# Prompt Injection Discovery on AgentDyn Source: [Prompt Injection Discovery on AgentDyn]() Collection index: [llms.txt]() Data updated: 2026-06-12T06:58:22.041Z Evaluation of model performance on discovering prompt injection vulnerabilities using the AgentDyn dataset, focusing on metrics like Source Diversity and Instruction Diversity. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Prompt Injection Discovery | | Dataset (including any stated split) | AgentDyn | | Metric | Source Diversity | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 14 of 14 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Source Diversity](): Higher is better; unit: PERCENT. - [Instruction Diversity](): Higher is better; unit: PERCENT. - [Source Recall](): Higher is better; unit: PERCENT. - [Instruction Recall](): Higher is better; unit: PERCENT. ## Recorded results | Method | Source Diversity | Result date | Report paper | | --- | --- | --- | --- | | PI-Hunter | 78 | Not recorded | 6a2b9278d33dfb23e783af62 | | PI-Hunter | 70 | Not recorded | 6a2b9278d33dfb23e783af62 | | PI-Hunter | 67 | Not recorded | 6a2b9278d33dfb23e783af62 | | Baseline | 13 | Not recorded | 6a2b9278d33dfb23e783af62 | | Baseline | 10 | Not recorded | 6a2b9278d33dfb23e783af62 | | Baseline | 10 | Not recorded | 6a2b9278d33dfb23e783af62 | | PI-Hunter | 0.526 | Not recorded | 6a2b9278d33dfb23e783af62 | | PI-Hunter | 0.473 | Not recorded | 6a2b9278d33dfb23e783af62 | | PI-Hunter | 0.44 | Not recorded | 6a2b9278d33dfb23e783af62 | | PI-Hunter | 0.318 | Not recorded | 6a2b9278d33dfb23e783af62 | | Baseline | 0.075 | Not recorded | 6a2b9278d33dfb23e783af62 | | Baseline | 0 | Not recorded | 6a2b9278d33dfb23e783af62 | | Baseline | 0 | Not recorded | 6a2b9278d33dfb23e783af62 | | Baseline | 0 | Not recorded | 6a2b9278d33dfb23e783af62 | ### PI-Hunter Source Diversity: 78. Result ID: 6a2bae07fbc373592c297e6c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Gemini-3.1-pro"}. This record does not assert independent reproduction. ### PI-Hunter Source Diversity: 70. Result ID: 6a2bae07fbc373592c297e70. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Claude-4.6-sonnet"}. This record does not assert independent reproduction. ### PI-Hunter Source Diversity: 67. Result ID: 6a2bae07fbc373592c297e6e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"GPT-5.4-mini"}. This record does not assert independent reproduction. ### Baseline Source Diversity: 13. Result ID: 6a2bae07fbc373592c297e6b. [Reporting paper]() Recorded evaluation configuration: {"Model":"Gemini-3.1-pro"}. This record does not assert independent reproduction. ### Baseline Source Diversity: 10. Result ID: 6a2bae07fbc373592c297e6d. [Reporting paper]() Recorded evaluation configuration: {"Model":"GPT-5.4-mini"}. This record does not assert independent reproduction. ### Baseline Source Diversity: 10. Result ID: 6a2bae07fbc373592c297e6f. [Reporting paper]() Recorded evaluation configuration: {"Model":"Claude-4.6-sonnet"}. This record does not assert independent reproduction. ### PI-Hunter Source Diversity: 0.526. Result ID: 6a2bae07fbc373592c297e7a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Defense":"None"}. This record does not assert independent reproduction. ### PI-Hunter Source Diversity: 0.473. Result ID: 6a2bae07fbc373592c297e7c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Defense":"Spotlight"}. This record does not assert independent reproduction. ### PI-Hunter Source Diversity: 0.44. Result ID: 6a2bae07fbc373592c297e7e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Defense":"MELON"}. This record does not assert independent reproduction. ### PI-Hunter Source Diversity: 0.318. Result ID: 6a2bae07fbc373592c297e80. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Defense":"PIGuard"}. This record does not assert independent reproduction. ### Baseline Source Diversity: 0.075. Result ID: 6a2bae07fbc373592c297e79. [Reporting paper]() Recorded evaluation configuration: {"Defense":"None"}. This record does not assert independent reproduction. ### Baseline Source Diversity: 0. Result ID: 6a2bae07fbc373592c297e7b. [Reporting paper]() Recorded evaluation configuration: {"Defense":"Spotlight"}. This record does not assert independent reproduction. ### Baseline Source Diversity: 0. Result ID: 6a2bae07fbc373592c297e7d. [Reporting paper]() Recorded evaluation configuration: {"Defense":"MELON"}. This record does not assert independent reproduction. ### Baseline Source Diversity: 0. Result ID: 6a2bae07fbc373592c297e7f. [Reporting paper]() Recorded evaluation configuration: {"Defense":"PIGuard"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.