All apps
VerisKit / Changelog
What shipped, version by version.
What shipped in each VerisKit release, mirrored from the public repository.
0.5.1 — 2026-07-13
Changed
--githubpublish errors now include GitHub's own message (for example "Resource not accessible by integration" when the workflow token lackspull-requests: writeorchecks: write), so a failed publish is diagnosable. The token still never appears in any error.homepagenow points to the VerisKit product page.
0.5.0 — 2026-07-13
Added
- Publish to GitHub.
veris verify --githubposts and updates one sticky PR comment with the verdict and report, and creates a Check Run (verified passes, failed fails, partial is neutral). Token read fromGITHUB_TOKEN; publishing never changes the verdict or exit code.veris badgewrites a shields.io endpoint JSON. GitHub API over built-in fetch, no new dependency. - Browser tests. A real Playwright runner, opt-in with
veris verify --browser(or abrowserentry in.veris/config.json). Detected Playwright now shows as an available capability. veris loglists past runs from the stored evidence records, andveris log --flakyflags checks that both passed and failed across recent runs. Local per-machine history.
0.4.1 — 2026-07-10
Added
- Signed evidence.
veris evidence keygencreates an Ed25519 keypair (via Node's built-in crypto, no new dependency);veris evidence sign <evidence.json>writes a detached signature;veris evidence verifychecks a sibling signature automatically and can assert the signer with--pubkeyor--key-id. Bundles carry the signature. Signing is opt-in; unsigned evidence still verifies for integrity.VERISKIT_SIGNING_KEYsupplies the key in CI.
Changed
veris initnow gitignoreskeys/.
0.4.0 — 2026-07-10
Added
- Evidence System. Every
veris verifyandveris affectedrun writes a canonical, git-anchored.veris/runs/<id>/evidence.json(schemaveriskit/evidence@1) with a sha256 integrity digest over the whole record and a sha256 of each per-check log. No new runtime dependencies. veris evidence verify <file>recomputes and checks a record or a bundle, and states plainly what an integrity digest does and does not prove.veris evidence bundlepackages the latest run (record, report, and logs, each digested) into one portable proof file under.veris/evidence/.veris evidence showprints the latest record's key facts.
Changed
- The report and terminal output now show the git commit and whether the tree was clean, plus the evidence digest in the report.
evidence.jsonreplaces the oldermetadata.json.veris initnow also gitignoresevidence/.
0.3.0 — 2026-07-10
Added
veris scan— import-graph map + untested areas, built from the project's own TypeScript (with a dep-free scanner fallback); writes.veris/graph.json.veris plan— prioritized recommendations (high-impact untested files, weak verification, risky changes). Analysis only — no code generation.
Changed
veris affected/veris watchare now graph-based: the unit run is narrowed to only the test files that transitively import your changes, with a conservative full-suite fallback (config/global change, unresolved file, or untested change) so an affected test is never skipped. No new runtime dependencies.
0.2.0 — 2026-07-09
Added
veris affected— run only the checks relevant to changed files (coarse: no import graph yet).--base <ref>for PR/CI diffs.veris watch— re-run affected checks as files change, using native fs.watch (no new dependency) with a--pollfallback and cross-tickcachedresults.- Honest scoped verdicts: affected/watch runs never report a bare "Verified"; unaffected capabilities are shown as "not affected by changes".
Changed
- Extracted the runner base (Runner/RunContext/localBin/runViaExec) into
src/runners/base.tsto eliminate an adapter import cycle.
0.1.0 — 2026-07-08
Added
veris init,doctor,test,verify,report.- Zero-config detection: package manager, TypeScript, Vitest, Jest, node:test, ESLint, Biome, Playwright (detected, not run).
- Parallel check orchestration with a three-state verdict (verified / failed / partial).
- Local evidence store under
.veris/and Markdown verification reports. - CI-correct exit codes (0 verified, 1 failed, 2 partial;
--partial-okoverrides).
More from the studio
Developer ToolsFree
ProjScan
ProjScan tells reviewers when to bootstrap, prove, or stop. Review Gate returns one decision; bootstrap is explicit. Local proof, no code upload.
ViewDeveloper ToolsFree
AgentLoopKit
The local control plane for low-token, verifiable agent loops. It owns scope, gates, and completion decisions, with a token receipt on every step.
ViewDeveloper ToolsFree
AgentFlight
agentflight guard watches local trust while your coding agent works; agentflight finish writes the Review Passport that proves the result.
View