` with a click handler |\n`
` + `` |\nan icon-only button with no name |\n`` with `` s |\nan ARIA-less custom dropdown |\n\n**Run an accessibility checker on your app before installing anything.** The result predicts how well this will work better than anything else you could measure — and the fixes it suggests are worth making regardless of whether you adopt this tool.\n\nNot supported yet:\n\n— a hosted payment widget is invisible, so an embedded checkout cannot be driven end to end`iframe`\n\ncontent**hover, scroll-to, drag and drop, file upload****multiple tabs**, and native`alert`\n\n/`confirm`\n\ndialogs\n\n**Windows is untested.** Development and CI run on Linux and macOS. Nothing is known to be broken and reports are welcome ([#8](https://github.com/hamc/blastproof/issues/8)).\n\nIf a critical journey needs one of these, that journey stays with your existing test suite. The two can coexist — nothing here replaces what you already have.\n\n**Use a seeded database, a staging environment you can reset, or a throwaway account. Do not gate on a run against production data.**\n\nWithin a step, an action that commits — a click, or pressing Enter — is never performed twice: the runner refuses the repeat and tells the agent it already did that. This closes the case that used to produce duplicate records, where a submit answered by a redirect came back to a reset form and the agent, seeing no evidence of its own work, submitted again.\n\nIt is **not** a guarantee of zero duplicate writes. An agent that reaches the same effect by a genuinely different route — another control that does the same thing — is not caught.\n\n**The boundary in the middle is the point.** Everything above it runs unattended on every pull request and ends in an exit code. Everything below it is something you choose to run, on your machine, and review before it lands.\n\nA route no test covers is *reported*, never failed — blocking on it would punish you for an incomplete map instead of teaching you to complete it. Turning that report into a test is the manual half, and the draft it produces is not trusted until a person has read it.\n\nTests live in `.blastproof/tests/`\n\nas plain-English YAML — no selectors:\n\n```\nsummary: Checkout with discount\npriority: P0\ntags: [checkout, discount]\nroutes: [\"/cart\", \"/checkout\"]\nsteps:\n - add item to cart\n - apply promo code SAVE20\n - verify a 20% discount is applied\n - complete checkout\n```\n\n`priority`\n\nis P0–P2 (default P1). `tags`\n\n, `setup`\n\nsteps and `auth`\n\nare optional — `auth: false`\n\nruns the test signed out, which a login test needs. `routes`\n\ndeclares the URLs a test covers, which is what `--impacted`\n\nselects on; write route strings consistently, since they compare by exact equality (`/cart`\n\n≠ `/cart/`\n\n). `run`\n\nwarns to stderr — non-fatal — when a test declares a route no `routes:`\n\nmapping declares, since that route contributes nothing to `--impacted`\n\nselection.\n\nThis is the one rule that decides whether a suite works. An outside evaluation took the **same application, same suite, same version from Score 64 to Score 100 by rewriting two steps** — nothing else changed:\n\n```\n# Fragile: a bare action. Nothing says what should be true afterwards.\n- submit the add-task form\n- verify the new task \"Fix the flaky test\" appears in the task list\n\n# Robust: the step carries its own outcome.\n- submit the add-task form and verify the task \"Fix the flaky test\" appears\n in the task list with priority High and status Open\n```\n\nThe bare version fails on a shape that is everywhere: the form POSTs, the server redirects back to the same page, and the form comes back empty. The agent is asked whether \"submit the add-task form\" happened, and is looking at a page that is indistinguishable from one where nothing did. A step that names the outcome gives it something to check that survives the action.\n\nWrite steps that end in an observable result — text on the page, a count, a state change — and this class of failure does not arise.\n\n**Inline error messages should be plain visible text.** `role=\"alert\"`\n\nis read correctly from the accessibility tree and needs no special handling, but note that an alert your page has cleared shows up as an empty element: if a verdict says an alert exists whose content is missing, the message was emptied, not hidden.\n\n```\n- fill the note field with Order not received # runs\n- fill the note field # cannot run\n```\n\nThe agent is **forbidden from inventing values** — one it types must come from the step, from the page, or from an `{{env.*}}`\n\nplaceholder — and the runner enforces it rather than asking. A `fill`\n\nor `select`\n\nwhose value appears in none of those is refused: it is not typed, the agent is told which sources it may draw from, and the step fails on the retry budget if it keeps insisting.\n\nThe rule used to live only in the prompt, and a prompt instructs rather than enforces. Run against a real model, `fill the note field`\n\ndid not fail — the agent made a value up, filled it, and the step **passed**, producing \"This is a test note.\" on two runs and \"This is a new note\" on a third. A test going green over a value nobody wrote, differing between runs, is worse than a failure.