v0.3.0

What's Changed

• chore: add GitHub issue/PR templates and Codex prompts by @aleksei-s-popov in #1
• Add L3 automation repository skeleton and L3 flow docs by @aleksei-s-popov in #2
• Add repository documentation skeleton for L3 automation and Codex integration by @aleksei-s-popov in #3
• Add repository skeleton: AGENTS.md and documentation placeholders for L3 automation by @aleksei-s-popov in #4
• Add CONTRIBUTING.md with L3 one-issue contribution guidelines (Closes #6) by @aleksei-s-popov in #34
• Add initial security policy by @aleksei-s-popov in #35
• Add focused pull request template by @aleksei-s-popov in #36
• Add AI-ready issue template guidance by @aleksei-s-popov in #37
• Add CODEOWNERS covering repo risk boundaries (closes #10) by @aleksei-s-popov in #38
• Add AI- and human-scoped issue templates; refine task and PR templates by @aleksei-s-popov in #40
• Document initial main branch ruleset checklist (closes #11) by @aleksei-s-popov in #41
• Add initial CI workflow lanes by @aleksei-s-popov in #42
• Add advisory PR risk review Codex prompt by @aleksei-s-popov in #43
• Improve issue and PR reference doc prompts by @aleksei-s-popov in #44
• Add repo automation ADR by @aleksei-s-popov in #45
• Add L3 automation boundary ADR by @aleksei-s-popov in #46
• fix(model): parse numeric-string confidence in extraction validation by @aleksei-s-popov in #47
• fix(model): cap generation-role output tokens to stop context blowout by @aleksei-s-popov in #48
• fix(model): convert reasoning_effort to atom before forwarding by @aleksei-s-popov in #49
• fix(console): make subject verdict buttons self-explanatory by @aleksei-s-popov in #50
• Add unattended governance consent for declared-auto accounts and deployments by @aleksei-s-popov in #51
• fix(governance): surface decision outcomes, add select-all, resolve conflicts inline by @aleksei-s-popov in #52
• fix(jobs): restore Oban peer leadership so retryable jobs stage again by @aleksei-s-popov in #53
• fix(pipeline): take model calls out of the Account transaction by @aleksei-s-popov in #54
• fix(retrieval): take model calls out of the account transaction in both rebuild lanes by @aleksei-s-popov in #58
• fix(database): provision a restricted role so row-level security actually enforces by @aleksei-s-popov in #57
• fix(governance): stop crashing on personal-sensitivity scope-subject knowledge by @aleksei-s-popov in #59
• fix(operations): scope edge metering to its Account so RLS admits it by @aleksei-s-popov in #61
• fix(pipeline): declare the Account so background jobs can see their own run by @aleksei-s-popov in #63
• fix(model): name why an incomplete provider response carried nothing by @aleksei-s-popov in #64
• docs: make repository prose concise by @aleksei-s-popov in #75
• Make ingest strictly asynchronous by @aleksei-s-popov in #84
• Preserve grounded inconclusive ask answers by @aleksei-s-popov in #81
• test(retrieval): rank against a distractor corpus by @aleksei-s-popov in #80
• docs: keep entity cache private by @aleksei-s-popov in #77
• fix(retrieval): make a scope with statements and no embeddings observable by @aleksei-s-popov in #78
• fix(retrieval): report strategies that ran and found nothing (FR-API-28/29) by @aleksei-s-popov in #76
• fix(retrieval): stop lexical search from ANDing every query term by @aleksei-s-popov in #79
• Keep provisional statements out of shared context projections by @aleksei-s-popov in #82
• Expose privacy-safe entity resolution diagnostics by @aleksei-s-popov in #83
• Add scope-bounded entity cards by @aleksei-s-popov in #85
• Add MCP tool workbench to console by @aleksei-s-popov in #86
• ci: publish tagged release artifacts by @aleksei-s-popov in #87

New Contributors

• @aleksei-s-popov made their first contribution in #1

Full Changelog: https://github.com/cartularyhq/cartulary/commits/v0.3.0