cofounder-watch

Watch a live lead funnel with a zero-token dispatcher and surface salient events as short advice: replied but no scheduling link, link sent with no booking in 24h, awaiting reply. Deduped so it never

SKILL.md

GitHub

/cofounder-watch — Phase 0 ambient cofounder (ping-only)

🧒 When reporting to a non-technical operator: finish with a child-simple "In plain words" recap in their language. Not inside the cofounder advice itself. 📖 Canon: [[decision-realtime-cofounder-2026-07-02]]. This is the EYES + filter of the real-time cofounder (Phase 0). It sends nothing — it only highlights.

What it does

A deterministic signal-dispatcher (0 tokens, stdlib only): reads the live funnel tg_followups.json → classifies salient events (the same rules as /pipeline) → dedups through cofounder_ledger.json (never pings twice) → writes a digest + an HTML dashboard. Silent when there are no new important events — that is the whole point: do not nag.

How to run

python "$IMPORTS_ROOT/cofounder/cofounder_watch.py" --stdout

  • Flags: --stdout (print the digest), --reset (clear the ledger → re-alert everything).
  • Outputs: digest $IMPORTS_ROOT/cofounder/cofounder-digest.md; dashboard $OBSIDIAN_VAULT/_Dashboards/Cofounder-Watch.html (the operator works visually).

Salience (rules = /pipeline priority)

  • 🔥 HIGH — they replied, Calendly NOT sent → "send the Calendly now".
  • ⏰ MEDIUM — Calendly sent >24h ago, no booking → "booking nudge".
  • 👀 LOW — pitch sent, no reply → "check inbound / follow up / drop".
  • booked/confirmed → skipped (we do not re-pitch).

Boundaries (Phase 0)

  • Ping-only — it NEVER messages leads. Actions are executed by the operator (or via /pipeline draft→approve). Human in the loop.
  • 0 tokens — the advice is templated from the rule. The LLM persona /cofounder for nuance is Phase 0.5 (later, HIGH events only).
  • Empty ≠ broken: "silent" is a valid result.

Next (Phase 0.5 / 1+, on the operator's greenlight)

  • Delivery into a Telegram signal chat (once the MCP is alive) instead of a file + dashboard only.
  • Scheduling on the HUB (always-on HUB-1) 2-3x/day — this is background work, so it belongs on the hub, not the laptop, via /schedule on the hub or the machine bus. Being lightweight (0 tokens), daytime runs are fine.
  • Phase 0.5: HIGH events → the LLM persona /cofounder (top model) for nuanced advice.
  • Phase 1: + urgent VC email + calendar conflicts (same mailbox + ledger).
  • To extend it, add a source to the SAME dispatcher; do NOT breed watchers ([[telegram-eventloop-listener]]: one live client, the AUTH_KEY pitfall).
<!--kit-footer-->

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.