Skip to content
OpenAgentSkill
Browsing is open · Catalog captured 2026-11-21 · Automatic updates and account services are paused.
Other skills

X SkillOpt learn

use this when running the X SkillOpt learn loop: log every post trajectory, refresh metrics, and nightly gated-edit best_skill.md so the bot only doubles down on what works

YannisKiefer · Static checks on file

Installation

Read the source and dependencies before running a command. Copying does not install or execute anything.

npx skills add YannisKiefer/grokbot-x --skill X SkillOpt learn

About this skill


name: X SkillOpt learn description: >- use this when running the X SkillOpt learn loop: log every post trajectory, refresh metrics, and nightly gated-edit best_skill.md so the bot only doubles down on what works

X SkillOpt learn loop

Use this whenever shipping X posts/replies for the operator, and on the nightly sleep routine. Treat skillopt/best_skill.md as the trainable skill (Microsoft SkillOpt method adapted to X).

Decide, then ship

Do not ask the operator for strategy permission. Report live URLs or real blockers only. No "want me to" questions.

After EVERY ship (rollout)

  1. Read skillopt/best_skill.md BEFORE drafting (during a live trial on a trial day: skillopt/proposed.md).
  2. Ship via the locked Typefully paths in docs/PUBLISH.md.
  3. Log it: python3 skillopt/sleep_cycle.py append --json '{...}' with id, ts, type, url, text, features (viral_ratio, viral_signal, topic, author_followers = YOUR own follower count now, source), metrics placeholder, split (train|val ~80/20), notes. Features are frozen at ship time; later refreshes touch metrics only. On a trial day, add --skill skillopt/proposed.md and use split val.
  4. Refresh metrics later at +1h / +6h / +24h / next night (views / likes / replies / bookmarks).

Nightly sleep (reflect -> edit -> gate)

  1. Refresh metrics on posts older than 1h (read-only).
  2. python3 skillopt/sleep_cycle.py status
  3. If ready_for_edit is false (scored rows < min_trajectories_before_edit): log an epoch skip. Do not edit.
  4. Else: write <= 4 bounded add/delete/replace edits into a copy, skillopt/proposed.md, from train successes vs failures + rejected_edits.md. Targeting rules go in the skillopt-policy block.
  5. GATE: python3 skillopt/sleep_cycle.py gate accept --summary "..." --edit "..." --skill skillopt/proposed.md. The gate decides on held-out rows (live trial if logged, else replay) and adopts only if the gain's interval is above zero. Every look since the last accept (gate or replay; a new proposal, or the same one on new data) raises the bar, so try few, well-reasoned edits and gate a trial only once. Prose-only edits need a live trial first. Never edit best_skill.md by hand.
  6. Mirror accepted rules into the next day's hourly behavior.
  7. Stay quiet unless a skill edit was accepted (one line) or a hard blocker.

Live trial (prose edits)

  1. Start: copy best_skill.md to skillopt/proposed.md and edit it.
  2. Alternate days: odd days draft from proposed.md and log with --skill skillopt/proposed.md; even days draft from best_skill.md and log normally. All trial rows use split val.
  3. Stop: once both sides have >= min_val_rows scored rows, run the gate with --skill skillopt/proposed.md. Accepted or not, the trial ends.

Score

Fixed in skillopt/sleep_cycle.py; see docs/SKILLOPT.md. Cold account: relative scores matter more than absolute views until you have a real ICP graph.

Reference

https://github.com/microsoft/SkillOpt Rollout, reflect, bounded edits, held-out gate, rejected buffer, slow update.