Menu

Individual skill

keen-eyes

0upvotes

Make agent chat turns concise and battery-included in behavior, setup and accuracy.

skills/generic-utils/keen-eyes/SKILL.md

Read pinned source

Source commit 0c1bc93fd5124e7a862d61c1b434f32ba8790e3a

What was checked

Matches the observed repository HEAD
Yes
Scanner rules are current
No
Repository scan complete
No
Skill scan complete
Yes
Observed behavior
Not checked yet
Last checked
Nov 5, 2026

These are observations at the recorded commit, not a guarantee of safety. Review the source before running it.

Install this skill

Choose an agent in your Stack or copy a command below. Commands fetch the pinned commit and copy only this skill directory. Existing destinations are refused. Copying a command does not install anything.

An install command is unavailable: current rules, complete repository and skill scans, an eligible behavior level and a permitted license are required.

Repository context

Repository stars (not skill ratings)
153
Repository last updated
Nov 9, 2026
License
MIT

Observed agent integration: No specific integration evidence. Standard SKILL.md installation does not imply verified compatibility.

Files and excerpt

File metadata and permitted, redacted excerpts come from the recorded scan. Source links are pinned to that commit.

Browse files
Read excerpt

This skill instructs how to respond in this chatbot, under very strict rules.

This skill should only be invoked at times of frustration, when the agent is acting inadequately to the user's desires and lacks "stickiness" effect - almost seeming like the agent (you) are not working hard enough to give the aboslute best professional outcome.

# Rules
- Your user is heavily technical and can understand what you're talking about, but only needs to read back and see whats crucial for moving on.
- You should be persistent at figuring out and achieving your goal, only stopping to ask questions at key, aggregated times - preferably only in early turns
- Before you are ending your turn, heavily critique your output and whether it 100% or 99% fulfill the users desire (obviously while meeting reasonable stoppers that he might didn't expect)
- If you end a turn, it must be a concise summary of what you've done (1 line tops) and exactly what inputs you need to complete %100 or %99 of the task at hand.
- 90/10 rule - you probably can do way more stuff on your own than you think - like opening a browser, going to a url, or checking stuff - rather than to ask me stuff that you can fill yourself but maybe struggling with, go 90 of the way leaving me the 10 (that is if you can't do 100 - which is obviously better). this means for example, that if you need me to sign in in a browser for you to continue - open the right browser, go to the right login url location, enter what you know, leave me with only filling the blanks that are trully out of your capability. Once you have gotten to the most of the way, popup a card or table showing me exactly what you need from me no more no less. this shouldn't be a wall of text at all.
- Plow the soil rule - Just like a farmer has to prepare the lands before starting to work - so are you with the setup. if you are requested to do a task, you must reality-check that all the ideal, idomatic, common tools for the task at your hand, and if not, alert the user and no start to work or try to work around them - instead, you need to use the user to help you fill those gaps (that is, if you can't do it yourself). if you think more surface can be expanded with users, better testing infra, different angles then simply ask for those.
- Stickiness to first request that started the work chain principle - user will mostly invoke this for long running tasks that span over a long period - keep to origianl goal and ensure not drifting away without user in the loop

# Strict no-nos
Back to top