github HolmesGPT/holmesgpt 0.40.0-alpha

pre-release3 hours ago

What's Changed

  • ROB-878 remove stale Sonnet 4.0/4.5 model recommendations from docs by @Sheeproid in #2378
  • Remove experimental Azure SQL toolset by @moshemorad in #2377
  • ROB-893 Fix command injection vulnerability in Kubernetes toolset by @naomi-robusta in #2382
  • Remove MariaDB MCP integration, consolidate to native MariaDB datasource by @moshemorad in #2389
  • Fix command injection in slab/kubevela/inspektor_gadget/aks toolsets (ROB-893 follow-up) by @naomi-robusta in #2386
  • Fix SSRF in internet/fetch_webpage and connectivity_check toolsets (ROB-896) by @naomi-robusta in #2385
  • Harden kubectl-run toolset: run commands without a host shell by @Sheeproid in #2391
  • Add LLM eval for AKS --query command injection by @naomi-robusta in #2393
  • bash toolset: validate command arguments & redirections, not just the prefix by @Sheeproid in #2388
  • Sign bash session-approval prefixes to prevent forgery attacks by @moshemorad in #2394
  • Update GitLab MCP to v2.1.46 and use whoami health check by @moshemorad in #2398
  • ROB-424 ROB-425 ROB-665: fix compaction on Bedrock Converse gateways (tools-attached summarization, user-role summary) by @moshemorad in #2214
  • Improve shell injection security in GitHub Actions workflows by @moshemorad in #2399
  • Fix SSRF via unrestricted redirect following in the http toolset (ROB-915) by @naomi-robusta in #2392
  • ROB-857 Surface skill titles for UI-added skills in fetch_skill results by @avshyz in #2380
  • ROB-1177 Only inject __robusta_user_approved into relay-routed remote MCP calls by @Avi-Robusta in #2408
  • Fix NetworkPolicy selectors to match Holmes pod labels by @moshemorad in #2410
  • Require an allowlist before tcp_check probes private networks (ROB-1114) by @naomi-robusta in #2402
  • Add diagnostic-pod target policy to prevent SSRF/exfiltration (ROB-910) by @naomi-robusta in #2412
  • ROB-900: wire kubernetes-remediation MCP bearer-token auth in Helm chart by @arikalon1 in #2396
  • ROB-831 Add personal skills and a per-account name-collision hierarchy by @alonelish in #2350
  • ROB-1158 Gracefully retire in-flight conversations on shutdown by @naomi-robusta in #2404
  • ROB-795: retry the startup Robusta models fetch by @avshyz in #2390
  • ROB-869 Cap oversized error tool results: 500-char gist inline, full content spilled to file by @Avi-Robusta in #2409
  • ROB-835 Prune the custom-skills mirror when the last skill is deleted by @alonelish in #2353

New Contributors

Full Changelog: 0.39.0...0.40.0-alpha

Don't miss a new holmesgpt release

NewReleases is sending notifications on new releases.