v0.5.0

What's Changed

• feat(website): overhaul marketing architecture, typography, blog, and navigation by @lostmartian in #30
• docs(readme): upgrade to product-grade README and polish website hero assets by @lostmartian in #31
• feat(gate): decoupled failure modes — hard blocks vs soft warnings by @lostmartian in #32
• feat(engine): statistical envelopes — N-run baselines and topological equivalence by @lostmartian in #36
• feat(cli): agentdiff init — zero-config wizard by @lostmartian in #34
• feat(bot): /agentdiff approve — in-PR re-baselining with branded identity by @lostmartian in #35
• feat(bot): zero-setup approve — GITHUB_TOKEN default + Checks-API green flip by @lostmartian in #37
• feat(worker): hosted identity service — zero-config agentdiff[bot] tokens by @lostmartian in #38
• docs(worker): App name is AgentDiff CI (AgentDiff reserved by repo) by @lostmartian in #39
• fix(worker): point TOKEN_SERVICE_URL at the deployed workers.dev URL by @lostmartian in #40
• feat(domain): token service live at token.agentdiff.app by @lostmartian in #41
• feat(site): migrate to agentdiff.app — CNAME, canonicals, sitemap, metadata by @lostmartian in #42
• chore(site): retire GitHub Pages — Cloudflare Pages serves agentdiff.app by @lostmartian in #43
• feat(site): host brand logo at /logo.png for email signatures by @lostmartian in #44
• chore(repo): update canonical URLs for kerrshift org transfer by @lostmartian in #45
• chore(site): update repo URLs for kerrshift org transfer by @lostmartian in #46
• fix(init): generated workflows invoke the diff CLI correctly by @lostmartian in #47
• fix(gate): provenance line reflects envelope-mode thresholds by @lostmartian in #48
• chore(release): bump version to 0.5.0 by @lostmartian in #49
• docs(sync): update guides, quickstart, and README for 0.5.0 release by @lostmartian in #50
• feat(site): update landing, features, quickstart, and action pages for 0.5.0 by @lostmartian in #51

Full Changelog: v0.4.0...v0.5.0

v0.4.0

What's Changed

• feat(docs): per-guide routes with full metadata, sitemap and robots by @lostmartian in #19
• feat(seo): technical foundation - structured data, OG cards, bundle diet, FAQ by @lostmartian in #20
• feat(site): agentdiff-style 404 report with typed terminal diff by @lostmartian in #21
• feat(blog): add MD-driven blog with hero, gradients and Goodhart post by @lostmartian in #22
• feat(cli): add 'record' subcommand — capture traces from any callable (E3) by @lostmartian in #23
• feat(governance): threshold-change flagging — the Goodhart guard (G6/G7) by @lostmartian in #24
• feat(governance): gate provenance in every report (G7) by @lostmartian in #25
• feat(governance): stale-baseline warning — surface baseline age (F7) by @lostmartian in #26
• docs: shell completion install instructions (E4) by @lostmartian in #27
• feat(website): site-wide redesign — new pages, editorial features page, tabbed surfaces by @lostmartian in #28
• chore(release): bump version to 0.4.0 by @lostmartian in #29

Full Changelog: v0.3.0...v0.4.0

v0.3.0

What's Changed

• chore(contrib): add contribution rules, PR template, and agent rules by @lostmartian in #1
• docs(readme): add FAQ and explicit non-goals by @lostmartian in #2
• docs(readme): add local-first privacy section by @lostmartian in #3
• feat(cookbooks): add live LangGraph e2e recipe through openinference adapter by @lostmartian in #4
• feat(cookbooks): add live CrewAI e2e recipe through openinference adapter by @lostmartian in #5
• feat(metrics): add Recovery Step Ratio (RSR) for post-error recovery effort by @lostmartian in #6
• docs(site): sync RSR/privacy/cookbook docs, add contributing guide by @lostmartian in #7
• docs(landing): add RSR to capabilities grid, sync snippets with new gates by @lostmartian in #8
• feat(adapters): add adapter registry with entry-point plugin discovery by @lostmartian in #9
• feat(suite): add programmatic scenario runner with per-scenario gates by @lostmartian in #10
• feat(suite): add opt-in parallel scenario execution with ordered results by @lostmartian in #11
• feat(adapters): add LangGraph adapter for native state/checkpoint ingestion by @lostmartian in #12
• feat(adapters): add CrewAI adapter for native kickoff-output ingestion by @lostmartian in #13
• feat(benchmark): add cross-framework A/B benchmark mode by @lostmartian in #14
• feat(bench): add pytest-benchmark suite with make bench target by @lostmartian in #15
• docs(site): document phase 2 features - suites, A/B, plugins, benchmarks by @lostmartian in #16
• feat(website): dark mode, header redesign, spec-sheet metrics, content refresh by @lostmartian in #17
• chore(release): bump version to 0.3.0 by @lostmartian in #18

New Contributors

• @lostmartian made their first contribution in #1

Full Changelog: v0.2.2...v0.3.0

v0.2.2

Full Changelog: v0.2.1...v0.2.2

v0.2.1

Full Changelog: v0.2.0...v0.2.1

AgentDiff v0.2.0

AgentDiff v0.2.0 — first PyPI release of a developer-first trajectory regression engine for AI agents.

• Trajectory diff engine (TDI / WEI, tool-loop detection)
• CI/CD gate: pytest plugin, CLI exit codes, GitHub Action
• Adapters: generic, openinference, langfuse, langsmith, openai_agents
• Config-as-code, PR reports, baseline rotation
• Python 3.10 ISO timestamp parsing fix

Install: pip install agent-trajectory-diff
Docs: https://lostmartian.github.io/agentdiff/