Reviewing from the CLI
entelligence review runs the same review engine that powers pull-request comments, but against your local git diff — no PR, no GitHub App, no comments posted anywhere. It's the fastest way to catch issues before you even push.
Install and authenticate
pip install entelligence-cli
entelligence auth loginRun a review
# Review your current changes (uncommitted first, falls back to committed)
entelligence review
# Review only committed changes vs. a base branch
entelligence review --committed-only --base-branch develop
# Control how much detail comes back
entelligence review --priority high
entelligence review --mode verbose
entelligence review --plain # plain text, no rich formatting
entelligence review --prompt-only # minimal, token-efficient output for feeding into an agentFindings print straight to your terminal with the same severity tags (CRITICAL / MAJOR / NIT) and confidence score used on PRs.
Using it in a script or agent pipeline
entelligence review --debug # print raw request/response, log to ~/.entelligence/cli-debug.log
entelligence review --mock # exercise the output format with fake data, no API call--prompt-only is specifically shaped for handing straight to another coding agent — minimal formatting, maximum signal-per-token — useful if you want an agent to read the findings and act on them without a human in the loop.
There's no dedicated exit code for "issues found" vs. "clean" — every run exits 0 on success and 1 on error (auth/network/git failures). If you're gating a script on findings, parse the output (or use --prompt-only/--plain for something easier to parse) rather than the exit code.
A typical workflow
git add .
entelligence review
git commit -m "Your message"or before opening a PR:
git checkout -b feature/my-feature
# ... make changes ...
entelligence review --priority high
git push origin feature/my-featureCode is analyzed transiently to generate the review — it is not permanently stored.
See the full CLI reference for every flag, or entelligence review --help.