Paper Citation Record · LEDGER

Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use

As of 14 August 2026, this Paper Citation Record lists 10 of 10 outbound references and 3 inbound Pith citation observations for arXiv:2603.03205.

A citation records a reference. It does not transfer a finding from one paper to another.

pith.paper-citation-record.v1
2603.03205 v2

Coverage vector

measured 10 of 10 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-02T19:13:30.613413Z

measured 13 of 13 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-14T06:32:32.682623+00:00

measured 3 of 3 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-08-06T00:39:13.428111Z

measured 0 of 1 external citation measurements

A source-named dated measurement, never combined with another source.

Source: pith, observed 2026-07-03T16:58:43.698040Z

Reference resolution

10 of 10 outbound references displayed

verified exact0
verified fuzzy0
unresolved10
parse uncertain0
malformed identifier0
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation f23738fe-33c0-4cff-b07f-34e4f42d4eb3 · outbound

This paper cites an unresolved cited work.

Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use Unresolved cited work

Reference 1

Resolution: unresolved; no resolver link, observed 2026-08-02T19:13:28.958381Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0911f224-96c9-4521-92a4-5d53ad35fe59 · outbound

This paper cites an unresolved cited work.

Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use Unresolved cited work

Reference 2

Resolution: unresolved; no resolver link, observed 2026-08-02T19:13:29.086757Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d274d1b5-ff10-4cb9-a4a3-e31bc35a5485 · outbound

This paper cites Donald Drewski.

Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use Donald Drewski

Reference 3

Resolution: unresolved; no resolver link, observed 2026-08-02T19:13:29.225323Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d5baa7df-fb59-4a51-80ed-8d5122e2c5ef · outbound

This paper cites </safety thoughts> if required to evaluate and reason about potential risks, safety, legality, and ethical implications.

Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use </safety thoughts> if required to evaluate and reason about potential risks, safety, legality, and ethical implications

Reference 6

Resolution: unresolved; no resolver link, observed 2026-08-02T19:13:29.547779Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7fe3dfa4-9e65-44c5-8dfe-46910c32f223 · outbound

This paper cites name”: “refuse unsafe task.

Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use name”: “refuse unsafe task

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-02T19:13:30.083801Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 305d3cce-6c0c-4591-b6ed-d4be52e08e18 · outbound

This paper cites </think> tags to plan or analyze the task at the start of a turn.

Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use </think> tags to plan or analyze the task at the start of a turn

Reference 10

Resolution: unresolved; no resolver link, observed 2026-08-02T19:13:30.233647Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 757d9b41-1823-4492-8d45-dc4eaab71103 · outbound

This paper cites name”: “function name.

Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use name”: “function name

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-02T19:13:30.361433Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation cc79bbeb-6520-426f-b721-3b82c650f001 · outbound

This paper cites </tool response>, which will be provided by the user.

Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use </tool response>, which will be provided by the user

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-02T19:13:30.488711Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5f64f239-fd25-48f2-8554-f43d4524ffed · outbound

This paper cites name”: “refuse unsafe task.

Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use name”: “refuse unsafe task

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-02T19:13:30.613413Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f535833d-7053-401e-89a0-25179ad8250c · outbound

This paper cites Challenges in Guardrailing Large Language Models for Science.

Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use Challenges in Guardrailing Large Language Models for Science

Reference 2025

Resolution: unresolved; no resolver link, observed 2026-08-02T19:13:28.853540Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Pith citing papers

Observation 395601be-1313-4331-a841-dddd7408b4f5 · inbound

Trust No Tool: Evaluating and Defending LLM Agents under Untrusted Tool Feedback cites this paper.

Trust No Tool: Evaluating and Defending LLM Agents under Untrusted Tool Feedback Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use

Reference 39

Resolution: verified exact; arxiv_id, observed 2026-06-04T02:07:11.549850Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 7428cdc3-7bc4-4140-a5f2-510b2f601230 · inbound

From Question Answering to Task Completion: A Survey on Agent System and Harness Design cites this paper.

From Question Answering to Task Completion: A Survey on Agent System and Harness Design Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use

Reference 176

Resolution: verified exact; local_arxiv, observed 2026-07-03T16:58:43.699693Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 4265f210-c9c7-4901-bab7-8e3a5bbc57ed · inbound

Don't Offer What Can't Be Done: Deterministic Executability Gating for LLM Skill Selection at Scale cites this paper.

Don't Offer What Can't Be Done: Deterministic Executability Gating for LLM Skill Selection at Scale Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use

Reference 1

Resolution: unresolved; no resolver link, observed 2026-08-06T00:39:13.428111Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.