Failed to get model definition wit `/compact` + Docker Model Runner

[k33g]

When running docker-agent against a model served by Docker Model Runner (provider dmr), the manual /compact command always fails with the error: Failed to get model definition

Additionally, automatic compaction never runs.

Environment

• Provider: dmr (Docker Model Runner, local inference via llama-server)
• Example model: huggingface.co/unsloth/qwen3.5-4b-gguf:Q4_K_M
• provider_opts.context_size: 32768 is set in the agent YAML
• base_url: http://localhost:12434/engines/v1

Steps to reproduce

1. Configure an agent that uses a local DMR model (see agent.yaml snippet below).
2. Start a conversation and let it grow.
3. Run /compact at any point.

  models:
    qwen-local:
      provider: dmr
      model: huggingface.co/unsloth/qwen3.5-4b-gguf:Q4_K_M
      base_url: http://localhost:12434/engines/v1
      max_tokens: 24576
      provider_opts:
        context_size: 32768

Expected behavior

• /compact should produce a summary of the conversation and free up context.
• When the conversation approaches the configured context_size (32768), proactive compaction should trigger automatically.

Actual behavior

• /compact emits Failed to get model definition and exits without compacting anything.
• The conversation continues to grow until the context window overflows; automatic compaction never fires.

Version of docker-agent: 1.50.0