🚀 Highlights
- Tracecat is now an MCP client with OAuth support for fine-grained scopes
- Agentic AI Slackbot -> works with Tracecat tools like any AI agent!
- Limit max number of tool calls an agent is allowed to make
💥 Breaking changes
- Always return list for deduplicate (#1467)
- Remove static inputs from workflows (#1427)
- Move runbooks into alpha (this is now disabled by default)
⚠️ Deprecation notice
All llm.xyz actions (llm.openai.chat, llm.anthropic.call, Bedrock, Gemini etc) will be replaced with the single ai.agent action. We're doing this to have one centralized place to control all LLM model logic. This will make it easier for us to add:
- Security features (e.g. approvals, agent specific audit trails)
- UI/UX improvements (e.g. automated model selection dropdown menu)
We will remove llm.xyz actions in the next 2-3 releases (i.e. before the end of October).
🧩 Integrations
- AI agent action (#1449)
- MCP actions (Linear, GitHub, Notion, Sentry, RunReveal) (#1458)
- Agentic AI Slackbot-as-an-action (#1461)
- Add wait UDF (#1474)
- Add GitHub OAuth authorization code flow (#1460)
- Add domain and email leaks research (leakcheck) (#1455)
- Codex and Claude code integrations (#1459)
- Gophish integration (#1444)
- Langufse integration (#1442)
- Add entities UDFs (#1439)
- Add Exa templates (#1432)
🚀 Performance improvements
- Support max tool calls and max requests in agents (#1462)
✨ Enhancements
- MCP OAuth providers (#1447)
- Add UI for offset in workflow schedules (#1415)
- Add cron schedule support and improve schedule UI (#1446)
- Add sortable column header component for data tables (#1443)
- Add user ID (temporal namespace) and release (app version) to Sentry event (#1416)
- Add bulk delete functionality in case table (#1448)
🐞 Bug fixes
- Complex secret expression loading (#1423)
- Improve git ssh url validation (#1470)
- Fix regression in controls header icon controls (#1457)
- Add Exa tools integration and fix API endpoints (#1441)
📖 Documentation
- Add Kandji pagination examples to tests and docs (#1424)
- Update cheatsheet to include if else (#1422)
- Update Okta API doc links (#1453)
📦 Build system
↪️ Refactor
- Overhaul LLM related logic into prompts submodules (#1454)
- Move agent code from integrations into main pkg (#1452)
🚧 Alpha features
- Add runbooks feature flag (#1468)
- Improve runbook execution and table metadata functionality (#1440)
- Add runbook alias support and improve chat interface integration (#1438)
- Add runbook chat interface (#1435)
- Add alias support to prompts/runbooks (#1433)
- Grid style runbooks UI (#1429)
Thank you to all our contributors for making this release possible!
@bishopdotdev, @daryllimyt, @s-anvar, @topher-lo and @y0no