Compress source documents into lossless, token-budgeted bullet distillates.
npx skills add https://github.com/curdx/curdx-flow --skill bmad-distillator-curdx
Please help me install this Agent Skill. Skill: bmad-distillator Source: https://github.com/curdx/curdx-flow/tree/main/.agents/skills/bmad-distillator Command: npx skills add https://github.com/curdx/curdx-flow --skill bmad-distillator-curdx
This Skill solves the problem of wasting context window tokens and cluttering prompts with bulky source documents, by turning multiple inputs into a single lossless, token-efficient distillate that an LLM workflow can consume without information loss.
Provide the Skill with your source documents and a destination output path, and it will generate a distillate that fits your token budget while preserving all required information.
High-intent search queries and answers about installing and using this skill.
Lossless document distillation compresses multiple inputs into a dense, bullet-based distillate optimized for downstream LLM consumption. It preserves every fact, decision, constraint, and relationship while stripping overhead to save context window tokens.
Semantic splitting by token budget automatically divides output into an index plus self-contained topic sections when results exceed a target size. This ensures generated distillates fit precisely within specified LLM context constraints without breaking topical coherence.
Verification and optional round-trip validation check distillate completeness and format rules, and can optionally reconstruct originals from the distillate to test losslessness. This confirms every fact survived the token optimization process.
Provide source documents and a destination output path to generate a token-efficient distillate. The process applies stage-based routing, theme extraction, and deduplication to produce a single lossless output tailored to your downstream consumer's needs.
When outputs exceed the token budget, distillation triggers semantic splitting into an index and self-contained topic sections rather than truncating data. Strict format compliance and optional completeness repair passes ensure no information is lost during this division.
Specifying a downstream consumer is optional but recommended for optimizing the distillate format. The caller may also define a token budget to trigger semantic splitting and an output path to save the compressed results for the target LLM workflow.
ImaginerLabsDistill mixed source documents into lossless, token-efficient distillates with frontmatter.
amitmishra-bylddCompress multi-file source documents into lossless, token-efficient distillates for LLM workflows.
joekhosbayarGenerate lossless distillates from mixed source documents for LLM workflows.
baobao0303Compress source documents into lossless token-efficient JSON distillates.
