@deepseek-ai/dsh-llm-deepseek

Minimal configuration

- name: '@deepseek-ai/dsh-llm-deepseek'
  config:
    apiKeyEnv: DEEPSEEK_API_KEY  # credential reference, resolved per request
    baseURL: https://api.deepseek.com # optional; $DEEPSEEK_BASE_URL then this default
    reasoningEffort: high        # optional; off | low | high | max
    maxTokens: 256000            # optional per-request output cap
    maxRequestFilesBytes: 134217728
    maxInlineRequestImageBytes: 20971520
    maxImagesPerRequest: 600
    filesApiTimeoutMs: 60000

A request selects the route with provider: deepseek-official; the model id passes through to the wire, so new DeepSeek models need no re-registration. Omitted models advertises the text- and image-capable deepseek-flash and deepseek-v4-flash-vision-exp alongside the text-only deepseek-v4-flash and deepseek-v4-pro, each with a 1,000,000-token context window. An explicit list replaces those defaults, and unlisted model ids still pass through as text-only routes. Clients, including model discovery tools, can read the advisory entries through ctx.llm.listModels('deepseek-official'). Image-capable entries may set imagePixelBudget to a positive integer or low, and may set imageMaxBytes. An entry may declare systemPromptUpdate: in-history when its endpoint reads the latest system message at any position of messages as the complete effective system prompt; the adapter reports the mode on the resolved model and the prepared call, and the agent loop then appends a changed prompt after the cached history instead of rewriting the leading system message (decision rule). The default deepseek-flash entry declares this mode; other models require an explicit models declaration, and any value other than in-history fails at load with llm-deepseek: catalog model "" systemPromptUpdate must be "in-history" when present.

Field	Default	Meaning
apiKeyEnv	DEEPSEEK_API_KEY	Credential reference resolved per request through the credentials seam, then the environment
baseURL	https://api.deepseek.com	Endpoint base; $DEEPSEEK_BASE_URL wins when set
thinking	enabled	Deployment policy; disabled locks every request to off
reasoningEffort	high	Default effort: off, low, high, or max
maxTokens	256,000	Per-request output cap; a model's own cap and explicit request values win
defaultContextWindow	1,000,000	Capacity fallback for models without an exact value
models	V41 Flash + V4 Flash + V4 Pro + V4 Flash Vision Exp	Advisory catalog shown by discovery consumers
streamIdleTimeoutMs	300,000	Maximum provider idle time per outstanding stream read
maxRequestFilesBytes	128 MiB	High watermark for retained request-image bytes before oldest-first offload
maxInlineRequestImageBytes	20 MiB	Independent base64 fallback high watermark
maxImagesPerRequest	600	High watermark for retained request-image count
imageOffloadByteQuantum	64 MiB	Files-mode oldest-prefix removal quantum
inlineImageOffloadByteQuantum	10 MiB	Inline-mode oldest-prefix removal quantum
imageOffloadCountQuantum	20	Count-overflow removal quantum
filesApiTimeoutMs	60,000	Per-image Files resolution deadline
fileExpiresAfterSeconds	604,800	Requested uploaded-image lifetime
fileRefreshMarginSeconds	3,600	Remaining lifetime below which an id is replaced
fileQuotaCleanupBatch	100	Oldest harness-owned files removed before one quota retry
retryPolicy	normal, 5 retries	Provider-owned retry policy executed by dsh-llm-retry

The generated configuration catalog is the exhaustive source for every accepted field and its JSDoc.

Dynamic configuration

Connection facts are re-read once per operation through the optional settings and credentials seams. A llm-deepseek: section in the user settings document overrides any field without a restart; a snapshot that fails a beyond-schema bound keeps the last good facts and logs the failure. The API key resolves per stream call from the same snapshot that supplies the endpoint, image and Files policies, and idle budget, so a rejected settings generation contributes none of them. Image requests resolve the attachment service at request time, so load order does not freeze image availability.