Tag: LLM agents

Showing 1-3 of 3

Oct 01, 20268 min readAgent PlaybookIntermediate240 min build

ProvenanceGuard: Preventing Cross‑Source Conflation in MCP LLM Agents

Adds a source-aware verification layer for MCP agents that checks claims against the named tool output, returns support_score and match_span, and flags cross-source conflation.

MCP ProvenanceGuard source-aware verification LLM agents provenance

+2 more

factuality cross-source conflation

NewsUnited Kingdom

Open

Mar 25, 20265 min readAgent PlaybookIntermediate90 min build

Litmus: Record and deterministically replay full LLM agent executions for debugging and CI validation

Litmus records complete LLM agent executions (prompts, tool calls, outputs) so teams can deterministically replay failures, inject faults, and gate regressions in CI.

litmus llm-agents observability testing ci

+3 more

replay fault-injection reliability

Model BreakdownsUnited States

Open

Feb 06, 20268 min readAgent PlaybookIntermediate5 min build

Active Epistemic Control: grounded-fact versus belief stores and SQ-BCP gating for verified planning

Summarizes AEC: split grounded facts for commitments from a belief store for pruning, query vs simulate by uncertainty, and gate commitments with SQ-BCP—reduces replanning.

active-epistemic-control planning partial-observability world-models verification

+3 more

agents llm-agents alfworld