Greg Brockman highlighted the method as a self-improvement prompt.
Codex Tip: ask Codex to look through your past sessions and turn repeated prompts into reusable skills + subagents
you’ll probably find the same stuff showing up again and again:
“check why CI failed” “review this PR” “write the changelog” “trace this bug” “clean up this diff”
make it a skill if it’s a repeatable workflow
or, make it a subagent if it’s a specific job you want to delegate
Positive users highlight the usefulness of the Codex prompt in converting repeated workflows into reusable skills and subagents, while negative users point to ongoing caching and compaction problems in Codex.
No Digg Deeper questions have been answered for this story yet.
UPDATE: Came up with an even better version of this prompt after the feedback
Ask Codex to look across your sessions, Memories, and Chronicle, identify patterns, reuse what already exists, and only create the smallest useful skill, subagent, or automation.
"Look back over my recent work from the last 30 days, or all available history if shorter, and identify repeated manual workflows worth packaging.
Use available evidence in this order: - Recent Codex sessions and task summaries. - Codex Memories and rollout summaries to find patterns repeated across sessions. - Chronicle, if enabled, to spot repeated work outside Codex. Use Chronicle for discovery only; confirm important details in the relevant source system when possible. - Existing skills, custom agents, and automations, so you reuse or extend what already exists instead of duplicating it.
Look broadly for work that is repeated, time-consuming, error-prone, context-heavy, or benefits from a consistent process. Include workflows across coding, research, writing, planning, communication, operations, analysis, and personal administration.
Only act on a candidate when it: - occurred at least twice, or is clearly likely to recur and costly to repeat; - has stable inputs, a repeatable procedure, and a clear output or stopping condition; - would materially improve speed, quality, consistency, or reliability; - is not already adequately covered.
Choose the smallest appropriate form: - Skill: a reusable workflow or playbook. - Custom subagent: a bounded specialist role or investigation task suitable for delegation. - Automation: a scheduled or recurring check, report, reminder, or monitor. - Skip: work that is too one-off, ambiguous, sensitive, or poorly evidenced to package.
First produce a compact shortlist with: - repeated workflow - supporting evidence and dates - frequency/confidence - recommended form: skill, subagent, automation, extend existing, or skip - why it is or is not worth creating
Then create only the high-confidence missing items. Keep them narrow, practical, source-aware, and easy to validate. Do not create speculative, overlapping, or overly broad assets.
Finish with: - what you created or extended - what you deliberately skipped - what needs more evidence before packaging"
Copy and paste this into your codex:
“Look through my recent Codex sessions and identify repeated workflows or repeated asks.
For anything I keep doing manually, suggest: 1. a skill if it is a reusable workflow 2. a custom subagent if it is a bounded role or investigation task
Focus on practical things like CI failures, PR reviews, changelogs, docs updates, release prep, debugging, and test triage.
Create the useful ones only. Keep them simple.”
self improvement prompt for codex
UPDATE: Came up with an even better version of this prompt after the feedback
Ask Codex to look across your sessions, Memories, and Chronicle, identify patterns, reuse what already exists, and only create the smallest useful skill, subagent, or automation.
"Look back over my recent work from the last 30 days, or all available history if shorter, and identify repeated manual workflows worth packaging.
Use available evidence in this order: - Recent Codex sessions and task summaries. - Codex Memories and rollout summaries to find patterns repeated across sessions. - Chronicle, if enabled, to spot repeated work outside Codex. Use Chronicle for discovery only; confirm important details in the relevant source system when possible. - Existing skills, custom agents, and automations, so you reuse or extend what already exists instead of duplicating it.
Look broadly for work that is repeated, time-consuming, error-prone, context-heavy, or benefits from a consistent process. Include workflows across coding, research, writing, planning, communication, operations, analysis, and personal administration.
Only act on a candidate when it: - occurred at least twice, or is clearly likely to recur and costly to repeat; - has stable inputs, a repeatable procedure, and a clear output or stopping condition; - would materially improve speed, quality, consistency, or reliability; - is not already adequately covered.
Choose the smallest appropriate form: - Skill: a reusable workflow or playbook. - Custom subagent: a bounded specialist role or investigation task suitable for delegation. - Automation: a scheduled or recurring check, report, reminder, or monitor. - Skip: work that is too one-off, ambiguous, sensitive, or poorly evidenced to package.
First produce a compact shortlist with: - repeated workflow - supporting evidence and dates - frequency/confidence - recommended form: skill, subagent, automation, extend existing, or skip - why it is or is not worth creating
Then create only the high-confidence missing items. Keep them narrow, practical, source-aware, and easy to validate. Do not create speculative, overlapping, or overly broad assets.
