| name | ml-llm-delivery-pipeline |
| description | Use when the task is to create, update, sequence, or review the ML/LLM delivery document set as a pipeline rather than a single document. Covers PRD, system design, ADR, service spec, evaluation plan, delivery plan, runbook, security note, and prompt template. |
ML / LLM Delivery Pipeline
Use this skill when the user asks for a delivery-document workflow for an ML or
LLM system, wants to know what document comes next, or wants consistency across
multiple artifacts.
Read First
Read:
../../../ML LLM Delivery Pipeline/ML LLM Delivery Pipeline and Required Documents.md
Then read only the specific template files you need for the requested document.
Document Order
Use this default sequence unless the user explicitly wants a subset:
01 PRD
02 System Design Doc
03 ADR
04 Service Spec
05 Evaluation Plan
06 Delivery Plan
07 Runbook
08 Security Note
09 Prompt Template
Working Rules
- Treat the pipeline document as process guidance and the numbered files as the
actual content templates.
- Keep cross-links between artifacts consistent.
- Reuse project-specific terminology from the repo instead of generic ML terms.
- If the repo already has canonical docs, adapt the pipeline output to the
existing format instead of introducing a second parallel documentation system.
- When writing one artifact, name the upstream dependencies and downstream
documents that must be updated next.
Output Expectations
For each requested artifact:
- keep the original template intent
- replace placeholders with project-specific content
- remove sections that are clearly out of scope instead of leaving junk
- call out open questions explicitly when information is missing