Adaptive prompt selection
Skill roronoazoroshao369/vibe-coding-os/skills/core/adaptive-prompt-selection
Vibe Coding OS — Claude/Codex/Cursor skill framework with 139 skills, 111 commands, 95 templates, 22 tracked sources, 28/28 validation gates PASS. Quality Shield, Engineering Discipline Pack, plugin marketplace.From the repository description
npx -y skills add roronoazoroshao369/vibe-coding-os --skill adaptive-prompt-selectionAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
5.3 KB, ~1.2k tokens by cl100k_base, as published. Nobody here has run it
Skill: Adaptive Prompt Selection
Purpose
Automatically classify a task into a type (feature, bugfix, refactor, security, migration) and select the most relevant quality packs from the skill registry so the agent applies proportional quality gates without manual curation each time. This replaces the "which checklists should I run?" decision with a data-driven lookup.
When to use
Use at the start of any non-trivial coding task to determine which quality packs (checklists, review prompts, specialized skills) should be composed into the working context. Triggers include "what quality checks should I run", "which packs apply to this task", or any time the agent begins implementation without having selected its quality gates.
Inputs
- Task description — what the user asked for
- Changed files / blast radius — which files or systems are affected
- Task type — classified as one of:
feature,bugfix,refactor,security,migration - Quality pack registry — the full list of available quality packs from
registry/skills.jsonand checklist skills
Quality pack matrix
The matrix maps task types to recommended quality packs. Use templates/adaptive-prompt-matrix.md as the canonical reference.
| Task Type | Recommended Quality Packs |
|---|---|
| feature | API quality (if endpoint added) · DB migration (if schema change) · Frontend state (if UI state involved) · Async jobs (if background processing) · Self-review · Adversarial review |
| bugfix | Bug-fix lifecycle · Disciplined diagnosis · Self-review · Diff audit (if risky area) |
| refactor | Self-review · Diff audit · Quality rubric · Adversarial review (if architecture-sensitive) |
| security | Auth quality · Adversarial review · Privacy filter · Diff audit |
| migration | DB migration quality · API quality (if endpoints change) · Self-review · Adversarial review |
Domain amplifiers: If the task touches a specific domain, add that domain's quality pack regardless of task type. Example: a feature that adds an auth endpoint gets API quality + Auth quality + Self-review + Adversarial review.
Workflow
- Classify the task. Read the task description and changed files. Assign a primary task type from the matrix above. If the task spans multiple types, use the heavier type and add domain amplifiers.
- Identify domain amplifiers. Check which systems the task touches: API endpoints, database schema, authentication, frontend state, async/background jobs. Each match adds its quality pack.
- Look up the matrix. For the classified task type, retrieve the base quality pack list from the matrix.
- Apply domain amplifiers. Add any domain-specific packs not already in the base list.
- Prioritize. If the resulting list is long (6+ packs), apply the
adaptive-flowtier logic: tiny/small tasks get the lightest subset; medium+ get the full set. - Output the recommended prompt stack. List the packs with their skill paths so the agent can load them.
- Compose into the task context. The agent loads each recommended pack and runs it as part of the task workflow.
Outputs
A recommended stack of quality packs (skill names and paths) tailored to the specific task type and domain, ready to be loaded and executed.
Failure modes
- Classifying a security-critical feature as plain "feature" and missing the auth quality pack
- Over-loading quality packs for tiny tasks, adding unnecessary overhead
- Ignoring domain amplifiers and only using the base matrix row
- Never updating the matrix when new quality packs are added to the registry
- Applying all packs unconditionally without considering task tier
Verification checklist
- Task type is classified and stated explicitly.
- Domain amplifiers are identified based on changed files or systems.
- Base quality packs are selected from the matrix for the task type.
- Domain-specific packs are added when relevant.
- Pack count is proportional to task tier (not too many for small tasks).
- Each recommended pack has a valid skill path in the registry.
- The recommended stack is confirmed with the user before execution.
Related skills/commands
commands/vibe-adaptive-prompt.md— command entry point for classifying and selecting packstemplates/adaptive-prompt-matrix.md— the canonical task-type-to-pack matrixskills/core/adaptive-flow/SKILL.md— tier-based workflow selectionskills/core/model-weakness-memory/SKILL.md— complementary weakness-aware checksskills/core/quality-execution-contract/SKILL.md— pre-coding quality contractskills/prompts/quality-rubric/SKILL.md— universal quality baseline
Ghi chú tiếng Việt
Tự động phân loại task (feature, bugfix, refactor, security, migration) và chọn các quality pack phù hợp nhất từ registry. Ma trận task-type → quality-pack được lưu trong templates/adaptive-prompt-matrix.md. Workflow: phân loại task → xác định domain amplifiers → tra cứu ma trận → ưu tiên theo tier → output prompt stack. Kết hợp với adaptive-flow để đảm bảo số lượng pack tỷ lệ với độ phức tạp task.
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.
Gives 0 of the 12 instructions most prompt engineering skills give in ~1.2k tokens
Counted across 542 of the 575 authors here whose files we hold, read 2026-09-06
- Provide few-shot examples for complex tasksin 17 of 542, across 16 files
- Ask clarifying questions if information is ambiguousin 16 of 542, across 14 files
- Output a complete optimized prompt for the userin 15 of 542, across 9 files
- Validate structured outputs against schemasin 15 of 542, across 13 files
- Analyze the draft prompt for intent and gapsin 14 of 542, across 8 files
- Detect project tech stack from local filesin 14 of 542, across 8 files
- Recommend a model based on task scopein 13 of 542, across 7 files
- Present results in the specified output formatin 13 of 542, across 7 files
- Match intent and scope to ECC componentsin 13 of 542, across 7 files
- Ask one question at a timein 13 of 542, across 12 files
- Respond in the same language as the user inputin 12 of 542, across 6 files
- Ask up to three clarification questions if context is missingin 11 of 542, across 5 files
Said here and by no other author read
- identify relevant domain amplifiers
- retrieve base packs from the matrix
- add domain specific packs
- prioritize packs based on task tier
- output the recommended prompt stack
- confirm the stack with the user
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.