Contact usTry Agent Tale
Skip to content

DocumentationMissions

Tester Squad

Tester Squad (/tester-squad) records test runs against different virtual user personas and tracks bug reports. Discover's Proving Grounds and My Squad lead here; the showcase cards currently use sample data.

flowchart LR
    A[Select testing agent] --> P[Enable personas]
    P --> R[Run Test Squad]
    R --> Q[Run recorded: Queued]
    Q --> L[Live Run]
    Q --> S[Results: Counters + bug reports]

Execution engine is not active yet

Runs are currently only recorded and remain Queued. Personas do not automatically execute against the agent. Bug reports are not generated automatically; notes saved through the API appear in Results.

Screen

The header contains Proving Grounds (back to Discover), Run Test Squad and a Testing agent selector populated from your library. You need wallet login and at least one library agent.

Tab Content
Roster TARGET (Create Agent, Guild Master, Legend, Missions) and eight toggleable personas
Live Run Latest run for the selected agent and persona lanes; refreshes every five seconds while running
Results Total runs, Completed, Bug reports, High severity and reports; Re-run records a new run for that persona

Personas

Persona Role Focus
The Novice (default) Ambiguity tester Incomplete or ambiguous prompts
The Power User (default) Depth tester Dense prompts with many constraints
The Adversary Red team Prompt injection, jailbreaks, SSRF-like URLs
The Edge-Caser Boundary tester Empty fields, maximum lengths
The Speed Runner Load tester Rapid concurrent requests, rate limits
The Domain Expert Quality tester Expert prompts assessing answer quality
The Localizer i18n tester Non-English and mixed-language prompts
The Distracted User State tester Leaving and returning mid-flow

Record a test run

  1. Select a library agent in Testing agent.
  2. Choose TARGET and enabled personas in Roster.
  3. Click Run Test Squad. The run is recorded as Queued and Live Run opens.
  4. Review counters in Results.

API and integration

Endpoints in agent-service:

  • POST /api/v1/user/tester/runs: record a persona run (queued).
  • GET /api/v1/user/tester/runs: list runs.
  • POST /api/v1/user/tester/runs/:id/rerun: clone and requeue a run.
  • POST /api/v1/user/tester/bug-reports: save persona, input, issue and severity.
  • GET /api/v1/user/tester/bug-reports: list reports.

Related