Skip to main content
A job is a prompt + trigger + optional delivery config. The daemon executes the prompt through the agent loop on schedule.

​
Job Format

Jobs are YAML files in .fliiq/jobs/: 
name: daily-email-summary
trigger:
  type: cron
  schedule: "0 9 * * *"        # 9am UTC daily
prompt: "Check my Gmail and summarize important emails from the last 24 hours"
playbook: blog-seo              # Optional: load a domain playbook
skills:
  - receive_emails
  - memory_read
  - memory_write
delivery:
  type: email
  to: me@example.com
enabled: true
FieldRequiredDescription
nameYesUnique identifier (lowercase, hyphens)
triggerYesWhen to run (trigger types)
promptYesThe instruction for the agent
playbookNoLoad a domain playbook into the job’s system prompt (e.g. blog-seo, coding)
skillsNoRestrict which skills the job can use (all if omitted)
deliveryNoWhere to send results (delivery config)
enabledNoSet to false to disable without deleting

​
Creating Jobs

​
Via CLI

fliiq job create                    # Interactive wizard

​
Via the Agent

In a chat session, ask the agent to create a job: 
"Remind me to exercise every day at 5pm"
The agent calls the create_job tool, writes the YAML file, and the daemon hot-reloads it.

​
Manually

Create a YAML file directly in .fliiq/jobs/: 
nano .fliiq/jobs/exercise-reminder.yaml
The daemon watches for new and changed files.

​
Managing Jobs

fliiq job list              # All jobs with status, next run, last run
fliiq job run <name>        # Manual trigger (runs immediately)
fliiq job logs <name>       # Recent run history
fliiq job output <name>     # Full response from last run
fliiq job delete <name>     # Remove job definition

​
Job Memory

Each job gets its own memory file at .fliiq/memory/jobs/<name>.md. The agent reads this at the start of each run and updates it when done. This enables progressive context — a fitness coaching job that tracks workout progression over weeks, or an email digest that remembers which emails it already summarized. 
# The job's system prompt automatically includes:
# "Your job memory is at .fliiq/memory/jobs/<name>.md —
#  read it at the start of each run to see what you processed
#  last time, and update it when done."

​
Job State

The daemon tracks runtime state in a _state block (managed automatically, don’t edit manually): 
_state:
  last_run_at: "2026-02-12T09:00:12Z"
  last_status: success          # success | error
  next_run_at: "2026-02-13T09:00:00Z"
  run_count: 14

​
Run Logs

Each run produces a JSON log in .fliiq/jobs/<name>/runs/: 
{
  "timestamp": "2026-02-12T09:00:12Z",
  "duration_ms": 4523,
  "status": "success",
  "iterations": 6,
  "final_text": "Summarized 3 important emails..."
}
Last 50 runs are kept. Older runs are pruned automatically.