Claude Skill

crm-sync

Sync CRM knowledge notes in a vault with the live CRM code repositories: pull the read-only repos, detect which changed, and refresh only the affected reverse-engineering notes and the keep/drop/decide note. Triggers: "/crm-sync", "refresh CRM knowledge", "what changed in the CRM

LLM Mart · 0 points · 1 views 0 listing impressions 0 install-command copies
Download tonydzi-second-brain-starter-kit-skills_crm-sync-7bc2efa.zip · 3 KB

Install

skills CLI npx skills add https://github.com/tonydzi/second-brain-starter-kit/tree/master/skills/crm-sync
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install tonydzi-second-brain-starter-kit@llmmart
Git git clone https://github.com/tonydzi/second-brain-starter-kit.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole tonydzi/second-brain-starter-kit collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

/crm-sync - keep the CRM knowledge layer fresh

The CRM (CHARM) is Anton's proven engine on GitLab. We reverse-engineered it once (2026-06-19) into a linked vault layer so his digital twin can /ask how the CRM works. Code keeps changing; this skill stops that knowledge from going stale silently. It is the recurring half of the standing rule evaluate-recurring-into-routine.

What this is NOT: not a re-decompose-from-scratch. It is a DIFF-DRIVEN refresh: pull, see what changed, update only the notes that map to the changed code.

Where everything lives

  • Code (read-only clone, 16 repos): <CRM_REPOS_ROOT> (memory crm-gitlab).
  • Decomposition audit (HTML, source of the notes): <CRM_REPOS_ROOT>\_AUDIT\ (Functional-Map.html, Know-How-Decomposition.html, Architecture-Deep.html).
  • Vault knowledge layer: $OBSIDIAN_VAULT/05-Resources/CRM-Engine/
    • _CHARM-CRM-Engine-MOC.md (hub)
    • crm-architecture-4-layers.md
    • crm-mtproto-engine.md
    • crm-core-autopilot.md
    • crm-admin-panel-api-brain.md
    • crm-data-model-mongo.md
    • crm-know-how-13-patterns.md
    • crm-landing-api-money.md
    • decision: $OBSIDIAN_VAULT/02-Decisions/decision-crm-keep-cc-drop-decide.md

Repo -> note map (which note to refresh when which code changes)

Code area in <CRM_REPOS_ROOT> Note(s) to refresh
admin-panel / admin-panel-api (FastAPI brain, filter-DSL, $facet) crm-admin-panel-api-brain
mtproto-api (Telethon hands, event-pool, dialog export, FloodWait) crm-mtproto-engine
core (aiocron autopilot: newsletters, gatekeeper, tapping, calls) crm-core-autopilot
landing-api (token sale, NOWPayments, CHEATING_MODE, webhook) crm-landing-api-money
shared models / Mongo schema (collections, fields) crm-data-model-mongo
cross-cutting architecture / generations crm-architecture-4-layers
reusable patterns / know-hows crm-know-how-13-patterns
any capability added/removed/disabled decision-crm-keep-cc-drop-decide (+ MOC)

Procedure

  1. RECALL first (don't duplicate). Read memory crm-gitlab and _CHARM-CRM-Engine-MOC.md so you refresh, not rewrite. Note the date of the last sync.
  2. Pull the repos. For each repo dir under <CRM_REPOS_ROOT>: git -C "<CRM_REPOS_ROOT>\<repo>" pull --ff-only (clone is read-only; this only updates). Collect the set of repos that actually moved (non-empty pull, or git -C <repo> log --oneline <old>..<new>). If a pull needs GitLab creds and fails, stop and tell Anton (creds in secrets\; do NOT hardcode).
  3. No changes -> stop. If nothing moved since last sync, report "CRM unchanged since
  4. For each changed repo, re-read ONLY the changed files/areas (use an Explore subagent with model: 'sonnet' - this is grunt extraction, per model-routing-sonnet-grunt; escalate to Opus only if a judgement call). Ask it to report: what behavior changed vs what the mapped note currently says, with file+line evidence.
  5. Backup before writing. python "$IMPORTS_ROOT/vault_backup.py" (rule vault-backup-rule). NEVER blanket-delete; edit in place.
  6. Update the mapped note(s) - the "How it works" / "Weak spots" sections only. PRESERVE frontmatter, origin: external, authored_by: (denis-udot / gleb-taigunov), and all [[wikilinks]]. If a capability moved KEEP<->DROP<->DECIDE, update the decision note too. New behavioral pattern worth a concept? Follow concept-creation-rules (create if >=3 repeats + entity noun + domain), and relink (no-orphan-notes-rule).
  7. Verify + reindex. Run the vault broken-link check; then python "$IMPORTS_ROOT/brain_embed_update.py" (cooldown 15min; nightly @04:00 will catch it otherwise) so the refreshed notes are /ask-searchable.
  8. Report post-hoc (not for permission): which repos moved, which notes changed, any new concept, any DECIDE item that flipped, and what still needs Anton/Denis.

