kortix-ai/suna v0.13.46

v0.13.46 — An agent permission fix, refresh-token replay protection, prompt authorship, and agents that can ask people (behind a flag).

on GitHub

An agent permission fix, refresh-token replay protection, prompt authorship, and agents that can ask people (behind a flag).

New

• Every prompt records who wrote it, and the transcript shows the author.
• Agents can ask a person a question or start a group conversation with people, behind the human_messaging project flag (off by default).

Security

• An agent whose policy denies bash or edit can no longer open a shell through the terminal tools or write files through the memory tool.
• A refresh token cannot be replayed, and logging out revokes the session.

Fixed

• Chat stays available while a connector waits for approval.
• The advanced panel preference keeps the panels you open.
• The cost explorer shows a loading state until usage arrives.
• Mobile renders model reasoning as a thinking row.
• Deleting an account no longer reports its sandboxes as lost work.
• A conversation with people is named at a word boundary.

What's Changed

• refactor(web): consolidate square dot matrix scaffolds (KRTX-655) by @agent-kortix in #8158
• fix(web): show cost explorer loading state until usage arrives (KRTX-791) by @agent-kortix in #8351
• fix(api): park test no longer leaks a session-worker process per run by @markokraemer in #8557
• refactor(web): extract sidebar panel components (KRTX-652) by @agent-kortix in #8556
• fix(mobile): render reasoning as a thinking row (KRTX-896) by @agent-kortix in #8540
• fix(web): keep chat available during connector approval (KRTX-911) by @agent-kortix in #8562
• fix(web): preserve advanced-preference panel opens (KRTX-398) by @agent-kortix in #8162
• feat(sessions): agents message people — asks, group chats, message authors (human_messaging) by @markokraemer in #8560
• fix(web): repair main after 0e82ff4 (KRTX-398) by @agent-kortix in #8564
• fix(tests): mode-gate test sets the panel mode through patchPreferences (#8162) by @markokraemer in #8566
• chore(release): staging VERSION → 0.13.46 [skip ci] by @github-actions[bot] in #8568
• chore(release): VERSION → 0.13.46 [skip ci] by @github-actions[bot] in #8569
• feat(brand): one agentic brand kit — router, verbal + visual rules, visual-system.json generates the tokens — and every surface on it by @markokraemer in #8567
• fix(sessions): name a conversation with people at a word boundary by @markokraemer in #8570
• feat(brand): close fresh-agent QA round 3 gaps; deck captions follow claims.md by @markokraemer in #8571
• fix(docs): quote the docs index description; test docs frontmatter YAML by @markokraemer in #8572
• docs(learnings): quote docs frontmatter values that contain a colon by @markokraemer in #8573
• Promote main da2c3cb to staging (v0.13.46) by @markokraemer in #8565
• refactor(composer): unify navigation and context usage (KRTX-899) by @agent-kortix in #8544
• fix(auth): prevent refresh replay and revoke logout session (KRTX-910) by @agent-kortix in #8559
• refactor(email): share consumer domain classification (KRTX-918) by @agent-kortix in #8576
• fix(api): Kortix-initiated sandbox removal is not a runtime.lost error by @markokraemer in #8580
• fix(tests): repair main after f605a96 (KRTX-910) by @agent-kortix in #8579
• fix(agents): a denied bash or edit also denies pty_* and memory (RUN-10) by @markokraemer in #8582
• Promote main 5f7b0bd to staging (v0.13.46, security fix) by @markokraemer in #8583
• v0.13.46 — An agent permission fix, refresh-token replay protection, prompt authorship, and agents that can ask people (behind a flag). by @github-actions[bot] in #8574

Full Changelog: v0.13.45...v0.13.46