vaelii.agents.llm.ollama — com.vaelii/vaelii-agents 0.32.0

The Ollama chat API backend, over java.net.http and cheshire, plus the host probes. Beyond the neutral request keys it reads :temperature and :keep-alive. See docs/llm.md § The local backend.

available? clj

(available?)

(available? opts)

Is an Ollama reachable? opts as version's.

base-url clj

(base-url)

The Ollama base URL: VAELII_OLLAMA_HOST, else OLLAMA_HOST when it names a real target, else default-host.

body clj

(body {:keys [model system messages tools num-ctx max-tokens temperature
              keep-alive think]
       fmt :format}
      {:keys [stream?]})

The neutral request -> Ollama's /api/chat body. keep_alive, num_ctx, num_predict and temperature (default 0) are always sent; format, tools and think only when given. :think false stops a thinking model (gemma4, qwen3) from spending num_predict on thinking and answering nothing; "low" etc. for gpt-oss.

capabilities clj

(capabilities model)

(capabilities model opts)

The capability keywords a model declares (#{:completion}, #{:completion :tools}, …), or nil when they cannot be read.

configured-generation-model clj

(configured-generation-model)

The model a generation turn runs: VAELII_OLLAMA_GENERATION_MODEL when set, else default-generation-model.

configured-keep-alive clj

(configured-keep-alive)

How long to ask the host to hold the model resident: VAELII_OLLAMA_KEEP_ALIVE when set, else default-keep-alive.

configured-model clj

(configured-model)

The model to run: VAELII_OLLAMA_MODEL when set, else default-model.

configured-num-ctx clj

(configured-num-ctx)

The context length, in tokens, to ask for: VAELII_OLLAMA_NUM_CTX when set and numeric, else default-num-ctx.

context-length clj

(context-length model)

(context-length model opts)

The model's native context length from /api/show's model_info, or nil.

default-generation-model clj

The model a generation turn runs on, a turn that writes new content rather than rewriting lines it was shown, when VAELII_OLLAMA_GENERATION_MODEL is unset.

default-host clj

The host base-url answers when neither variable names one: Ollama's own default.

default-keep-alive clj

How long the host holds the model resident after a turn, when VAELII_OLLAMA_KEEP_ALIVE and :keep-alive are unset.

default-model clj

The model a turn runs when VAELII_OLLAMA_MODEL and :model are unset.

default-num-ctx clj

The context length, in tokens, a request asks for when VAELII_OLLAMA_NUM_CTX and :num-ctx are unset.

default-num-predict clj

num_predict when a request carries no :max-tokens.

generation-provider clj

(generation-provider)

(generation-provider opts)

provider on configured-generation-model, for a generation turn; an opts :model still wins.

normalize-host clj

(normalize-host host)

A host string -> the base URL to call, read as Ollama reads OLLAMA_HOST: a host with no scheme gets http://, and :11434 when it names no port, while a host with a scheme keeps that scheme's own port. The scheme is matched in any case and written in lower case, an empty host is 127.0.0.1, an unbracketed IPv6 address is bracketed and names no port, and trailing slashes are dropped.

parse-response clj

(parse-response m)

Ollama's chat JSON -> the neutral response map (vaelii.agents.llm.protocol).

probe-client clj

The one HttpClient, delayed, that version, show and warm send on. See docs/llm.md § What both HTTP backends share.

provider clj

(provider)

(provider {:keys [host model num-ctx keep-alive timeout-ms think]
           :or {timeout-ms 300000}})

An Ollama-backed Provider. opts:

:host default base-url :model default VAELII_OLLAMA_MODEL, else default-model :num-ctx default VAELII_OLLAMA_NUM_CTX, else default-num-ctx :keep-alive default configured-keep-alive :timeout-ms default 300000; bounds the whole turn, the body included :think sent as Ollama's think when given: false, true, or "low"/"medium"/"high"

A request's :model, :num-ctx, :keep-alive and :think override these. Building one opens no socket; available? is the reachability check. meta carries :conn.

show clj

(show model)

(show model {:keys [host timeout-ms] :or {timeout-ms 10000}})

/api/show for one model — its template, parameters, and declared capabilities. nil when the host or the model is not reachable.

supports-tools? clj

(supports-tools? model)

(supports-tools? model opts)

Does this model declare :tools? False when the host cannot be reached.

version clj

(version)

(version {:keys [host timeout-ms] :or {timeout-ms 2000}})

The Ollama server's version string, or nil when it is not reachable within :timeout-ms (default 2000).

warm clj

(warm)

(warm model)

(warm model
      {:keys [host timeout-ms keep-alive num-ctx prefill-tokens]
       :or {timeout-ms 300000 prefill-tokens 2048}})

Load model (default configured-generation-model) and hold it resident: one :prefill-tokens (2048) prefill at the request's context length, generating one token. Returns {:loaded? bool :elapsed-ms n :model m :prefill-tokens n} and never throws.

opts: :host, :timeout-ms (300000), :keep-alive, :num-ctx, :prefill-tokens.