Index of every page on this site: llms.txt
All skills
hagaiw avatar

/select-blind-targets

@33ac0e9
Run it yourself

Select and preregister small external Agent Skills repositories for an independent SkillRoll trial before any run outcomes are inspected.

  • 4 files
  • 7.3 KB
  • Updated 2 weeks ago
  • GitHub
Use this Skill: https://skilld.dev/gh/hagaiw/skillroll/select-blind-targets

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

referencescontext.md

≈407 tokens on demand. Your agent reads this file only when SKILL.md points to it.

Selection context

A blind target is selected before its SkillRoll outcome is known. Discovery may use repository metadata and the active skill tree, but not old run reports, leaderboards, issue discussions about eval results, or unpublished conclusions. Do not inspect or copy filenames whose names reveal a prior verdict. When the user supplies a preregistered ordering rule, apply it without substituting a result-informed ranking.

Prefer useful small-repository evidence

  • Favor roughly one to ten active skills when comparable candidates exist.
  • Count skills only below the repository's intended skills root; fixtures and vendored examples are not active skills.
  • Seek variation in task shape, repository layout, age, and skill complexity.
  • Do not choose a famous repository merely because it is easy to recognize.
  • Reject a candidate whose license or access terms do not permit the proposed temporary clone and local changes.

Preregister the target

Record one readable entry per candidate:

- Repository: <canonical URL>
  Revision: <full immutable commit SHA>
  Skills root: <relative path>
  Active skills: <count and paths>
  Why selected: <small-team fit and variation>
  Limitations: <freshness, access, or none>

Resolve mutable branch names to commits before the trial. If remote freshness was not checked, say so rather than calling a local revision latest. Keep the selection artifact free of inference results so it can be handed to a fresh preparation or authoring worker. Mark an unavailable URL, root, path, license, or revision as a limitation rather than guessing it.

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No third-party reports yet.

Signed by skilld at 33ac0e9. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 6 days ago.

Activeupdated 2 weeks ago

README badge

README badge for hagaiw/skillroll/select-blind-targets