operator's pain: the session died mid-work (a crash / the computer went off) → the new session is empty. Before: remember by hand, then separately fire /arch, /sync-check, /mcp, then work out "where did we stop". Now it is one word: /1 (or /!).
What this is NOT: /retro is for a CLEAN end of session (inventory of what was built → routing to homes → /compact). After a sudden crash there is nothing left to package, the context is already gone → you need resurrection, NOT /retro. Two different tools.
The TurnState black box writes EVERY turn into SQLite (a Stop hook), and it survives any crash. Read the last turns:
python "$IMPORTS_ROOT/turnstate/turnstate_show.py" --n 12
From the output pull out: what the operator asked for last · which files were touched · which DECISIONS were made · the next step. That is "where we were cut off". (Flags: --stats, --session <id>, --n N.)
⚠️ Fallback if the box is EMPTY (turnstate_show.py --stats shows turns: 0 — it happens when the Stop hook hasn't fired yet after a Claude Code restart; see [[crash-recovery-command]]): do NOT say "there is no data". Derive "where we were" from the LAST session — its full seed is already assembled in step 3 (_Dashboards/sessions-md/_continue/<cli>.seed.md): read the TAIL of that file (the last 1-2 human→assistant exchanges) and pull the task plus the last step out of them. So with an empty box, step 1 leans on the result of step 3.
Three fast deterministic checks:
python "$IMPORTS_ROOT/arch/arch_status.py"
powershell -NoProfile -ExecutionPolicy Bypass -File "$IMPORTS_ROOT/sync_check\sync_check.ps1"
MCP — IN-SESSION ONLY (⚠️ NOT claude mcp list and NOT a second Telethon client → AUTH_KEY_DUPLICATED logs the account out; memory [[mcp-health-check]]): fire one cheap read call per live server — Telegram mcp__telegram__get_me (expect the work account), WhatsApp mcp__whatsapp__get_my_profile, n8n mcp__n8n__n8n_health_check. If a server's tools are missing from the session, it never loaded (diagnostics live in /mcp).
Assemble this machine's previous human session into a seed → the clipboard (the /resume-last engine):
python "$IMPORTS_ROOT/claude_sessions/continue_session.py" --last
Tell the operator: "Open a New session and hit Ctrl+V — you get the whole conversation back." If it prints clipboard: FAILED, the seed is on disk at _Dashboards/sessions-md/_continue/<cli>.seed.md.
🔄 Resurrection
📍 Where we were: <1-2 lines from TurnState — the task, the last file/decision, the next step>
💚 System: arch <✅/⚠️/🔴> · sync <✅/⚠️/🔴> · mcp <tg ✅ / wa ✅ / n8n ✅>
📋 The previous session's full history is in the clipboard (New session → Ctrl+V).
➤ Continue from: <the next step>?
Then a 🧒 "In plain words" line (memory [[eli5-always]]).
/arch / /sync-check / /mcp separately.1, so /1 works natively; /! is an alias trigger (a special character can't be a folder name): when you see /!, run this same skill.python3; with a non-standard vault path — env CLAUDE_VAULT_ROOT=<...>.Memory [[crash-recovery-command]]. Building blocks: [[turnstate-ledger]] (the black box), [[claude-desktop-sessions-per-account]] (continue_session), [[system-architect]] (/arch), syncthing-desktop-laptop-sync (/sync-check), [[mcp-health-check]] (/mcp). Paired with the SessionStart hook session_resume_hook (that one shows the previous session at startup on its own — this one gathers recall + health + full history on command).
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.
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
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.
The English summary is being prepared. Showing the original description.
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".
The English summary is being prepared. Showing the original description.
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.
The English summary is being prepared. Showing the original description.
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".
The English summary is being prepared. Showing the original description.
- 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“
The English summary is being prepared. Showing the original description.
LLM-judge stage of an alpha-mining pipeline: take a miner's deterministic candidate list, keep only real signal, drop noise, and propose additions to that miner's curated note. Use after any zero-token detector run. Reads only the candidate digest and the home note, never the corpus. Triggers: "/alpha-judge <miner>", "judge the candidates".
The English summary is being prepared. Showing the original description.