agentleFS
Sign inStart free
ryanportfolioryanportfolio/Harness-Firmware/.agents/skills/why/SKILL.md

why

Skill1 starsChanged 21 days agoMITView on GitHub

Fresh critical second look at a recommendation: on $why, or before presenting your own weighty pick (hard to undo, two or more real options, or real time or money at stake). Never from ordinary why questions.

We split this skill into the rules it gives an agent

2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 87 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If the independent reviewer is unavailable or fails, explicitly report the missing independent check.

    line 14
  2. Distinctive:

    Check the candidate criticisms locally, then explain the reasoning, real weaknesses and whether you still recommend the choice.

    line 16

“Common” is measured: kept by at least 6 in 10 of the 87 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Second look at a recommendation

Scope to the recommendation the user names, otherwise the immediately preceding recommendation. If none exists, say so. Ask only if several materially different choices cannot be disambiguated.

Self-run mode: when you invoke this yourself before presenting a weighty pick, review the recommendation you are about to give, with the user message that prompted it as context. Then present the refined recommendation in the session's normal style with one added line, Pressure-tested ($why): <what the reviewer changed or confirmed>, leading with the new pick if the check overturned it. Make no file edits for the pick until the user answers.

Verify the few facts the recommendation depends on. With an exposed collaboration tool, give one fresh read-only leaf reviewer the recommendation and minimal necessary context using fork_turns: "none". Honor an explicit user model choice; otherwise inherit the configured model. Check actual exposure and disclose an unavailable requested model rather than silently substituting it. Ask for missed assumptions, costs and circumstances that would change the decision. No nested review or mutation.

If the independent reviewer is unavailable or fails, explicitly report the missing independent check. You may provide clearly labeled personal reasoning, but cannot claim the independent workflow completed. Do not silently replace it with self-review.

Check the candidate criticisms locally, then explain the reasoning, real weaknesses and whether you still recommend the choice. Keep it short and readable; a clean finding is valid. This is analysis, not authorization to implement a different approach.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free