# Tokengeist: Multi-Turn Attribution Tracing in Agentic Conversations

arXiv: https://arxiv.org/abs/2607.22610 · HTML: https://mlsanity.com/paper/2607.22610/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use.
- Builds on 20 of its 36 references that are in the map.
- First posted on arXiv 14 Jun 2026.

## Placed in

- LLM agents and tool use / [General LLM agent frameworks and benchmarks](https://mlsanity.com/agents/general-llm-agent-frameworks-and-benchmarks/)

## Builds on (most cited first, 20 of 20)

- [A Unified Approach to Interpreting Model Predictions](https://mlsanity.com/paper/1705.07874/) (2017)
- ["Why Should I Trust You?": Explaining the Predictions of Any Classifier](https://mlsanity.com/paper/1602.04938/) (2016)
- [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023)
- [Efficient Streaming Language Models with Attention Sinks](https://mlsanity.com/paper/2309.17453/) (2023)
- [τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains](https://mlsanity.com/paper/2406.12045/) (2024)
- [Enabling Large Language Models to Generate Text with Citations](https://mlsanity.com/paper/2305.14627/) (2023)
- [LLMs Get Lost In Multi-Turn Conversation](https://mlsanity.com/paper/2505.06120/) (2025)
- [MT-Bench-101: A Fine-Grained Benchmark for Evaluating Large Language Models in Multi-Turn Dialogues](https://mlsanity.com/paper/2402.14762/) (2024)
- [ContextCite: Attributing Model Generation to Context](https://mlsanity.com/paper/2409.00729/) (2024)
- [Interpreting Language Models with Contrastive Explanations](https://mlsanity.com/paper/2202.10419/) (2022)
- [Summary of a Haystack: A Challenge to Long-Context LLMs and RAG Systems](https://mlsanity.com/paper/2407.01370/) (2024)
- [AttributionBench: How Hard is Automatic Attribution Evaluation?](https://mlsanity.com/paper/2402.15089/) (2024)
- [Towards Verifiable Generation: A Benchmark for Knowledge-aware Language Model Attribution](https://mlsanity.com/paper/2310.05634/) (2023)
- [CONFETTI: Conversational Function-Calling Evaluation Through Turn-Level Interactions](https://mlsanity.com/paper/2506.01859/) (2025)
- [Verifiable Generation with Subsentence-Level Fine-Grained Citations](https://mlsanity.com/paper/2406.06125/) (2024)
- [LAQuer: Localized Attribution Queries in Content-grounded Generation](https://mlsanity.com/paper/2506.01187/) (2025)
- [AttnTrace: Contextual Attribution of Prompt Injection and Knowledge Corruption](https://mlsanity.com/paper/2508.03793/) (2025)
- [Follow the Flow: Fine-grained Flowchart Attribution with Neurosymbolic Agents](https://mlsanity.com/paper/2506.01344/) (2025)
- [Switchcraft: AI Model Router for Agentic Tool Calling](https://mlsanity.com/paper/2605.07112/) (2026)
- [Are Finer Citations Always Better? Rethinking Granularity for Attributed Generation](https://mlsanity.com/paper/2604.01432/) (2026)

## Abstract (arXiv)

When a language model produces a response in a multi-turn conversation, which tokens from prior turns shaped that answer, and how did those dependencies propagate across prior turns? Existing context attribution methods process the full context in a single pass, recovering surface-level dependencies but missing the layered, non-linear structure of real-world dialogues and multi-step reasoning tasks. We introduce multi-turn context attribution (MTCA): given a target span in a model response, the task of tracing attribution backward across turns to identify not only which prior turns were directly relevant, but also how those turns themselves depended on earlier context. We propose Tokengeist, an attribution-method-agnostic and scalable framework that recovers full dependency paths by casting attribution as a recursive traversal of a directed acyclic graph (DAG) over conversation turns. We will release MTCABench, a benchmark of 3,845 target spans across 665 multi-turn conversations, annotated with gold provenance graphs reaching depths of up to 14, across four dependency types. Across four open-weight models, flat attribution methods fail to recover multi-hop dependencies, achieving under 20% source recall, while Tokengeist reaches 90%. Our results reveal systematic failure modes of single-pass attribution -- which we term provenance collapse -- and motivate attribution methods that reason recursively across turns.