Commit Graph

37 Commits

Author SHA1 Message Date
dorien-koelemeijer 826cce0257 Prompt injection mitigation: update pattern-based detection (#9198) 2026-05-14 02:19:48 +00:00
Rodolfo Olivieri 826a437d08 chore: replace lazy_static with std::sync::LazyLock (#8815)
Signed-off-by: Rodolfo Olivieri <rolivier@redhat.com>
2026-05-11 22:47:31 +00:00
dorien-koelemeijer a0fa5f5791 Update patterns for pattern-based command injection (#8546) 2026-04-30 03:21:26 +00:00
Lifei Zhou 96fa25bce8 fix: remove defaults yaml (#8854) 2026-04-27 21:57:32 +00:00
dorien-koelemeijer 2e529b7497 Add prompt injection mitigation logging back (#8518) 2026-04-14 00:44:49 +00:00
dorien-koelemeijer dc249f7b62 feat(security): add egress logging inspector (#8149) 2026-04-01 00:59:28 +00:00
dorien-koelemeijer 6d25161b19 Remove unused tool call json in logs (#8147) 2026-03-27 03:09:14 +00:00
Michael Neale 754c214df4 feat: adversarial agent for preventing leaking of info and more (#7948) 2026-03-17 06:38:45 +00:00
dorien-koelemeijer 6d6c6b3228 More logging for command injection classifier model training (#7779) 2026-03-11 13:03:18 +00:00
Adrian Cole b4546f62d9 fix: restore smart-approve mode (#7690)
Signed-off-by: Adrian Cole <adrian@tetrate.io>
2026-03-06 15:45:23 +00:00
Jack Amadeo 325bf396af Update to rmcp 1.1.0 (#7619)
Co-authored-by: Alex Hancock <alexhancock@block.xyz>
2026-03-06 02:40:52 +00:00
Bradley Axen ced5c1b108 feat: simplify developer extension (#7466)
Co-authored-by: Alex Hancock <alexhancock@block.xyz>
2026-02-26 11:30:36 +00:00
dorien-koelemeijer e32720fcfb Fix: Minor logging uplift for debugging of prompt injection mitigation (#7195) 2026-02-17 23:23:50 +00:00
Adrian Cole 4a5848ea76 fix(otel): use monotonic_counter prefix and support temporality env var (#7234)
Signed-off-by: Adrian Cole <adrian@tetrate.io>
2026-02-16 21:45:15 +00:00
dorien-koelemeijer 5b15593e0a Fix: Only send command content to command injection classifier (excluding part of tool call dict) (#7082) 2026-02-09 22:09:12 +00:00
dorien-koelemeijer a3be331fe6 Fix: Small update UI settings prompt injection (#6830) 2026-01-30 12:57:55 +07:00
dorien-koelemeijer 66e4a1a43c Different approach to determining final confidence level of prompt injection evaluation outcomes (#6729) 2026-01-29 07:41:54 +07:00
dorien-koelemeijer e9024b7c19 Logging uplift for model training purposes (command injection model) [Small change] (#6330) 2026-01-27 12:25:11 +07:00
dorien-koelemeijer 3dae12765f BERT-based command injection detection in tool calls (#6599) 2026-01-27 10:46:02 +07:00
Alex Hancock 8233f0afac chore: upgrade to rmcp 0.14.0 (#6674) 2026-01-23 14:52:43 -05:00
Alex Hancock 4191826569 chore: upgrade rmcp (#6516) 2026-01-16 12:09:51 -05:00
Adrian Cole 7d4a6bd8ff feat(goose-acp): enable parallel sessions with isolated agent state (#6392)
Signed-off-by: Adrian Cole <adrian@tetrate.io>
2026-01-14 14:19:17 -08:00
dorien-koelemeijer 8a9b090aa8 Fix: Small update in how ML-based prompt injection determines final result (#6439) 2026-01-13 08:20:40 +10:00
dorien-koelemeijer 3d415fc1ea Update confidence levels prompt injection detection to reduce false positive rates (#6390)
Signed-off-by: Dorien Koelemeijer <dkoelemeijer@squareup.com>
2026-01-08 15:51:53 +10:00
dorien-koelemeijer 9dc548ee2f Add ML-based prompt injection detection (#5623) 2026-01-08 11:55:59 +10:00
Douwe Osinga 85aee9c5b7 MCP Apps Plumbing (#6286)
Co-authored-by: Douwe Osinga <douwe@squareup.com>
2026-01-05 10:34:52 -05:00
Rabi Mishra fc8a4b4adf fix(google): Fix 400 Bad Request error with Gemini 3 thought signatures (#6035)
Signed-off-by: rabi <ramishra@redhat.com>
2025-12-17 08:38:46 -05:00
dorien-koelemeijer 469f99f9c9 Update vars to be capitalised to be in line with other variables in config file (#6085) 2025-12-11 16:12:45 -08:00
Salvatore Testa cfdf01567d fix: support Gemini 3's thought signatures (#5806)
Signed-off-by: Salvatore Testa <sal@withpersona.com>
2025-11-20 16:28:27 +11:00
dorien-koelemeijer 387d76c5ff Update prompt injection detection metrics (due to errors exporting to datadog) (#5692) 2025-11-14 08:40:29 +10:00
dorien-koelemeijer 786e7084b7 Set up Datadog metrics for prompt injection detection (#5385) 2025-10-28 09:11:34 +10:00
dorien-koelemeijer e6a5692656 Feat: Add prompt injection detection settings UI + update logging (#4651) 2025-10-03 09:01:37 -04:00
Alex Hancock cb0eca9966 alexhancock/mcp-crate-cleanup (#4885) 2025-09-29 17:00:17 -04:00
Alex Hancock b9ba8dca29 Revert "Internal MCP Crate Cleanup (#4800)" (#4883) 2025-09-29 14:21:30 -04:00
Alex Hancock 146cf3185a Internal MCP Crate Cleanup (#4800) 2025-09-29 10:39:43 -04:00
Jack Amadeo 757ceb6109 chore: turn clippy on for test code (#4817) 2025-09-26 00:06:07 -04:00
dorien-koelemeijer 916ba902dc Prompt injection detection (simplified - only pattern matching) (#4237)
Signed-off-by: Dorien Koelemeijer <dkoelemeijer@squareup.com>

merging as looks good, lets keep an eye on it.
2025-09-10 10:41:30 +10:00