Skip to content

cli-reference.md

Latest commit

 

History

History
147 lines (108 loc) · 5.78 KB

cli-reference.md

File metadata and controls

147 lines (108 loc) · 5.78 KB
Raw

CLI reference

crie is a command-line tool, invoked as crie <command> [arguments] [flags] — there is no server or web interface. This page documents every command and flag in full; for everyday examples, see the README's Usage section.

Quick reference

Command Purpose
crie analyze [PATH] Run the full evaluation and write reports — the main command.
crie ask PATH QUESTION Investigate a free-form engineering question.
crie impact PATH TARGET Estimate the blast radius of changing one file.
crie report JSON_PATH Re-render a saved report.json in another format.
crie list-analyzers List every registered analyzer and its domain(s).
crie init [PATH] Scaffold a commented crie.config.yaml.
crie version Print the installed version.

Every command accepts --help for its flags directly in the terminal, e.g. crie analyze --help.

crie analyze [PATH]

Analyze a repository and write risk reports. PATH defaults to ..

Flag Default Meaning
--config, -c FILE auto-detected Path to a config file (see configuration.md).
--output, -o DIR crie-report (or config's output.directory) Output directory for reports.
--format, -f LIST config's output.formats Comma-separated report formats, e.g. json,markdown.
--analyzers LIST all enabled Comma-separated analyzer names to run; everything else is disabled for this run. See crie list-analyzers.
--exclude GLOB none Additional exclude glob(s); repeatable. Merged with config's exclude.
--max-commits N full history Cap how many commits git-history-based analyzers walk.
--weight-profile NAME balanced balanced | regression-focused | greenfield.
--profile NAME auto-detected Override project-type detection, e.g. backend_service. See software-evaluation.md for the full list.
--focus NAME none debugging | performance | readability | maintainability | architecture | reliability | testing. Restricts and prioritizes analyzers; --analyzers, if also given, wins.
--fail-on-score N none Exit code 1 if any file meets/exceeds N on --fail-on-dimension (config's fail_on_dimension, default overall_risk).
--top-n N 25 How many files/modules the console summary and Markdown report highlight.
--quiet, -q off Suppress the console summary table (reports are still written).

Exit codes: 0 success, 1 a --fail-on-score threshold was breached, 2 a usage error (bad path, invalid config, unknown --focus/--profile).

crie analyze .
crie analyze . --format markdown --output ./out --max-commits 5000
crie analyze . --analyzers git_history,complexity_python --weight-profile regression-focused
crie analyze . --focus debugging
crie analyze . --profile backend_service
crie analyze . --fail-on-score 0.9   # for CI gating

crie ask PATH QUESTION

Investigates a free-form engineering question using a deterministic evidence-driven plan, optionally synthesized into prose by an AI provider. See custom-inquiries.md.

Flag Default Meaning
--config, -c FILE auto-detected Path to a config file.
--model NAME claude-sonnet-5 AI model used for synthesis, if available.
--no-ai off Skip AI synthesis; show the investigation plan and evidence only.
--output, -o FILE none Also write the full investigation as JSON to this file. 
crie ask . "Why is this project difficult to debug?"
crie ask . "What parts are most dangerous to modify?" --no-ai
crie ask . "How difficult would replacing src/db.py be?" --output investigation.json

AI synthesis requires pip install "code-risk-intelligence-engine[ai]" and ANTHROPIC_API_KEY — without either, crie ask still runs fully, showing structured evidence and the plan instead of prose.

crie impact PATH TARGET

Estimates what would likely be affected by changing TARGET (a file path relative to PATH): direct/indirect dependents, related tests, config references, and a categorical change-radius estimate. See impact-analysis.md.

Flag Default Meaning
--config, -c FILE auto-detected Path to a config file. 
crie impact . src/auth/session.py

Exit codes: 0 success, 2 TARGET isn't among the discovered/analyzed files.

crie report JSON_PATH

Re-render a previously generated report.json as another format, without re-running analysis. Useful for iterating on report formatting or generating a different format after the fact.

Flag Default Meaning
--format, -f NAME markdown json | markdown (or a registered plugin reporter).
--output, -o DIR alongside JSON_PATH Output directory.
--top-n N 25 Same as analyze --top-n. 
crie report crie-report/report.json --format markdown

crie list-analyzers

Lists every registered analyzer — built-in and plugin — with the languages it applies to (or language-agnostic) and which software-evaluation domain(s) it feeds, directly or via the category→domain bridge (see software-evaluation.md).

No flags.

crie init [PATH]

Scaffolds a fully commented crie.config.yaml into PATH (default .).

Flag Default Meaning
--force off Overwrite an existing config file. 
crie init
crie init . --force

crie version

Prints the installed version. No arguments, no flags.