CLI Usage
Autonima exposes four main commands:
autonima create-sample-configautonima validate CONFIG [OUTPUT_FOLDER]autonima run CONFIG [OUTPUT_FOLDER]autonima meta OUTPUT_FOLDER
See the CLI reference for generated argument and option details.
Create a Starting Config
autonima create-sample-config > config.yaml
This writes the canonical sample YAML to stdout, so shell redirection produces an editable config file directly.
Validate a Config
autonima validate config.yaml
With an explicit runtime output folder:
autonima validate config.yaml runs/review_a
What it does:
- parses and validates the YAML
- prints a short summary of key config values
- resolves the runtime output folder exactly as
runwould
Run the Pipeline
autonima run config.yaml
With an explicit runtime output folder:
autonima run config.yaml runs/review_a
Useful options:
--dry-runto validate without running-v/--verbosefor more logging--debugfor post-mortem debugging on errors-j/--num-workersto control parallel screening workers--force-reextract-incomplete-fulltextto re-run full-text screening for cachedfulltext_incompletestudies using current files
Omitted Output Folder
If you omit OUTPUT_FOLDER, Autonima derives it from the config path:
autonima run projects/cue_reactivity/default.yaml
Runtime output folder:
projects/cue_reactivity/default/
Run Meta-Analysis
autonima meta expects the folder containing nimads_studyset.json and nimads_annotation.json.
For standard pipeline output, that is usually the outputs/ directory:
autonima meta projects/cue_reactivity/default/outputs
Optional parameters let you change the estimator, corrector, and include-ID filtering.
Common Failure Modes
Invalid Config
Typical causes:
- missing
objectiveorinclusion_criteriain an enabled screening stage - empty
output.directory - empty
search.querywithoutpmids_fileorpmids_list
Fix by running:
autonima validate config.yaml
Missing API Keys
LLM-backed workflows require API credentials:
OPENAI_API_KEY- or
OPENROUTER_API_KEYwhen using an OpenRouter path in the OpenAI-compatible client
Missing Meta Dependencies
If autonima meta fails with an import error, install:
pip install -e .[meta]
Missing Readability Support
Enhanced HTML cleaning needs:
pip install -e .[readability]
and a working Node.js installation.