Search results

Generalizable agent_factory→grader context channel (retire agent_state as the per-rollout side-channel)

Status: Open.

#310 In letta-ai/letta-evals;

· devanshrj opened on Jun 30, 2026

Runner: agent_state not provided to reward composers (retrieval is grader-driven only)

Status: Open.

#301 In letta-ai/letta-evals;

· devanshrj opened on Jun 27, 2026

Add a pre-grade hook to Runner.run_sample (seam between generation and grading)

Status: Open.

#294 In letta-ai/letta-evals;

· devanshrj opened on Jun 24, 2026

Support halt_on_tool for letta_code targets

Status: Open.

#286 In letta-ai/letta-evals;

· devanshrj opened on Jun 17, 2026

Fix multiturn per-turn grading E2E for letta_code target

Status: Open.

#285 In letta-ai/letta-evals;

· devanshrj opened on Jun 17, 2026

APIConnectionError when running letta-evals with letta_agent target

Status: Open.

#158 In letta-ai/letta-evals;

· jasonlarkin opened on Jan 14, 2026

Can xAI models like Grok 4 be added to Context-Bench?

Status: Open.

#131 In letta-ai/letta-evals;

· exigentix-dev opened on Nov 17, 2025