Skip to content

ROUTINES.md

Latest commit

 

History

History
230 lines (172 loc) · 7.49 KB

ROUTINES.md

File metadata and controls

230 lines (172 loc) · 7.49 KB

EvoNexus — Routines & Scheduled Tasks

Routines are automated workflows that run on a schedule via the ADW Runner.

Core vs Custom

Type Location Tracked Description
Core ADWs/routines/ Yes Essential system routines shipped with the repo
Custom ADWs/routines/custom/ No (gitignored) User-created, workspace-specific routines

Core Routines

Routine Script Agent Schedule
Good Morning good_morning.py @clawdia Daily 07:00
End of Day end_of_day.py @clawdia Daily 21:00
Memory Sync memory_sync.py @clawdia Daily 21:15
Weekly Review weekly_review.py @clawdia Friday 08:00
Memory Lint memory_lint.py @clawdia Sunday 09:00
Daily Backup backup.py systematic Daily 21:00

Memory Sync follows the LLM Wiki pattern: extracts knowledge from daily logs, meetings, and git changes, then propagates updates across related memory files (e.g., a role change updates people/, glossary.md, and CLAUDE.md). Updates memory/index.md (catalog) and memory/log.md (operation log) after each run.

Custom Routines

Custom routines live in ADWs/routines/custom/ (gitignored) and are scheduled via config/routines.yaml (also gitignored).

To create a custom routine, say "create a routine" and the create-routine skill will guide you.

config/routines.yaml

daily:
  - name: "My Routine"
    script: my_routine.py
    time: "19:00"
    enabled: true

weekly:
  - name: "Weekly Report"
    script: weekly_report.py
    day: friday
    time: "09:00"
    enabled: true

monthly:
  - name: "Monthly Close"
    script: monthly_close.py
    day: 1
    time: "08:00"
    enabled: true

Scheduled Tasks (One-Off)

Scheduled tasks are non-recurrent actions that execute once at a specified date/time. Unlike routines (which repeat on cron), a scheduled task runs once and is done.

Use cases:

  • "Post no LinkedIn sexta 10h"
  • "Roda o financial pulse amanha 8h"
  • "Envia resumo pro Telegram as 14h"

Creating Scheduled Tasks

Via CLI: Say "schedule this for" or use the schedule-task skill.

Via Dashboard: Go to /tasks → click "New Task" → fill the form.

Via API:

POST /api/tasks
{
  "name": "Post LinkedIn — Summit",
  "type": "skill",           # skill | prompt | script
  "payload": "social-post-writer LinkedIn post about Summit",
  "agent": "pixel-social-media",
  "scheduled_at": "2026-04-11T16:00:00Z"
}

Task Lifecycle

pending → running → completed
                  → failed (can retry)
pending → cancelled

The scheduler checks for pending tasks every 30 seconds and executes them in background threads.

Dashboard

Tasks are managed at /tasks in the dashboard with:

  • Filter by status (pending, running, completed, failed)
  • Create/edit/cancel/delete actions
  • "Run Now" for immediate execution
  • View result/error output

Triggers (Reactive)

Triggers are event-driven actions that execute in response to external events — unlike routines (cron-based) and tasks (one-off scheduled).

Types

Type Description
Webhook Receives HTTP POST from external services, validates HMAC signature, executes action
Event Reacts to integration events (configured in YAML or Dashboard UI)

Supported Sources

Source Signature Example Events
GitHub X-Hub-Signature-256 push, pull_request, issues, release
Stripe Stripe-Signature charge.succeeded, subscription.created, invoice.paid
Linear X-Linear-Signature Issue (create/update), Comment
Telegram Token-based message, callback_query
Discord Ed25519 message_create
Custom X-Webhook-Signature Any user-defined

Configuration

Triggers are defined in config/triggers.yaml (synced to DB on startup) or created via Dashboard UI / Skill CLI.

webhooks:
  - name: "Deploy Notification"
    source: github
    event_filter:
      event: push
      branch: main
      repo: "EvolutionAPI/evolution-api"
    action_type: skill
    action_payload: "discord-send-message Deploy da evolution-api na main"
    agent: pulse-community
    enabled: true

API Endpoints

Method Endpoint Action
GET /api/triggers List triggers
POST /api/triggers Create trigger
PUT /api/triggers/<id> Update trigger
DELETE /api/triggers/<id> Delete trigger
POST /api/triggers/<id>/test Test trigger
POST /api/triggers/webhook/<id> Webhook receiver (public, HMAC-validated)

Webhook URL Format

After creating a webhook trigger, configure the external service to POST to:

https://your-dashboard.example.com/api/triggers/webhook/<trigger_id>

Use the secret (shown on creation) to configure HMAC signing in the source service.

Dashboard

Triggers are managed at /triggers in the dashboard with:

  • Filter by type (webhook/event), source, enabled/disabled
  • Create/edit/delete/test actions
  • Toggle enable/disable
  • View execution history
  • Copy webhook URL
  • Regenerate webhook secret

Dynamic Routine Discovery

Routines are discovered dynamically from script files. The system:

  1. Scans ADWs/routines/*.py (core) and ADWs/routines/custom/*.py (custom)
  2. Extracts agent from docstring (via AgentName pattern)
  3. Builds make-IDs automatically (e.g. financial_pulse.py → fin-pulse)

No hardcoded mappings needed — add a new script and it's automatically available.

How It Works

  1. Scheduler runs embedded in the dashboard (make dashboard-app)
  2. Core routines are hardcoded in scheduler.py
  3. Custom routines are loaded from config/routines.yaml
  4. Scheduled tasks are stored in SQLite and checked every 30 seconds
  5. Each routine/task invokes Claude Code CLI via the ADW Runner (ADWs/runner.py)
  6. Runner logs to ADWs/logs/ (JSONL + metrics)
  7. Reports saved to workspace folders

Manual Execution

# Core routines (dedicated targets)
make morning      # Good Morning
make eod          # End of Day
make memory       # Memory Sync
make memory-lint  # Memory Lint
make weekly       # Weekly Review
make backup-daily # Daily Backup

# Any routine (core or custom) via dynamic runner
make run R=fin-pulse        # Financial Pulse
make run R=community-week   # Community Weekly
make run R=licensing        # Licensing Daily

# List all available routines
make list-routines

# All commands
make help

Agent Teams (Experimental, Opt-in)

Parallel multi-agent versions of consolidation routines. Instead of one agent collecting data sequentially, Agent Teams spawn domain-specific teammates that work in parallel.

Trade-off: ~3-5x higher token cost, but faster execution and each agent uses its own domain expertise.

Target Normal equivalent Lead Teammates
make team-strategy make run R=strategy-digest @sage @atlas, @flux, @pulse, @pixel
make team-dashboard make run R=dashboard @clawdia @atlas, @flux, @pulse, @dex
make team-weekly make run R=weekly-review @clawdia @atlas, @flux, @pulse

Scripts live in ADWs/routines/teams/. These are never scheduled — run manually when you want speed or richer cross-domain analysis.

Requires Claude Code v2.1.32+ and the experimental flag (already enabled in .claude/settings.json):

{ "env": { "CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS": "1" } }