Guardrails

  • Read-only by default; the only writes are git pull (updates the clone) + the mapped vault notes, always backup-first.
  • Don't touch the live CRM server or its Mongo - that is the prod-access track, separate.
  • AK-47: this is one markdown procedure. If the CRM grew a whole new service, that is a new note + MOC link, still just notes - flag it, don't over-build.

When to run

On demand (after Anton hears the CRM changed), or as a light routine. Pairs with the crm-gitlab DD-audit and the decision-crm-keep-cc-drop-decide open items.


Like this skill? It is part of 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.

Files (second-brain-starter-kit)
  • SKILL.md 5.9 KB
    ---
    name: crm-sync
    description: >-
      Sync CRM knowledge notes in a vault with the live CRM code repositories: pull the read-only
      repos, detect which changed, and refresh only the affected reverse-engineering notes and the
      keep/drop/decide note. Triggers: "/crm-sync", "refresh CRM knowledge", "what changed in the
      CRM code".
    license: MIT
    ---
    
    # /crm-sync - keep the CRM knowledge layer fresh
    
    The CRM (CHARM) is Anton's proven engine on GitLab. We reverse-engineered it once
    (2026-06-19) into a linked vault layer so his digital twin can `/ask` how the CRM
    works. Code keeps changing; this skill stops that knowledge from going stale silently.
    It is the recurring half of the standing rule `evaluate-recurring-into-routine`.
    
    **What this is NOT:** not a re-decompose-from-scratch. It is a DIFF-DRIVEN refresh:
    pull, see what changed, update only the notes that map to the changed code.
    
    ## Where everything lives
    - **Code (read-only clone, 16 repos):** `<CRM_REPOS_ROOT>` (memory `crm-gitlab`).
    - **Decomposition audit (HTML, source of the notes):** `<CRM_REPOS_ROOT>\_AUDIT\`
      (`Functional-Map.html`, `Know-How-Decomposition.html`, `Architecture-Deep.html`).
    - **Vault knowledge layer:** `$OBSIDIAN_VAULT/05-Resources/CRM-Engine/`
      - `_CHARM-CRM-Engine-MOC.md` (hub)
      - `crm-architecture-4-layers.md`
      - `crm-mtproto-engine.md`
      - `crm-core-autopilot.md`
      - `crm-admin-panel-api-brain.md`
      - `crm-data-model-mongo.md`
      - `crm-know-how-13-patterns.md`
      - `crm-landing-api-money.md`
      - decision: `$OBSIDIAN_VAULT/02-Decisions/decision-crm-keep-cc-drop-decide.md`
    
    ## Repo -> note map (which note to refresh when which code changes)
    | Code area in <CRM_REPOS_ROOT> | Note(s) to refresh |
    |---|---|
    | `admin-panel` / `admin-panel-api` (FastAPI brain, filter-DSL, $facet) | `crm-admin-panel-api-brain` |
    | `mtproto-api` (Telethon hands, event-pool, dialog export, FloodWait) | `crm-mtproto-engine` |
    | `core` (aiocron autopilot: newsletters, gatekeeper, tapping, calls) | `crm-core-autopilot` |
    | `landing-api` (token sale, NOWPayments, CHEATING_MODE, webhook) | `crm-landing-api-money` |
    | shared models / Mongo schema (collections, fields) | `crm-data-model-mongo` |
    | cross-cutting architecture / generations | `crm-architecture-4-layers` |
    | reusable patterns / know-hows | `crm-know-how-13-patterns` |
    | any capability added/removed/disabled | `decision-crm-keep-cc-drop-decide` (+ MOC) |
    
