Search results

red team: end-to-end behavior test for GOAT dynamic technique selection

Status: Open.

#338 In langwatch/scenario;

· Aryansharma28 opened on Apr 14, 2026

red team: regression test for KeyError helper in _generate_attack_plan

Status: Open.

#337 In langwatch/scenario;

· Aryansharma28 opened on Apr 14, 2026

red team: success_score=9 default may be too strict; needs benchmark

Status: Open.

#336 In langwatch/scenario;

· Aryansharma28 opened on Apr 14, 2026

red team: Python ↔ TypeScript prompt-text parity test

Status: Open.

#335 In langwatch/scenario;

· Aryansharma28 opened on Apr 14, 2026

red team: GOAT metaprompt + injection_probability double-encoding

Status: Open.

#334 In langwatch/scenario;

· Aryansharma28 opened on Apr 14, 2026

red team: edge cases — empty techniques list, empty metaprompt_plan, scorer failure cascade

Status: Open.

#333 In langwatch/scenario;

· Aryansharma28 opened on Apr 14, 2026

red team: phase ontology divergence breaks cross-strategy analytics

Status: Open.

#332 In langwatch/scenario;

· Aryansharma28 opened on Apr 14, 2026

red team: _MAX_BACKTRACKS = 10 is hardcoded; should scale with total_turns

Status: Open.

#331 In langwatch/scenario;

· Aryansharma28 opened on Apr 14, 2026

red team: capture which GOAT technique the attacker chose per turn (telemetry)

Status: Open.

#330 In langwatch/scenario;

· Aryansharma28 opened on Apr 14, 2026

red team: shared mutable state on RedTeamAgent makes parallel scenario.run() unsafe

Status: Open.

#329 In langwatch/scenario;

· Aryansharma28 opened on Apr 14, 2026

red team: strategies should own their default metaprompt template

Status: Open.

#328 In langwatch/scenario;

· Aryansharma28 opened on Apr 14, 2026

red team: _attack_plan cache survives across scenarios with different descriptions

Status: Open.

#327 In langwatch/scenario;

· Aryansharma28 opened on Apr 14, 2026