\n\nA placeholder counts as a source **only when the step names that variable**. `fill the password field with {{env.TEST_PASSWORD}}`\n\nworks; `fill the password field`\n\ndoes not become valid because the agent supplies `{{env.SOMETHING}}`\n\nitself. An agent cannot know the name of a variable nobody showed it, so one it produces is a guess — and a guessed variable would put a live credential into a field your test never pointed one at, in output that cannot redact a secret it was never told about.\n\nTwo limits worth knowing. A value the page shows in one format and the field wants in another — `1234`\n\nin the step, `1,234.00`\n\nin the box — is refused, and the fix is to write the value the way it is typed. And a very short value (`3`\n\n) appears somewhere in almost any page, so it will pass; this closes fabricated content, not every fabricated character.\n\n`run`\n\nalso warns about it first — the same rule caught earlier, from the test file, on every path, before launching a browser or asking for a key:\n\n```\nAuthoring (a step enters a value but names none):\n Add a note (.blastproof/tests/notes.yaml) step 2:\n fill the note field\n → fill the note field with \n```\n\nNon-fatal by default — `--fail-on-authoring`\n\nturns it into exit 1 for teams enforcing it in CI. Taking the value from the page is fine and is not flagged: `fill the recipient field with the address shown on the confirmation page`\n\n.\n\n**The warning reads English only.** A suite written in another language runs exactly as well but is not inspected, and prints no warning saying so — silence from this check means \"nothing found in English\", never \"this suite is clean\". The runner's refusal has no such limit: it compares text rather than parsing grammar, so a suite in any language is still held to the rule at run time.\n\n```\nblastproof init # scaffold .blastproof/\nblastproof run # run every test\nblastproof run --impacted --base main # run only what the diff can affect\nblastproof plan --base main # draft tests for uncovered routes\nblastproof test --base main # run what covers the diff, then draft the gaps\n```\n\nCommon flags — `blastproof --help`\n\nhas the full list:\n\n| flag | |\n|---|---|\n`--dry-run` |\nPrint the selection (or, for `plan` , the routes it would draft) and exit. No browser, no API key |\n`--tag` · `--priority` · `--query` |\nSelect a subset of tests |\n`--url ` |\nOverride `base_url` for this run (e.g. a PR preview) |\n`--min-score ` |\nGate on a weighted score instead of all-must-pass |\n`--fail-on-unmapped` |\nFail when a changed file matches no `routes:` or `ignore:` glob |\n`--fail-on-authoring` |\nFail when a step enters a value but names none (warns by default) |\n`--junit [path]` · `--html [path]` |\nWrite reports |\n`--concurrency ` |\nRun tests at once —\n|\n\n`--write`\n\n`plan`\n\nonly — persist drafts instead of previewing`--max-llm-calls`\n\n· `--max-tokens`\n\n· `--max-duration`\n\n[Bound what a run may spend](/hamc/blastproof/blob/main/docs/configuration.md#budget--bounding-what-a-run-spends)Exit codes: **0** pass, **1** the gate failed, **2** usage or config error.\n\n**Generated drafts are never executed and never affect the score.** An unreviewed model-written test in the merge path fails in two directions: a hallucinated expectation blocks a correct PR, and a credulous one waves a broken change through while looking like coverage. `plan`\n\nmakes the gap visible with a draft to review; it does not make an uncovered route safe.\n\n`--impacted`\n\nruns only the tests whose `routes:`\n\nintersect the routes your diff can affect, mapped from changed files by globs you maintain in `.blastproof/config.yaml`\n\n:\n\n```\nroutes:\n \"src/cart/**\": [\"/cart\", \"/checkout\"]\nignore:\n - \"**/*.md\"\n```\n\nThe key is the file glob and the value is the routes it can affect — the opposite way round from a test file's own `routes:`\n\n, which is a plain list of the routes that test covers. Inverted, the map matches nothing at all and every run reports a diff that affected no page, so `blastproof`\n\nrefuses a map written that way rather than running green against it.\n\nEvery changed file lands in one of three buckets: it matches `routes:`\n\nand contributes them, matches `ignore:`\n\nand is knowingly irrelevant, or **matches neither — nobody has said what it affects**. `--fail-on-unmapped`\n\nblocks on that third case, naming the files and both ways to resolve them.\n\n**Nothing is ignored by default**, on purpose: a default that guesses on your behalf would hide the first files worth thinking about. The flag is additive — a run can meet `--min-score`\n\nand still be blocked here, because \"the tests I ran passed\" and \"something changed that nobody classified\" are different claims.\n\nIts limit is worth knowing: it catches files that are *unclassified*, not *misclassified*. A shared module mapped to one route when it can break five still slips through. Impact by import graph is the fix, and blastproof does not do it yet.\n\nEach run scores the percentage of executed test **weight** that passed, weighing 3 at P0, 2 at P1, 1 at P2 — so a failing checkout costs three times a failing tooltip.\n\n```\nblastproof run # any failure exits 1 (strict, the default)\nblastproof run --min-score 80 # one failing P2 is tolerated\n```\n\n`--min-score`\n\n**replaces** the all-must-pass rule rather than adding to it. Only executed tests count: filtered and unrouted tests are neither numerator nor denominator, and a run that executed nothing scores 100, so a docs-only PR is never blocked. JUnit carries the score as a ``\n\n, and unrouted tests appear as `