    ## Procedure
    1. **RECALL first (don't duplicate).** Read memory `crm-gitlab` and `_CHARM-CRM-Engine-MOC.md`
       so you refresh, not rewrite. Note the date of the last sync.
    2. **Pull the repos.** For each repo dir under `<CRM_REPOS_ROOT>`:
       `git -C "<CRM_REPOS_ROOT>\<repo>" pull --ff-only` (clone is read-only; this only updates).
       Collect the set of repos that actually moved (non-empty pull, or
       `git -C <repo> log --oneline <old>..<new>`). If a pull needs GitLab creds and fails,
       stop and tell Anton (creds in `secrets\`; do NOT hardcode).
    3. **No changes -> stop.** If nothing moved since last sync, report "CRM unchanged since
       <date>, 0 notes touched" and exit. (Cheapest correct answer.)
    4. **For each changed repo,** re-read ONLY the changed files/areas (use an `Explore`
       subagent with `model: 'sonnet'` - this is grunt extraction, per `model-routing-sonnet-grunt`;
       escalate to Opus only if a judgement call). Ask it to report: what behavior changed vs
       what the mapped note currently says, with file+line evidence.
    5. **Backup before writing.** `python "$IMPORTS_ROOT/vault_backup.py"` (rule
       `vault-backup-rule`). NEVER blanket-delete; edit in place.
    6. **Update the mapped note(s)** - the "How it works" / "Weak spots" sections only.
       PRESERVE frontmatter, `origin: external`, `authored_by:` (denis-udot / gleb-taigunov),
       and all `[[wikilinks]]`. If a capability moved KEEP<->DROP<->DECIDE, update the decision
       note too. New behavioral pattern worth a concept? Follow `concept-creation-rules` (create
       if >=3 repeats + entity noun + domain), and relink (`no-orphan-notes-rule`).
    7. **Verify + reindex.** Run the vault broken-link check; then
       `python "$IMPORTS_ROOT/brain_embed_update.py"` (cooldown 15min; nightly @04:00
       will catch it otherwise) so the refreshed notes are `/ask`-searchable.
    8. **Report post-hoc** (not for permission): which repos moved, which notes changed,
       any new concept, any DECIDE item that flipped, and what still needs Anton/Denis.
    
    ## Guardrails
    - Read-only by default; the only writes are git pull (updates the clone) + the mapped
      vault notes, always backup-first.
    - Don't touch the live CRM server or its Mongo - that is the prod-access track, separate.
    - AK-47: this is one markdown procedure. If the CRM grew a whole new service, that is a new
      note + MOC link, still just notes - flag it, don't over-build.
    
    ## When to run
    On demand (after Anton hears the CRM changed), or as a light routine. Pairs with the
    `crm-gitlab` DD-audit and the `decision-crm-keep-cc-drop-decide` open items.
    
    <!--kit-footer-->
    
    ---
    
    **Like this skill?** It is part of [second-brain-starter-kit](https://github.com/tonydzi/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](https://github.com/tonydzi/secondop-panel) (a second opinion from a panel of external models), [claude-memory-tidy](https://github.com/tonydzi/claude-memory-tidy) (stop your agent's memory from rotting), [telegram-mcp-kit](https://github.com/tonydzi/telegram-mcp-kit) (your own Telegram over MCP in about 15 minutes).
    
    Author: **Anton Dziatkovskii**, Palo Alto AI Research Lab. Telegram [@tonydzi](https://t.me/tonydzi) - WhatsApp [+1 341 222 9178](https://wa.me/13412229178) - X [@Tony_Stef_](https://x.com/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.
    

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.