Finish with: - what you created or extended - what you deliberately skipped - what needs more evidence before packaging"
Copy and paste this into your codex:
“Look through my recent Codex sessions and identify repeated workflows or repeated asks.
For anything I keep doing manually, suggest: 1. a skill if it is a reusable workflow 2. a custom subagent if it is a bounded role or investigation task
Focus on practical things like CI failures, PR reviews, changelogs, docs updates, release prep, debugging, and test triage.
Create the useful ones only. Keep them simple.”
Codex Tip: ask Codex to look through your past sessions and turn repeated prompts into reusable skills + subagents
you’ll probably find the same stuff showing up again and again:
“check why CI failed” “review this PR” “write the changelog” “trace this bug” “clean up this diff”
make it a skill if it’s a repeatable workflow
or, make it a subagent if it’s a specific job you want to delegate
Came up with an even better version of this here:
UPDATE: Came up with an even better version of this prompt after the feedback
Ask Codex to look across your sessions, Memories, and Chronicle, identify patterns, reuse what already exists, and only create the smallest useful skill, subagent, or automation.
"Look back over my recent work from the last 30 days, or all available history if shorter, and identify repeated manual workflows worth packaging.
Use available evidence in this order: - Recent Codex sessions and task summaries. - Codex Memories and rollout summaries to find patterns repeated across sessions. - Chronicle, if enabled, to spot repeated work outside Codex. Use Chronicle for discovery only; confirm important details in the relevant source system when possible. - Existing skills, custom agents, and automations, so you reuse or extend what already exists instead of duplicating it.
Look broadly for work that is repeated, time-consuming, error-prone, context-heavy, or benefits from a consistent process. Include workflows across coding, research, writing, planning, communication, operations, analysis, and personal administration.
Only act on a candidate when it: - occurred at least twice, or is clearly likely to recur and costly to repeat; - has stable inputs, a repeatable procedure, and a clear output or stopping condition; - would materially improve speed, quality, consistency, or reliability; - is not already adequately covered.
Choose the smallest appropriate form: - Skill: a reusable workflow or playbook. - Custom subagent: a bounded specialist role or investigation task suitable for delegation. - Automation: a scheduled or recurring check, report, reminder, or monitor. - Skip: work that is too one-off, ambiguous, sensitive, or poorly evidenced to package.
First produce a compact shortlist with: - repeated workflow - supporting evidence and dates - frequency/confidence - recommended form: skill, subagent, automation, extend existing, or skip - why it is or is not worth creating
Then create only the high-confidence missing items. Keep them narrow, practical, source-aware, and easy to validate. Do not create speculative, overlapping, or overly broad assets.
Finish with: - what you created or extended - what you deliberately skipped - what needs more evidence before packaging"
@mc_ees that’s what setting this as an automation would do :)
@reach_vb It should do that automatically
@reach_vb the caching/compaction issue that was eating a lot of usage appears to still be present, can you pass this to the team? thanks
@thesherlocker yes, this plus an automation which refines your skills every week to ensure that they keep updating as your work and delegation style changes
Act like a senior engineer responsible for long-term system quality, not just task completion.
Before coding:
- understand the architecture, - identify tradeoffs, - and plan the cleanest approach.
While coding:
- prefer simplicity over cleverness, - reduce technical debt, - follow existing patterns, - and optimize for readability + maintainability.
After coding:
- review your own work critically, - identify weaknesses or edge cases, - and suggest improvements that increase long-term leverage.
Optimize for durable systems, not just working code.
@reach_vb C excellent, surtout quand on intègre pas forcément un SDK.. pour ma maintenance d’un client REST maison
@gdb The prompt is the easy part. The hard part is the memory layer — and it has to be native to the agent runtime, not bolted on. Most agent frameworks treat session persistence as optional. It's not. Without it, self-improvement is just confident drift with extra steps.
@argofowl Say more? Do you have a repro? Will pass along
@reach_vb great automation to run every 2-3 days
@gdb yikes
The self-reflection prompt works.
The more durable version: externalize the improvement into a file instead of a prompt.
A SKILL.md captures what the agent learned in session and makes it available to every future session, without re-prompting.
Prompts improve sessions. Files improve the system.
@reach_vb or run watchmen and have it do this continuously in the background instead of asking. mines past sessions, writes skills and subagents to disk, cross-agent.
All open source. https://github.com/firstbatchxyz/watchmen
@reach_vb Does this work with the cli as well ?
This is something that bothers me with my setup. Both codex and claude I open in containers everytime. So it's a clean environment to work without any permissions.
But that means the memories they both save on their ~/.codex and ~/.claude are gone and I need to specifically tell them to save them eslewhere.
@emanueledpt actually try this one, it’s much more exhaustive:
@reach_vb i think it also does this automatically, at least it created several skills for me under ~/.codex/memories/skills without me doing anything