Skip to content

feat: improve skill scores across 84 agent skills#4871

Closed
popey wants to merge 1 commit intoudecode:mainfrom
popey:improve/skill-review-optimization
Closed

feat: improve skill scores across 84 agent skills#4871
popey wants to merge 1 commit intoudecode:mainfrom
popey:improve/skill-review-optimization

Conversation

@popey
Copy link

@popey popey commented Mar 16, 2026

Hullo @zbeyens 👋

I ran your skills through tessl skill review at work and found some targeted improvements. Here are the ten most improved:

score_card

Here's the full before/after in text form:

Skill Before After Change
heal-skill 0% 91% +91%
proof 0% 90% +90%
generate-command 0% 89% +89%
internal-workspace-devdeps-hoist 0% 86% +86%
resolve-todo-parallel 0% 79% +79%
translate 22% 86% +64%
resolve-parallel 0% 63% +63%
react 0% 61% +61%
testing 0% 61% +61%
lfg 39% 100% +61%
workflows-compound 36% 95% +59%
plan 28% 86% +58%
ce-brainstorm 43% 100% +57%
workflows-brainstorm 39% 95% +56%
changeset 32% 83% +51%
ce-work 43% 93% +50%
docs-plugin 39% 89% +50%
sync-testing-skill 43% 93% +50%
deepen-plan 38% 83% +45%

84 skills improved across .claude/skills/ and .codex/skills/ — average score went from 47% to 81% (+34%).

Changes made:

Validation fixes (0% → 60-91%):

  • Fixed allowed-tools field format: converted YAML arrays to comma-separated strings (compound-docs, heal-skill, proof)
  • Fixed name field: replaced underscores with hyphens (generate-command, resolve-parallel, resolve-todo-parallel)
  • Fixed double frontmatter in .codex/skills/react/SKILL.md

Description improvements (across 80+ skills):

  • Replaced generic "Skill: name" placeholders with specific descriptions including concrete actions, trigger terms, and "Use when..." clauses
  • Added explicit trigger terms users would naturally say when needing each skill
  • Added clear scope boundaries to reduce skill conflict risk

Content improvements:

  • Added brief overview sections to .claude/skills/ thin wrappers before @file.mdc references
  • Improved workflow clarity and structure in lower-scoring skills
  • Fixed typos and informal language

Honest disclosure — I work at @tesslio where we build tooling around skills like these. Not a pitch - just saw room for improvement and wanted to contribute.

Want to self-improve your skills? Just point your agent (Claude Code, Codex, etc.) at this Tessl guide and ask it to optimize your skill. Ping me - @popey - if you hit any snags.

Thanks in advance 🙏

Checklist

  • pnpm typecheck
  • pnpm lint:fix
  • bun test
  • pnpm brl
  • [-] pnpm changeset
  • [-] ui changelog

pnpm typecheck

 Tasks:    53 successful, 53 total
Cached:    0 cached, 53 total
  Time:    19.19s

pnpm lint:fix

> plate@ lint:fix /Users/alan/Projects/auto-p-o/udecode/plate
> biome check . --fix

Checked 2691 files in 6s. Fixed 3 files.

bun test

 2353 pass
 0 fail
 20 snapshots, 3992 expect() calls
Ran 2353 tests across 381 files. [3.90s]

pnpm brl


 Tasks:    51 successful, 51 total
Cached:    0 cached, 51 total
  Time:    4.118s

Hullo @zbeyens 👋

I ran your skills through `tessl skill review` at work and found some targeted improvements. Here's the full before/after:

| Skill | Before | After | Change |
|-------|--------|-------|--------|
| heal-skill | 0% | 91% | +91% |
| proof | 0% | 90% | +90% |
| generate-command | 0% | 89% | +89% |
| internal-workspace-devdeps-hoist | 0% | 86% | +86% |
| compound-docs | 0% | 84% | +84% |
| resolve-todo-parallel | 0% | 79% | +79% |
| translate | 22% | 86% | +64% |
| resolve-parallel | 0% | 63% | +63% |
| react | 0% | 61% | +61% |
| testing | 0% | 61% | +61% |
| lfg | 39% | 100% | +61% |
| workflows-compound | 36% | 95% | +59% |
| plan | 28% | 86% | +58% |
| ce-brainstorm | 43% | 100% | +57% |
| workflows-brainstorm | 39% | 95% | +56% |
| changeset | 32% | 83% | +51% |
| ce-work | 43% | 93% | +50% |
| docs-plugin | 39% | 89% | +50% |
| sync-testing-skill | 43% | 93% | +50% |
| deepen-plan | 38% | 83% | +45% |

84 skills improved across .claude/skills/ and .codex/skills/ — average score went from 47% to 81% (+34%).

Changes made:

Validation fixes (0% → 60-91%):
- Fixed allowed-tools field format: converted YAML arrays to comma-separated strings (compound-docs, heal-skill, proof)
- Fixed name field: replaced underscores with hyphens (generate-command, resolve-parallel, resolve-todo-parallel)
- Fixed double frontmatter in .codex/skills/react/SKILL.md

Description improvements (across 80+ skills):
- Replaced generic "Skill: name" placeholders with specific descriptions including concrete actions, trigger terms, and "Use when..." clauses
- Added explicit trigger terms users would naturally say when needing each skill
- Added clear scope boundaries to reduce skill conflict risk

Content improvements:
- Added brief overview sections to .claude/skills/ thin wrappers before @file.mdc references
- Improved workflow clarity and structure in lower-scoring skills
- Fixed typos and informal language

Honest disclosure — I work at @tesslio where we build tooling around skills like these. Not a pitch - just saw room for improvement and wanted to contribute.

Want to self-improve your skills? Just point your agent (Claude Code, Codex, etc.) at this Tessl guide: https://docs.tessl.io/evaluate/optimize-a-skill-using-best-practices and ask it to optimize your skill. Ping me - @popey - if you hit any snags.

Thanks in advance 🙏
@popey popey requested a review from a team March 16, 2026 09:08
@codesandbox
Copy link

codesandbox bot commented Mar 16, 2026

Review or Edit in CodeSandbox

Open the branch in Web EditorVS CodeInsiders

Open Preview

@vercel
Copy link

vercel bot commented Mar 16, 2026

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated (UTC)
plate Ready Ready Preview, Comment Mar 16, 2026 9:13am

Request Review

@dosubot dosubot bot added the size:XL This PR changes 500-999 lines, ignoring generated files. label Mar 16, 2026
@changeset-bot
Copy link

changeset-bot bot commented Mar 16, 2026

⚠️ No Changeset found

Latest commit: af5d793

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

This PR includes no changesets

When changesets are added to this PR, you'll see the packages that this PR includes changesets for and the associated semver types

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

@zbeyens
Copy link
Member

zbeyens commented Mar 16, 2026

Those skills are from claude plugins. Should be fixed upstream.

@zbeyens zbeyens closed this Mar 16, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

maintenance size:XL This PR changes 500-999 lines, ignoring generated files.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants