Skip to content

Latest commit

Β 

History

History
50 lines (41 loc) Β· 2.32 KB

File metadata and controls

50 lines (41 loc) Β· 2.32 KB

Hullo @itsmostafa πŸ‘‹

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

Skill Before After Change
rds 74% 100% +26%
eventbridge 77% 100% +23%
bedrock 74% 94% +20%
sns 75% 94% +19%
ec2 83% 100% +17%
dynamodb 75% 90% +15%
step-functions 77% 90% +13%
lambda 83% 94% +11%
s3 83% 94% +11%
iam 83% 94% +11%
cloudwatch 83% 94% +11%
ecs 83% 94% +11%
eks 83% 94% +11%
secrets-manager 83% 94% +11%
cognito 83% 90% +7%
api-gateway 90% 94% +4%
sqs 89% 90% +1%
cloudformation 90% 90% +0%
Changes made

Across all 18 skills:

  • Removed introductory paragraphs explaining what each service is (Claude already knows)
  • Removed Table of Contents sections (unnecessary token overhead for agents)
  • Trimmed Core Concepts sections where they explained fundamentals Claude already understands
  • Moved last_updated and doc_source frontmatter keys to metadata: block (fixes validation warnings)
  • Added validation checkpoints to key workflows (e.g., verify resource created before proceeding)

Description improvements for 6 skills scoring below 100%:

  • bedrock: Added AWS-specific trigger terms (boto3 bedrock-runtime, InvokeModel API, Amazon Titan, knowledge bases)
  • rds: Added RDS-specific terms (Multi-AZ, parameter groups, RDS snapshots, DB instances)
  • dynamodb: Added DynamoDB-specific terms (partition key, sort key, GSI, LSI, read/write capacity units)
  • sns: Added user-friendly terms (push notifications, SMS alerts, publish-subscribe, fan-out patterns)
  • eventbridge: Made actions more concrete (writing event pattern rules, creating cron-based or rate-based schedules)
  • step-functions: Made actions more concrete (writing ASL state machine definitions, configuring retry and catch policies)

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

If you want to run evals yourself, click here.

Thanks in advance πŸ™