Limbo v2026.4.15

• chore: retrigger CI (12e0751)
• chore: switch release workflow to version-from-package-json model (f0d0b70)
• fix: promote staging to main (#252) (c276c2b)
• chore: migrate CI/CD back to GitHub with GitLab as dormant fallback (#253) (61f79df)
• refactor: remove trivial JSDoc and historical comments (825a001)
• refactor: consolidate shared MCP utilities and strengthen types (3021167)
• fix: clean up legacy references and unnecessary defensive code (33e60dd)
• fix: cloud activate before config + release v2026.5.0 (ce6f73c)
• fix: serve install script from get.heylimbo.com instead of GitHub raw (ebf9b24)
• chore: release v2026.4.14 (e674a3b)

Limbo v1.30.0

• chore: release v1.30.0 [skip ci] (76679f8)
• feat: promote staging to main (#239) (e6ab055)

Limbo v1.29.0

• chore: release v1.29.0 [skip ci] (c854c9e)
• feat: add agent-first GitHub issue templates and label system (979fef9)

Limbo v1.28.1

• chore: release v1.28.1 [skip ci] (030200b)
• fix: require workspace_read on USER.md before first response (83327b4)
• fix: document cron/schedule tools in TOOLS.md for non-Claude models (db8a620)

Limbo v1.28.0

• chore: release v1.28.0 [skip ci] (938b354)
• feat: overhaul setup wizard for non-technical users (27aba90)
• chore(evals): add cron eval run history (includes overload failures + opus pass) (326ca06)
• fix: detect workspace ownership issues and add recurring cron eval (04ebe18)

Limbo v1.27.0

• chore: release v1.27.0 [skip ci] (db01b68)
• fix(evals): remove named data volume and fix resetVault bind mount (83ec4d6)
• feat: add workspace_read/write tools for agent self-awareness (#201) (98124e9)
• feat(evals): add manual Telegram eval tests for PDF + audio (#196) (40e6d09)
• test(evals): add subdirectory seed note and search eval case for #203 (74aab51)
• docs: production-grade README with banner (#209) (cb258fa)
• feat(evals): per-case baselines for independent eval comparison (#208) (a42f830)
• chore: bump ZeroClaw v0.6.3 → v0.6.5-codex-parity-custom (#207) (fc968ba)
• test(evals): add workspace_read/write eval cases (#205) (23f64d1)
• fix: sync FTS index with filesystem on startup (#203) (101fa49)
• feat: improve eval transcript and user profile handling (#197) (f263084)
• fix: USER.md template envsubst defaults (#200) (5bee290)
• feat: replace zeroclaw-state named volume with bind mount (#198) (32c1ebf)
• chore(evals): promote staging baseline — 94.0% (FTS5 + v0.6.3 + web search enabled) (a5975de)
• fix(evals): add brave_api_key secret to eval compose + fix search-implicit regex (df9c78d)
• fix(dashboard): add horizontal scroll to history chart (86cd11d)
• Set Codex reasoning effort default to medium (#194) (9c9e78e)
• feat: replace regex search with SQLite FTS5 (#188) (295c504)
• chore(evals): promote FTS5 + v0.6.3 run as baseline (92.5%) (65e816b)
• chore(evals): add FTS5 + v0.6.3 eval run (92.5%) (386b687)
• chore(evals): add Sonnet baseline (76.1%) + ZeroClaw v0.6.3 eval (91.0%) (127a77b)
• chore: bump ZeroClaw v0.5.3 → v0.6.3 (#182) (b9d78e8)
• feat: make eval dashboard model-aware (#195) (e6f9ed3)
• feat(evals): track vault_search execution time, show in Speed dashboard (3b773f4)
• feat(dashboard): redesign with accuracy/speed split, inline details, cleaner UI (031d578)
• chore(evals): update baseline — 95.5% with 20 cases, 6 tools, latency tracking (603aff9)
• chore: add docker log rotation to all compose files (#193) (e1a7147)
• fix(evals): fix (?i) regex double-flag bug + relax overly strict assertions (8c3c93a)
• feat: add vault file storage (images, PDFs, documents) (#185) (3c36880)
• feat: eval speed benchmarks, cron cleanup, latency tracking + dashboard (#191) (a72c63f)
• feat(evals): add local eval dashboard for visualizing test results (#190) (249a77f)
• feat(evals): add file tool eval cases + test asset (#187) (88344d3)
• feat: restructure personality files — move processing flow to TOOLS.md (#186) (710d9cb)
• feat: enable rag-pdf for native PDF text extraction (#179) (e600ffd)
• fix: align auth-profiles.json with ZeroClaw's actual Rust schema (#183) (4c3f47a)
• chore(evals): add initial baseline — 3 runs, 14 cases, ~63% avg pass rate (eb96ccf)
• feat: add difficulty tiers (easy/medium/hard) to eval cases (#184) (4b30ac4)
• fix: eval framework bugs + cron assertion support (#181) (155aa8b)
• feat: add eval framework for end-to-end capability testing (#180) (acfb86f)
• docs: remove provider recommendations from README (9103966)

Limbo v1.26.0

• chore: release v1.26.0 [skip ci] (d0f003e)
• feat: promote staging to main (#189) (2e99e51)

Limbo v1.25.0

• chore: release v1.25.0 [skip ci] (89f3f1c)
• feat: promote staging to main (#177) (4fb12e9)

Limbo v1.24.9

• chore: release v1.24.9 [skip ci] (8b19515)
• docs: add shields.io badges to README (58988c8)
• docs: rewrite README — VPS-first install, cleaner structure (37e6baa)
• docs: add server install one-liner to README (47d8359)
• chore: rewrite install.sh — lean bootstrap, CLI does the rest (aa1b16f)
• fix: export OAuth tokens as ANTHROPIC_OAUTH_TOKEN, not API_KEY (78125f4)

Limbo v1.24.8

• chore: release v1.24.8 [skip ci] (8899eab)
• fix: subscription auth — secrets for Anthropic, auth-profiles for OpenAI (b687d02)