v0.5.0

Highlights

• Interactive terminal UX redesign — live review lanes on stderr with a verdict banner on stdout, an interactive reviewer-set editor, and an autocomplete model picker.
• Model catalogs — curated model catalogs for codex, cursor, opencode, pi, copilot, and droid, with opt-in live catalog tests.
• Host-local config overlay — layer machine-specific settings over a syncable base config, with doctor and init support.
• Launch prep — refreshed README with demo GIF, stability contract, security intake, and documented win32 caveat.

Fixes

• In-lane reviewer failures are surfaced and --out is written even on failed runs.
• Fail fast when the review prompt exceeds the codex app-server input cap.
• Init wizard produces a working config on plain-Enter; recovery hints for missing --target and existing configs.
• Numerous terminal-rendering fixes: cell-accurate truncation, narrow-width banner clamping, live pane caps.
• Config overlay edge cases: symlink handling, local-only reviewer id rejection, default-set guards.

No output-schema change — schema_version stays at 2.

v0.4.0

Highlights

Opt-in reviewer debug output capture and streaming (#15)

• New diffwarden review --debug-reviewer-output: CLI-transport reviewer artifacts gain a bounded debug_output transcript (separate stdout/stderr, 256 KiB per stream per reviewer, UTF-8-safe truncation; retried runs share the budget so failing first attempts survive). With --ndjson, bounded reviewer_debug_output events stream while reviewers run. Defaults are byte-identical without the flag; debug output is non-authoritative and never affects results, validation, gating, or exit codes.

Cross-engine structured-output repair pipeline (#14)

• Schema-parse failures now trigger a short same-engine repair request before falling back to one labeled full re-run; engine-reported spend is summed across invocations.

Interactive setup and reviewer management (#6, #7, #13)

• Host-aware reviewers discover plus interactive-by-default init and reviewers add/edit/remove/set flows (@clack/prompts) with atomic writes and concurrency guards.
• Interactive model catalog picker in setup, backed by live engine listModels.

Claude reviewer improvements (#8–#12)

• xhigh effort passthrough with catalog-aware top-tier clamp; diffwarden-default effort high on claude reviewers.
• Diffwarden review contract delivered as a system prompt on claude transports.
• First-class CLAUDE_CODE_OAUTH_TOKEN support.
• fallbackModel, maxTurns, and maxBudgetUsd reviewer settings.

Review UX

• Diff-backed focus review lanes (--focus), reshaped output modes with a human review display, and batch report aggregation fixes.

Full changelog: v0.3.2...v0.4.0

v0.3.2

Full Changelog: v0.3.1...v0.3.2

v0.3.1

Full Changelog: v0.3.0...v0.3.1

v0.3.0

What's Changed

• AUR-504 tune Claude review tool policy by @aurokin in #3
• AUR-505 tune Cursor review controls by @aurokin in #4

Full Changelog: v0.2.9...v0.3.0

v0.2.9

What's Changed

• Centralize metadata provenance and live doctor executable reporting by @aurokin in #2

New Contributors

• @aurokin made their first contribution in #2

Full Changelog: v0.2.8...v0.2.9

v0.2.8

Changes

• Add Codex web-search controls for CLI and app-server reviewers, including disabled-by-default behavior and inherit/enabled options.
• Add experimental Codex native review mode support for app-server transport.
• Add diffwarden reviewers list for inspecting configured reviewers and reviewer sets without preflight.
• Port the Codex review rubric into Diffwarden prompts.
• Clarify Claude SDK read-tool behavior and Codex app-server documentation.
• Make quality gates easier to run locally and in CI by moving deterministic coverage into unit tests and reducing real Git, process, and e2e suites to canaries.
• Move pnpm onlyBuiltDependencies configuration into pnpm-workspace.yaml for current pnpm behavior.

Verification

• pnpm test:full
• pnpm test:e2e
• node dist/cli.js --version -> 0.2.8

v0.2.7

Changes

• Reuse Codex app-server sessions through the shared CODEX_HOME Unix socket by default.
• Add appServerOptions.mode for auto, attach, launch, and stdio-isolated lifecycle control.
• Allow appServerOptions.codexHome so users can point Diffwarden at a stable alternate Codex home.
• Preserve ephemeral read-only Codex review turns while documenting that shared mode inherits Codex auth, config, plugins, apps, and daemon state.
• Keep the previous temporary CODEX_HOME stdio behavior available as stdio-isolated.
• Add regression coverage for shared socket reuse, bad socket upgrade handling, and isolated auth-source preflight.

Verification

• pnpm check
• pnpm build
• node dist/cli.js --version -> 0.2.7
• pnpm vitest run --config /dev/null test/codex-app-server-adapter.test.ts

v0.2.6

Changes

• Add internal adapter preparation context so preflight can pass private run context without serializing it into artifacts.
• Reuse resolved CLI executables and Claude runtime/auth state across preflight and run.
• Keep Droid SDK preflight as a real executable smoke test while reusing the resolved path for the run.
• Add macOS executable trust diagnostics with diffwarden macos doctor.
• Document macOS quarantine, codesign, Gatekeeper, and Spotlight triage guidance.

Verification

• pnpm lint
• pnpm typecheck
• pnpm vitest run test/droid-adapter.test.ts test/cli-adapter.test.ts test/runner.test.ts test/cli.test.ts
• pnpm build && node dist/cli.js macos doctor --path /bin/ls --json
• Full suite passed under elevated Vitest timeout after default-timeout flakes under local machine load: INTEGRATION_TEST_ON=0 pnpm vitest run --testTimeout 15000

v0.2.5

Changes

• Prefer Claude Code subscription auth over Anthropic API keys by default for Claude SDK and CLI reviewers.
• Add explicit Claude auth mode controls with API-key and Claude Code modes.
• Strip Anthropic API credentials when Claude Code auth is selected so reviewer runs do not silently consume API credits.
• Preserve explicit environment isolation in Claude auth resolution and CLI invocation setup.

Verification

• pnpm check