Troubleshooting

`model not found` With Ollama

Common error:

model 'gemma4:e4b' not found (status code: 404)

Check:

ollama list

The model name must match the name shown by Ollama exactly.

`OLLAMA_API_KEY` Is Set, But I Want Local Ollama

Use ollama:<model>. The harness forces http://localhost:11434 by default:

agno-deep-agent --model ollama:gemma4:e4b "Say hello"

For cloud, use ollama-cloud:<model>.

Ollama Runs On Another Host

Pass --ollama-host:

agno-deep-agent --model ollama:gemma4:e4b \
  --ollama-host http://localhost:11434 \
  "Say hello"

`OPENAI_API_KEY is not set`

This warning appears when the provider is OpenAI:

export OPENAI_API_KEY="your-key-here"

Or use local Ollama:

agno-deep-agent --model ollama:gemma4:e4b "Say hello"

The Agent Tried To Run A Blocked Command

CodingTools blocks commands outside the allowlist and dangerous shell operators in restricted mode. For a more conservative run, disable shell:

agno-deep-agent --no-shell "Analyze the files without running commands"

Memory Does Not Seem To Persist

Confirm that you are using the same --db-file, --user-id, and --session-id.

Example:

agno-deep-agent --user-id dev@example.com --session-id local-project \
  "Remember that I prefer concise answers"

agno-deep-agent --user-id dev@example.com --session-id local-project \
  "Continue the session"

GitHub Pages Did Not Update