feat: improve skill scores (+46% avg across 5 skills)

[rohan-tessl]

Hey @edwinhu 👋

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
wrds	17%	93%	+76%
ds-implement	40%	89%	+49%
writing-validate	44%	83%	+39%
dev-explore	41%	75%	+34%
dev-implement	45%	77%	+32%

This PR covers your 5 lowest-scoring skill(s) in the repo. The remaining 87 skills can be reviewed incrementally via the included GitHub Action on future PRs.

What changed in wrds

• Description rewritten from a massive wall of trigger terms into a concise quoted string with an explicit "Use when..." clause
• Added triggers array to frontmatter, moving all trigger terms out of the description
• Consolidated the separate Rationalization Table and Red Flags sections into single unified tables for both Query and SAS enforcement
• Preserved all domain-specific content (table names, filter checklists, connection patterns, SAS code validation)

What changed in ds-implement

• Description rewritten with concrete actions (run regressions, create plots, build models) and natural trigger terms instead of internal workflow jargon
• Added triggers array for frontmatter compliance
• Consolidated 4 EXTREMELY-IMPORTANT blocks, the Rationalization Table, Drive-Aligned Framing, and Red Flags into a single concise "STOP If You Think" section
• Trimmed redundant DO/DON'T table, verbose Topic Change Protocol, motivational framing sentences, and edge case lists
• Moved ETL Strategy Enforcement from a large inline table to a brief pointer with reference links
• Preserved flowchart (declared as authoritative spec), gate logic, deviation rules, and all reference file links

What changed in dev-explore

• Description rewritten with natural trigger terms ("understand the code", "how does this work", "code structure")
• Added triggers array for frontmatter compliance
• Consolidated Rationalization Table + Drive-Aligned Framing + "Why Skipping Hurts" into a single concise section
• Trimmed verbose Test Infrastructure Discovery section (removed rationalization table, kept gate function)
• Trimmed Code Path Discovery section (removed 3 inline examples, kept checklist and reference link)
• Reduced EXTREMELY-IMPORTANT blocks from 5 to 1, keeping enforcement in concise directives
• Preserved key files list format, agent launch patterns, gate checks, and workflow sequencing

What changed in writing-validate

• Description rewritten with concrete actions (map claims, verify coverage, run checks, produce VALIDATION.md) and natural trigger terms
• Added triggers array for frontmatter compliance
• Consolidated Rationalization Prevention + Drive-Aligned Framing into a single "STOP If You Think" table with a "Do Instead" column
• Preserved flowchart, claim classification logic, VALIDATION.md template, and all gate/constraint check details

What changed in dev-implement

• Description rewritten with concrete actions and natural trigger terms ("start coding", "kick off implementation")
• Added triggers array for frontmatter compliance
• Condensed Iron Law of TDD from 7-row rationalization table to 3-line directive
• Condensed Iron Law of Delegation from 14-row rationalization table to concise rule with table of allowed/disallowed actions
• Trimmed Failure Recovery Protocol from ~80 lines to a 4-step summary
• Trimmed Step 3 Verification from verbose checklists with rationalization table to compact checklist
• Removed "Why Skipping Hurts" motivational table (redundant with enforcement directives)
• Trimmed "No Pause Between Tasks" and Test Gap Validation sections
• Preserved prerequisite gates, deviation rules, sub-skills reference, agent team protocol link, and all workflow logic

Tessl Skill Review GitHub Action ✅

I've also included a GitHub Action (.github/workflows/skill-review.yml) that automatically reviews any SKILL.md changed in future PRs and posts scores as a PR comment.

What this gives you:

• 🔍 Automatic tessl skill review runs on every PR touching SKILL.md
• 💬 One updated PR comment with scores and improvement feedback
• 🔓 Zero extra accounts — contributors don't need a Tessl login; only GITHUB_TOKEN is used
• ✅ Non-blocking by default — feedback-only, no surprise red CI (add fail-threshold: 70 later if you want a hard gate)
• 📈 Covers future skills incrementally as contributors edit them

Want automatic AI optimization on every SKILL.md change? 🚀

The action I've added gives you review scores on PRs. We also have a more powerful variant — tesslio/skill-review-and-optimize — that can:

• Run AI-powered optimization suggestions on every SKILL.md PR (requires adding TESSL_API_TOKEN as a repo secret)
• Let contributors accept suggested improvements by commenting /apply-optimize
• Still works in review-only mode with zero secrets

Interested? Tick the box and I'll raise a follow-up PR:

• Yes please! Add the tesslio/skill-review-and-optimize action so every SKILL.md PR gets AI optimization suggestions + the /apply-optimize flow
• No thanks — the review scores action is enough for now

---

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 — @rohan-tessl — if you hit any snags.

Thanks in advance 🙏