with one click
int-sync-meetings
// Sync meetings from Fathom — fetch new recordings, save JSON, generate transcripts and summaries, update indexes. Use when user says 'sync meetings', 'sync fathom', 'update meetings', 'sync calls', or similar.
// Sync meetings from Fathom — fetch new recordings, save JSON, generate transcripts and summaries, update indexes. Use when user says 'sync meetings', 'sync fathom', 'update meetings', 'sync calls', or similar.
Review GitHub repos — open PRs, community issues, stars/forks, releases, contributors. Use when user says 'check github', 'github review', 'repo status', 'open PRs', 'github issues', 'repository status', or any reference to checking GitHub repos status.
Review Linear projects — check issues in review, blockers, stale items, sprint progress, and assigned tasks. Use when user says 'check linear', 'linear review', 'sprint status', 'issues in review', 'project status', 'what is blocked', or any reference to checking project/issue status in Linear.
Monthly close kickoff — initiates the month-end closing process with a checklist, simplified P&L (Stripe + Omie + Evo Academy), pending reconciliations, receivables, payables, and action items for the finance team. Trigger when user says 'monthly close', 'start closing', 'closing kickoff', or on the 1st of each month.
Weekly financial report — consolidates Stripe, Omie and Evo Academy data for the week: revenue (courses, subscriptions, tickets), expenses, cash flow projection, overdue accounts, and variance analysis. Trigger when user says 'financial weekly', 'weekly financial report', or 'financial summary of the week'.
Triage and prioritize inbox emails. Summarize unread messages, classify by urgency and category, propose actions (reply, archive, schedule, create task). Use when user wants to review inbox, process unread emails, or needs help prioritizing messages. Outputs structured summary with top priorities and suggested next actions.
Daily consolidation dashboard — reads outputs from all routines (morning, linear, github, community, strategy, meetings, metrics) and generates a unified HTML dashboard. Trigger when user says 'dashboard', 'generate dashboard', 'overview', 'consolidation', or 'general panel'.
| name | int-sync-meetings |
| description | Sync meetings from Fathom — fetch new recordings, save JSON, generate transcripts and summaries, update indexes. Use when user says 'sync meetings', 'sync fathom', 'update meetings', 'sync calls', or similar. |
Complete pipeline to sync Fathom meetings and organize them in workspace/meetings/.
FATHOM_API_KEY configured (see skill fathom)jq installedfathom.sh available at .claude/skills/fathom/fathom.shWhen triggered, execute the steps below in order:
By default, fetch only today's meetings:
# Fetch today's meetings with summary and action items
{project-root}/.claude/skills/fathom/fathom.sh meetings --after "$(date +%Y-%m-%d)" --include-summary --include-actions
If the user specifies a different period (e.g., "sync this week", "sync yesterday"), adjust --after and add --before as needed:
--after "$(date -v-1d +%Y-%m-%d)" --before "$(date +%Y-%m-%d)"--after "$(date -v-7d +%Y-%m-%d)"--after "$(date -v-1m +%Y-%m-%d)"The API already returns default_summary.markdown_formatted and complete action_items — no extra calls needed.
Read the file of already processed IDs:
{project-root}/workspace/meetings/.state/fathom-processed-recording-ids.txt
Compare with the returned recording_id. Process only IDs that do not exist in this file.
IMPORTANT: This step is mandatory and cannot be skipped. If the ID already exists in the file, the meeting MUST NOT be reprocessed under any circumstances — no summary, no tasks, no notification.
If there are no new meetings, stop immediately and stay silent — do NOT send any Telegram notification. Do not continue to the following steps.
For each new meeting, save the complete JSON to:
{project-root}/workspace/meetings/fathom/YYYY-MM-DD/YYYY-MM-DD__{recording_id}__{slug-do-titulo}.json
Where:
YYYY-MM-DD = date from created_atslug-do-titulo = title in lowercase, spaces→hyphens, without special charactersDetermine the project based on the meeting title:
| Title pattern | Project |
|---|---|
| Main API, API | main-api |
| CRM, Product | crm-product |
| Academy, Course | academy |
| Partner, Partnership | partner |
| Financial, NF, Invoice | foundation |
| Planning, Sprint, Grooming | infer from context |
| Sales, Partnership | sales |
| Operations, Internal, Daily | internal |
| (default) | other |
Use the default_summary.markdown_formatted that already came in the API response (Step 1).
Read the template at .claude/templates/meeting-summary.md and fill with the meeting data.
Save to:
{project-root}/workspace/meetings/summaries/{project}/YYYY-MM-DD__{project}__meeting__{slug}__{recording_id}.summary.md
File format (based on the template):
---
date: YYYY-MM-DD
title: {original title}
project: {project}
type: meeting
status: summary
tags: [fathom, meeting]
recording_id: {recording_id}
recording_url: {url or share_url}
people: [{names from calendar_invitees}]
---
{content from default_summary.markdown_formatted}
## Action Items
{list of action_items formatted as checklist:}
- [ ] **{assignee.name}** — {description} ([{recording_timestamp}]({recording_playback_url}))
For each processed meeting, extract the action_items and create tasks in Todoist.
BEFORE CREATING ANY TASK — mandatory anti-duplication check:
Check the local state file:
{project-root}/workspace/meetings/.state/fathom-todoist-sync.json
If the recording_id already has synced tasks, DO NOT create new tasks. Skip to Step 7.
Search Todoist for existing tasks with the meeting title or recording_id in the comment:
todoist list --filter "search: {meeting title}"
If you find tasks that clearly correspond to the same action items, DO NOT duplicate. Record the existing IDs in fathom-todoist-sync.json and skip.
Triage rules (only if passed the check above):
Evolution — all tasks go to the Evolution project in Todoist, unless explicitly instructed otherwiseTodoist task format:
Title: {translated and clear action in PT-BR}
Project: Evolution
Priority: p3 (default) — raise to p2 if blocker or near deadline
Comment:
📋 Origin: {meeting title} ({date})
🎯 Objective: {what this action resolves}
➡️ Next step: {concrete action}
🔗 Reference: {recording_playback_url link}
Execute directly, without intermediate report. Do not list tasks before creating — create and confirm at the end.
CRITICAL: This step must be executed immediately after processing EACH meeting individually, NOT at the end of all. This prevents a crash mid-processing from causing reprocessing.
Add the recording_id to the state file:
{project-root}/workspace/meetings/.state/fathom-processed-recording-ids.txt
One ID per line. Append, do not overwrite.
Also update fathom-todoist-sync.json with the created task IDs.
Order per meeting: Step 3 → 4 → 5 → 6 → 7 (write state) → next meeting.
When finished, present a short summary:
## Sync Fathom — Completed
**Period:** {oldest date} → {most recent date}
**New:** {N} meetings processed
**Already processed:** {M} skipped
**Tasks created:** {T} in Todoist (Evolution project)
### Synced meetings:
| Date | Title | Project | Tasks |
|------|--------|---------|---------|
| ... | ... | ... | {N tasks} |
Without listing tasks one by one — just counts. If the user wants details, they ask.
Only if at least one new meeting was processed, write this as the last line of your output:
TELEGRAM_MSG: 🎙️ Sync Fathom — N reunião(ões) processada(s) | N tarefas criadas
If no new meetings were processed (stopped at Step 2), do NOT write a TELEGRAM_MSG line.
fathom-processed-recording-ids.txtstatus: pendingfathom/YYYY-MM-DD/, raw/{project}/, summaries/{project}/)raw/ and summaries/ before saving