npm / modelsOfficial
DSH Agent Default Model
by DeepSeek
Default model selection shared by Agent entry points
CAPABILITY INDEX
Models
09Compare model DSH plugins for provider adapters, vision, retries, token measurement, and inference configuration.
npm / modelsOfficial
DSH Compaction Basic
by DeepSeek
Token-meter-driven compaction policy and LLM summarization backend for the DeepSeek Harness
npm / modelsOfficial
DSH Compaction Tool Result Pruner
by DeepSeek
Replay-safe model-free head/middle/tail pruning for tool-result surface nodes
npm / modelsOfficial
DSH LLM
by DeepSeek
Provider-neutral LLM service interface for the DeepSeek Harness
npm / modelsOfficial
DSH LLM DeepSeek
by DeepSeek
DeepSeek chat-completions adapter for the DeepSeek Harness LLM seam
npm / modelsOfficial
DSH LLM Pi AI
by DeepSeek
pi-ai-backed DeepSeek adapter for the DeepSeek Harness LLM seam (design-verification twin of dsh-llm-deepseek)
npm / modelsOfficial
DSH LLM Retry
by DeepSeek
Provider-routed LLM request retry policy for the DeepSeek Harness
npm / modelsOfficial
DSH Session Title First Prompt LLM
by DeepSeek
First-message LLM provider plugin for DeepSeek Harness session titles
npm / modelsOfficial
DSH Token Meter
by DeepSeek
Replay-aware token measurement service (ctx.tokenMeter) for the DeepSeek Harness
DSH CATEGORY GUIDE
How to choose Models DSH plugins
Model DSH plugins add provider adapters, inference behavior, vision support, retry logic, token measurement, or model-selection controls to DeepSeek Harness. They can change both output quality and operating cost, so provider compatibility and failure behavior matter as much as the model name.
Where these plugins help
- Provider and model adapters
- Vision, retry, and fallback behavior
- Token usage and model selection controls
How to compare
Compare supported providers and modalities, streaming, tool calling, context limits, retry and fallback rules, secret handling, usage reporting, and configuration precedence. Confirm whether the adapter targets the current Harness version and whether provider-specific features degrade safely.
Before you install
Store API keys in the supported secret mechanism, set conservative usage limits, and test tool calls as well as plain text responses. Review retry settings carefully because repeated requests can increase latency and cost without improving the result.
FAQ / 02
Frequently asked questions
Can a model plugin replace the default DSH model?
Some adapters can change the active provider or model, while others only add retries, metrics, or selection UI. Check the package role and configuration precedence before expecting it to replace the default.
What should I test after installing a model plugin?
Test normal prompts, tool calls, long context, streaming, authentication failures, rate limits, and fallback behavior. Also confirm usage reporting so unexpected retries or model changes are visible.