fix(ai-red-teaming): record custom-target attacks via assessment - #162
Merged
Merged
Conversation
rdheekonda
changed the base branch from
fix/agentic-goal-category-recording
to
main
September 30, 2026 00:48
multistep_tool_attack, agentvigil_attack and eva_attack ran their custom loops and returned a report without going through the assessment, so they emitted no study/trial spans and materialized no findings even on success. Call the new assessment.record_attack_result(...) after each so the outcome is traced and mapped to the right OWASP-Agentic category (data exfil / goal hijacking). Requires the SDK method from dreadnode-tiger#2668.
rdheekonda
force-pushed
the
fix/agentic-custom-target-recording
branch
from
September 30, 2026 00:53
dbdb6b7 to
1c2b800
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Capability wiring for bug #1 (custom-target/multistep attacks recorded 0 spans / 0 findings).
multistep_tool_attack,agentvigil_attack, andeva_attackran their custom loops and returned a report without going through the assessment, so nothing was traced or materialized even on success.Each generated workflow now calls the new SDK method
assessment.record_attack_result(...)right after the attack, so the outcome emits a study + trial span linked to the assessment and materializes a finding with the right OWASP-Agentic category:multistep_tool_attack->agentic_data_exfilagentvigil_attack/eva_attack->agentic_goal_hijackingDependency / ordering
Requires the SDK method from dreadnode-tiger#2668 at runtime. That SDK PR must merge + release before this capability version is published, or the generated scripts will
AttributeErroronrecord_attack_result.Stacked on
fix/agentic-goal-category-recording(PR #161) to avoid acapability.yamlversion-bump conflict; base retargets tomainonce #161 merges.Validation
compile()passes with the new calls).pre-commit(ruff, ruff-format, yaml, gitleaks): pass.just validate->ai-red-teaming@1.17.5OK.1.17.4 -> 1.17.5.Test plan
study:/trialspan in ClickHouse for the assessment and a materialized finding in the UI (correct category + score). Honeytoken flows (TUI tools) are out of scope here - tracked separately.