augur docs

augur reads a git diff and tells you how risky it is, and whether a human should look, as a deterministic, scriptable verdict: proceed, review, or block. No API key, no LLM in the core. AI is optional and additive.

It’s built for the world where agents write most of the code: humans can’t hand-review the volume, and agents have no native sense of “I’m out of my depth here, escalate.” augur is that missing primitive: language-agnostic and CI-agnostic.

  • Humans use it to triage: spend review attention on the risky 10% of a 40-file PR.
  • Agents use it to gate: augur gate exits non-zero so an agent escalates instead of merging blind.

Quick start

swift build -c release
install -m 0755 .build/release/augur /usr/local/bin/augur

augur check                         # assess working-tree changes
augur check --range main..HEAD      # assess a range (range-first)
augur gate --threshold review       # exit 1 if verdict >= review (CI / agents)

Requires Swift 6 and git on PATH. augur runs on macOS and Linux.


How it scores

Every signal is derived from git history and the filesystem. No model, no network:

SignalWhat it catches
sensitivityTouches secrets, auth, crypto, payments, migrations, infra, CI, or dependency manifests.
test-gapCode changed with no test in the changeset, or, with a coverage report, the fraction of changed lines left uncovered.
churnHot files that change constantly are fragile.
couplingA file’s usual co-change partner is absent from the change.
diff-shapeLarge single-file edits are harder to review.
ownershipBus-factor (single author) or diffuse ownership (many authors).
incidentThe file’s own history of reverts / hotfixes.
codeownersA changed file with no declared owner in the repo’s CODEOWNERS.

Scoring has two layers: a transparent heuristic prior with documented weights (always applies, even on a brand-new repo), and a history calibration that scales the incident signal by how much the repository’s own revert/hotfix record backs it. Every assessment reports calibration (prior-onlyweakhistory-backed) so you know whether a score is guessing or grounded.


Learn more

New to augur?Going deeper
Quickstart: install and first verdictSignals: every signal, weight, and tuning knob
CLI reference: every command and flagConfiguration: the full .augur.toml reference
Architecture: AugurKit vs the CLI, the pipelineCoverage: LCOV / Cobertura / JaCoCo / Go
CI integration: gate, SARIF, pre-commit, attestDogfooding: augur scores augur (real captured proof)
View on GitHub