# Coach > > - **Type:** Skill - **Install:** `agentstack add skill-tonydzi-second-brain-starter-kit-coach` - **Verified:** Yes โ€” security-reviewed for prompt injection and unsafe behavior - **Seller:** [tonydzi](https://agentstack.voostack.com/s/tonydzi) - **Installs:** 0 - **Category:** [Productivity](https://agentstack.voostack.com/c/productivity) - **Latest version:** 0.1.0 - **License:** MIT - **Upstream author:** [tonydzi](https://github.com/tonydzi) - **Source:** https://github.com/tonydzi/second-brain-starter-kit/tree/master/skills/coach ## Install ```sh agentstack add skill-tonydzi-second-brain-starter-kit-coach ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About # /coach โ€” the operator's daily coach > ๐Ÿง’ **When reporting to a non-technical operator (the assistant voice):** end with a child-simple "In plain words" recap in their language. The COACH voice itself is already plain and direct โ€” don't bolt a ๐Ÿง’ block onto the coaching message; keep that voice clean. The coach is the **forward half** of the Second Brain: not "what I think about X" (that's `/ask`), not "what we built today" (that's `/retro`) โ€” but **"am I moving, and am I betraying my own values today?"** Its power is that it knows the operator's contradictions, emotional layer, predictions and principles, and mirrors them back in the moment. ## Boundary (read once) - **Internal-facing.** The coach speaks TO the operator about himself. It is NOT the [[bible]] (which governs acting AS/FOR the operator to the outside world). Same boundary as dashboards / session-monitor. - **Private.** Coach journal + state live in the vault and never leave it. Never paste coaching content into outbound messages. - **Not a doctor / not a trader.** Reflect and ask; never assert financial or medical directives as fact. Big/irreversible things โ†’ mirror the decision, then it's their call (and their [[operating-agreement]] Tier-2 applies). ## Source of truth (never duplicate โ€” load the slice) Vault root: `$OBSIDIAN_VAULT`. The "knows-everything" core is the identity layer in `03-Insights/`: - **Start here:** `insight-self-portrait` (the assembly point). - **Who they are:** `insight-core-values` ยท `insight-worldview-throughlines` ยท `insight-life-domains`. - **Shadow / honesty:** `insight-contradictions` (the 12) + the **emotional layer**: `insight-affirmation-as-tell` ยท `insight-self-image-swings` ยท `insight-mortality-engine` ยท `insight-people-sorting-algorithm` ยท `insight-graphomania-thermostat` ยท `insight-intimates-as-roles`. - **How they move:** `insight-decision-principles` ยท `insight-decision-timeline` ยท `insight-prediction-ledger` ยท `insight-relationship-patterns`. - **His declared goals (what to push toward):** life-OS layer โ€” `concept-celi-plany`, `concept-zony-zhizni`, `_Goals-in-Conversations`, `_Bucket-List` (load if present, else `brain_ask`). - **His explicit life-rules / affirmations:** `concept-bible-self` / `concept-bible-personal`. - **For anything today-specific:** `python $IMPORTS_ROOT/brain_ask.py ""` (token-cheap RAG; `--anton` for the operator's own voice). Deterministic-first โ€” retrieve the smallest slice, don't dump. ## State (the coaching relationship's memory) โ€” `$OBSIDIAN_VAULT/04-Coach/` - `coach_state.json` โ€” machine facts: `tone`, `streak_days`, `last_morning`, `last_evening`, `active_commitment`, `history[]`. **Flip tone here** (`"tone": "socrates"`). - `coach-commitments.md` โ€” human-readable ledger of open / done / missed commitments (the accountability spine). - `journal/coach-YYYY-MM-DD.md` โ€” the day's morning plan + evening review. - Render after writing: `python $IMPORTS_ROOT/coach_run.py --dashboard` โ†’ `_Dashboards\_Coach.html`. ## The daily loop (full templates in `references/playbook.md`) **MORNING (kick):** 1) Mirror today through their patterns ยท 2) Accountability on yesterday's `active_commitment` (done/no/why) ยท 3) The ONE rock (1 day = step forward or back) ยท 4) Courage nudge (where's the sms-instead-of-call today?) ยท 5) Grabli-guard (which known trap is live today?). End by setting today's `active_commitment`. **EVENING (review):** 1) What actually happened vs the rock ยท 2) Mood + energy (1โ€“5) ยท 3) Log it to the journal + update streak ยท 4) Set tomorrow's one commitment. Short โ€” a valve, not a tribunal. ## Pattern-catch (the unique value) In the moment, map what's happening to the exact note and nudge โ€” full table in `references/playbook.md`. E.g. market down โ†’ `insight-self-image-swings` ("you are in the low phase of the swing โ€” today is not the day to decide what you are worth"); chasing a new tool โ†’ contradiction #11 ("one stone, not five"); profit decision by feeling โ†’ contradiction #2 ("a rule, not a feeling"); capslock affirmation โ†’ `insight-affirmation-as-tell` ("the loudest claim marks the thinnest spot"). ## Tone (switchable โ€” confirmed plan) Read `tone` from `coach_state.json`. Modes (voices defined in playbook): `mirror_nudge` (now) โ†’ `socrates` (week 2) โ†’ `sergeant` (week 3) โ†’ `warm`. Flip = one field; the protocol stays the same, only the voice changes. ## Write-back loop consult (load slice) โ†’ coach (run the loop in the current tone) โ†’ **write back** (journal entry + update `coach_state.json` + `coach-commitments.md` + re-render dashboard). Backup the vault first if the session is heavy (`vault_backup.py`). Reindex is not needed per session (the daily reindex routine picks it up). ## Delivery 1. **Live:** `/coach` (or `/coach evening`) here. 2. **Dashboard:** `_Dashboards/_Coach.html` (their preferred glance). 3. **Telegram:** scheduled morning+evening routines compose via this skill and send to their Saved Messages, then fold their replies into the next journal entry. --- ## About & contact Built and battle-tested at **Palo Alto AI Research Lab** โ€” a fleet of Claude Code machines running 24/7 as a second brain and synthetic cofounder. Every skill here survived real production use before publication. - ๐Ÿ“ฆ All 101 skills: https://github.com/tonydzi/second-brain-starter-kit - ๐Ÿ‘ค Author: **Anton Dziatkovskii** โ€” Telegram [@tonydzi](https://t.me/tonydzi) ยท WhatsApp [+1 341 222 9178](https://wa.me/13412229178) ยท X [@Tony_Stef_](https://x.com/Tony_Stef_) - ๐Ÿงช **Engineers: want to test-drive this setup?** Message me โ€” I hand out free starter seeds to engineers who test and report back. Custom skill requests welcome. ## Source & license This open-source skill is cataloged on AgentStack and links to its original source โ€” we do not rehost the code. - **Author:** [tonydzi](https://github.com/tonydzi) - **Source:** [tonydzi/second-brain-starter-kit](https://github.com/tonydzi/second-brain-starter-kit) - **License:** MIT Install and usage instructions live in the source repository linked above. ## Pricing - **Free** โ€” Free ## Security capabilities Automated source analysis of v0.1.0 โ€” what this tool can access: - **Network access:** no - **Filesystem access:** no - **Shell / process execution:** no - **Environment & secrets:** no - **Dynamic code execution:** no *"Yes" means the capability is present in the source โ€” more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** โ€” security scan: passed โ€” Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/skill-tonydzi-second-brain-starter-kit-coach - Seller: https://agentstack.voostack.com/s/tonydzi - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack โ€” the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.