← All skills
loop
Use when the user runs /ar:loop or asks to run an autoresearch experiment continuously on a schedule, with a user-selected interval: 10min, 1h, daily, weekly, or monthly.
About this skill
Use when the user runs /ar:loop or asks to run an autoresearch experiment continuously on a schedule, with a user-selected interval: 10min, 1h, daily, weekly, or monthly.
SKILL.md
loop.SKILL.mdPlace it in your Syntic skills folder. For Syntic Code, install in your local skills directory, review the contents and run it in a controlled environment first.
--- name: loop description: Use when the user runs /ar:loop or asks to run an autoresearch experiment continuously on a schedule, with a user-selected interval: 10min, 1h, daily, weekly, or monthly. category: Engineering version: 1.0.0 tools: [] --- # /ar:loop — Autonomous Experiment Loop Start a recurring experiment loop that runs at a user-selected interval. ## Usage - `/ar:loop <experiment>` — start a loop (prompts for interval) - `/ar:loop <experiment> 10m` — every 10 minutes - `/ar:loop <experiment> 1h` — every hour - `/ar:loop <experiment> daily` — daily at ~9am - `/ar:loop <experiment> weekly` — weekly on Monday ~9am - `/ar:loop <experiment> monthly` — monthly on the 1st ~9am - `/ar:loop stop <experiment>` — stop an active loop ## What It Does ### Step 1: Resolve experiment If no experiment is specified, list the available experiments and let the user pick. ### Step 2: Select interval If the interval isn't provided, present the options and map to a schedule: | Interval | Cron Expression | Shorthand | |----------|----------------|-----------| | 10 minutes | `*/10 * * * *` | `10m` | | 1 hour | `7 * * * *` | `1h` | | Daily | `57 8 * * *` | `daily` | | Weekly | `57 8 * * 1` | `weekly` | | Monthly | `57 8 1 * *` | `monthly` | ### Step 3: Create the recurring job Schedule a job that, on each run, does exactly ONE iteration: - Review the experiment's results history: what worked, what failed, what hasn't been tried - Make ONE change to the target based on a strategy escalation tied to the run count - Record the change with a clear description - Evaluate the change against the experiment's metric and direction - Read the outcome (KEEP/DISCARD/CRASH) Rules for each iteration: - Only ONE change per experiment run - NEVER modify the evaluator itself - If 5 consecutive crashes appear in the results history, stop the loop and alert - After every 10 experiments, update the strategy summary with what's been learned ### Step 4: Track loop metadata Record the schedule ID, interval, start time, and experiment name so the loop can be checked or stopped later. ### Step 5: Confirm to the user Report the interval, that recurring jobs auto-expire after 3 days, and how to check progress or stop the loop. ## Stopping a Loop When asked to stop a loop: cancel the recurring job, clear its metadata, and confirm how many experiments completed. ## Important Limitations - **3-day auto-expiry**: recurring jobs expire after 3 days. For longer experiments, restart the loop — results persist and the new loop picks up where the old one left off. - **One loop per experiment**: don't start multiple loops for the same experiment. - **Concurrent experiments**: multiple experiments can loop simultaneously only if they're isolated from each other (e.g., on separate branches), which they are by default.
Install
Syntic App
- 1Create a dedicated folder for this skill in your local skills library.
- 2Place SKILL.md into that folder.
- 3Restart Syntic and invoke this skill on matching tasks.
Syntic Code (CLI)
- 1Save SKILL.md in your local Syntic Code skills directory.
- 2Keep related files in the same skill folder.
- 3Run in a safe environment and validate outputs.