CtrlK
BlogDocsLog inGet started
Tessl Logo

mcollina/skill-optimizer

Optimizes AI skills for activation, clarity, and cross-model reliability. Use when creating or editing skill packs, diagnosing weak skill uptake, reducing regressions, tuning instruction salience, improving examples, shrinking context cost, or setting benchmark and release gates for skills. Trigger terms: skill optimization, activation gap, benchmark skill, with/without skill delta, regression, context budget, prompt salience.

87

1.14x
Quality

87%

Does it follow best practices?

Impact

87%

1.14x

Average score across 5 eval scenarios

SecuritybySnyk

Passed

No known issues

Overview
Quality
Evals
Security
Files

Evaluation results

100%

Improve a Poorly-Activating Database Migration Skill

Criteria
Without context
With context

Explicit triggers at top

100%

100%

Numbered top-level checklist

100%

100%

Imperative wording in checklist

100%

100%

Fuzzy phrasing removed

100%

100%

Strict phrasing for mandatory rules

100%

100%

Integrated example — realistic input

100%

100%

Integrated example — shows output format

100%

100%

Non-negotiables front-loaded

100%

100%

No split core behavior

100%

100%

Explicit omission warning

100%

100%

100%

36%

Benchmark Report for a Commit Message Skill

Criteria
Without context
With context

Table column order

100%

100%

Delta computed

100%

100%

Universal failures section

0%

100%

Regressions section

50%

100%

Identifies ModelC 0% criterion

100%

100%

Identifies ModelB regressions

100%

100%

High baseline + tiny delta interpretation

100%

100%

Low baseline + high delta interpretation

80%

100%

Low baseline + low skill-on interpretation

10%

100%

Negative delta patch-immediately

0%

100%

100%

Slim Down an Oversized API Documentation Skill

Criteria
Without context
With context

Trigger list preserved

100%

100%

Background prose trimmed

100%

100%

Duplicate examples removed

100%

100%

Single integrated example retained

100%

100%

Decision table or checklist kept

100%

100%

Layered structure created

100%

100%

SKILL.md links to rules

100%

100%

Paragraphs converted to checklists

100%

100%

Rationale moved to rules

100%

100%

audit-notes what was cut

100%

100%

audit-notes what was kept

100%

100%

Token reduction achieved

100%

100%

86%

3%

Fix a Skill That's Making Things Worse

Criteria
Without context
With context

Regressions identified

100%

100%

Regression cause classified

100%

100%

Exact instruction lines cited

100%

100%

Fuzzy language replaced in fix

100%

100%

Must/should boundaries in fix

100%

100%

Positive example in fix

100%

100%

Negative example in fix

30%

60%

Provider-config not broken

100%

100%

Exit criteria documented

0%

0%

Root cause + diff summary

100%

100%

100%

32%

Prepare a Skill Update for Release

Criteria
Without context
With context

Identifies 0% criterion failure

100%

100%

Identifies negative delta

70%

100%

Benchmark run entry recorded

100%

100%

Follow-up issues required

40%

100%

Go/no-go recommendation

100%

100%

Soft pass: measurable gain noted

100%

100%

PR checklist: SKILL.md links updated

50%

100%

PR checklist: benchmark run log entry

25%

100%

PR checklist: validation commands

100%

100%

PR checklist: tracking issues linked

25%

100%

Post-merge loop mentioned

60%

100%

Stale guidance pruning mentioned

0%

100%

Evaluated
Agent
Claude
Model
Claude Opus 4.6

Table of Contents