A short tour of the ideas the rest of the docs build on.
Two kinds of check
The structural scan (free). Given a change, Shipmoor checks whether the code is real and self-consistent: does every import resolve, does every called symbol exist, does a handler actually do its work? These are the high-confidence failure modes agents introduce. See Findings & rules.
Claim Check. Given a change and the intent behind it, Claim Check asks the question a linter can’t: did this change do what the task asked? Free includes five invocations per month; Pro is unlimited.
Severity and findings
Each finding carries a severity (critical, high, medium, …). You decide what blocks with --fail-on, and you can ask for the reasoning behind any rule with shipmoor explain <rule>. Findings are emitted as human output, deterministic JSON, and SARIF from the same evidence. See Output formats & exit codes.
The three claim-check axes
The claim-check result carries three independent measures. Collapsing them into one number is the most common misread:
| Axis | Answers | Where it shows |
|---|---|---|
| confidence | How sure are we of the intent? | the Source: line |
| coverage | What fraction of expectations could we check here? | the badge |
| maturity | How well-checked is the claim? (verified / partial / gap_disclosed / unprobed / inferred) | the badge headline |
A weak result is allowed to look weak. “Not yet checked” is honest silence, not a pass. See Reading the verdict.
Deterministic decides; an LLM only advises
Shipmoor’s core decision rule: only deterministic, falsifiable evidence can ever block a merge, and only if you opt in. An optional LLM second opinion runs in your own coding agent (BYO-Judge), is labeled as inferred, is excluded from the score, and is structurally unable to gate. Shipmoor hosts and calls no model.
Plans, entitlements, and allowances
Every installation is authorized to an account. Entitlements describe access, while four independent Free allowances govern Claim Check, Review, Test Evidence, and Blast Radius:
| Entitlement | Unlocks |
|---|---|
scan | Unlimited on Free and Pro |
claim_check | 5/month on Free; unlimited on Pro |
code_review | 5/month on Free; unlimited on Pro |
test_evidence | 5/month on Free; unlimited on Pro |
blast_radius | 5/month on Free; unlimited on Pro |
agent_skills | Unlimited on Free and Pro |
agent_harness | Unlimited on Free and Pro |
ide_extension_pro | The IDE extension |
repair_guidance | Repair guidance for findings |
Read account state with shipmoor auth status and allowances with shipmoor usage. See Capabilities, entitlements & allowances.
Local-first, no source upload
Across every plan, analysis runs locally. By default, source, diffs, prompts, findings, file paths, repository names, branches, tests, and acceptance criteria are not uploaded. Shipmoor contacts its services for authentication, entitlements, command allowances, and features you explicitly enable. See Security & privacy.