| name | deepseek-reasoning-transition-agent |
| description | Reference older DeepSeek R1-specific reasoning examples and migrate them to current DeepSeek API operations guidance. Use when preserving an open-model reasoning example without treating raw chain-of-thought exposure as a stable product contract. |
| measurable_outcome | Reframe a DeepSeek R1 example as a current provider-aligned reasoning workflow with one safety or compatibility note. |
| allowed-tools | ["read_file","run_shell_command"] |
DeepSeek Reasoning Transition Agent
This directory is a transition reference for model-specific reasoning examples.
Workflow
- Use
Skills/AI_Providers/DeepSeek_API_Operations_2026/ for current API, compatibility, and rollout guidance.
- Preserve the useful part of the example: strong reasoning and code-oriented problem solving.
- Do not assume that exposing raw reasoning traces is a durable or desirable application contract.
- Validate structured outputs, retries, and compatibility against the current DeepSeek API docs before production use.
Output Requirements
- State the provider-current DeepSeek skill that should anchor the implementation.
- Name one reasoning workload where the example still helps.
- Include one caution about exposing raw reasoning traces or relying on model-specific behavior.
