Compress SKILL.md tokens via a deterministic 8-transform playbook with hash-based validation.
npx skills add https://github.com/bma-d/skill-optimiser --skill skill-optimiser
Please help me install this Agent Skill. Skill: skill-optimiser Source: https://github.com/bma-d/skill-optimiser/tree/main/skill-optimiser Command: npx skills add https://github.com/bma-d/skill-optimiser --skill skill-optimiser
Reduces the token footprint of SKILL.md files by compressing prose and consolidating structure without losing meaning, enabling faster context loading and lower cost.
Run the optimiser on a skill folder to produce a tighter SKILL.md with the same semantics.
High-intent search queries and answers about installing and using this skill.
To compress SKILL.md tokens, apply a deterministic 8-transform playbook that consolidates structure and compresses prose without losing meaning. This reduces the token footprint, enabling faster context loading and lower costs.
The best way to optimize large skills is to indexify them into a lightweight router paired with separate data files. This approach enables just-in-time context loading, ensuring only necessary information is processed at any given time.
You keep a SKILL.md file concise by extracting long procedures into dedicated scripts. This maintains the file's maintainability and reduces its token size while preserving the original procedural information.
Automated prose compression is safe because the process requires backups, uses hash-based change detection, and performs validation to prevent data loss. These controlled transformations ensure the original meaning is preserved.
Yes, you can apply the optimization transforms to all installed skills simultaneously. The optimizer supports both individual skill folders and batch processing of all installed skills using the same controlled playbook.
You need backups before compressing skill files to prevent irreversible data loss during the transformation process. Hash-based change detection and validation work alongside backups to ensure format compression preserves all original information.
lrbjkAssess Unity projects for version, render pipeline, packages, asmdefs, and folder structure.
slabgorbManage context windows with just-in-time loading and subagent patterns.
AesopScottManage prompt template libraries and injection-order rules for AI harnesses.
junechen7414Compress assistant responses into terse, token-efficient prose while preserving technical accuracy.
