AegisRT
Evaluate
Investigate
Probes
Compare
Copy Run ID
Compare Prev
JSON
HTML
SARIF
No run selected
Select a run from the sidebar to evaluate or investigate it
Run
aegisrt run -c aegisrt.yaml
to get started
Loading run...
Target
Judge
Attacker
Investigate
findings
All
Failures
Passes
All triage
Untriaged
All probes
25
50
100
Prev
Next
Missing Metadata
All
/
Result
Probe
Input (attack prompt)
Output (model response)
Grading
Score:
Inspect trace
Execution Context
Target Model
type
Judge Model
llm-as-judge
Attacker Model
adversarial generation
Endpoint
run-level target
Kill Chain
1
Seed
gen:
▶
2
Transforms
none
converter chain / prompt mutation
▶
3
Prompt
▶
4
Response
▶
5
Detect
▶
Detection Evidence
Matched:
Policy:
Score:
/ threshold:
Remediation
•
No results match filters.
Investigate
Security Findings
Review exploit paths, assign ownership, and move findings through triage without leaving the run context.
Run
Timestamp
Active
/ Closed
Open Findings
not fixed or false positive
Needs Review
new or unassigned triage
Critical + High
highest-impact findings
Closed
fixed or false positive
Missing Metadata
Attack Sessions
Grouped multi-step traces, conversation attacks, and future agent/tool chains.
active session views
Sessions
grouped attack chains
Multi-step
sessions with more than one hop
Tool Reach
sessions invoking tools
Boundary Crossings
sessions with trust boundaries
session
Attack
open
findings
steps
Participants
Tools / Memory
Boundaries
Representative
Filter findings
Open trace
Open findings
All findings
Needs review
Acknowledged
Accepted risk
Closed
Fixed
False positive
All severity
Critical
High
Medium
Low
All probes
All sessions
12
24
48
Prev
Next
Back to Evaluate
No failing findings in this run.
No findings match the current filters.
Attack Prompt
Observed Impact
Owner
Detections
Last seen
Note
Score
Confidence
Open Trace
Security Probes
Suites
Compare Runs
Need at least two runs.
Baseline
Select...
Compare
Select...
Regressions
Resolved
Persistent
Stable
Skipped
Baseline
Compare
Attack Trace
case
confidence
score
Close
Detector Hits
Generator
Runtime
Model
Previous Failure
Next Failure
Compare Previous Run
Copy Finding Key
Missing Metadata
Finding Triage
Finding
Status
Untriaged
Owner
Seen
First:
Last:
Note
Updated
Attack Timeline
Intent
No pre-transform seed recorded
Delivered Prompt
Executed prompt wrapper
Model Response
Conversation Trace
Detector Outcomes
No detector evidence captured
Judge Evaluations
No judge trace captured
Context Metadata
No additional case metadata
Response Metadata
No additional response metadata
Remediation
•
Evidence Snapshot
Raw Trace
Expand to inspect the full stored trace payload.
Delete Eval
Remove this saved run?
This removes the run from the dashboard list and compare views.
Run
Timestamp
Current behavior deletes the stored run record. Artifact files under
.aegisrt/runs/
are not removed by this UI action.
Cancel