Merge remote-tracking branch 'origin/main' into copilot/standardize-deprecation-calls

# Conflicts:
#	pyrit/prompt_target/common/prompt_chat_target.py

Author: romanlutz

.github/instructions/style-guide.instructions.md

@@ -192,6 +192,33 @@ def calculate_score(
     """
 ```
 
+### Code references in docstrings
+
+The PyRIT docs build uses **MyST** (Markdown-flavoured), not reStructuredText.
+Do **not** use reST cross-reference roles in docstrings or module comments —
+they render as raw text under MyST and are inconsistent with the rest of the
+codebase, which uses plain double-backtick code spans for symbol names.
+
+```python
+# WRONG — reST roles render as literal `:class:\`SeedPrompt\`` under MyST
+"""Returns a :class:`SeedPrompt` instance."""
+"""Delegate to :func:`download_files_async` (deprecated alias)."""
+"""See :meth:`PromptTarget.apply_capabilities` for details."""
+
+# CORRECT — plain double-backtick code span (matches existing convention)
+"""Returns a ``SeedPrompt`` instance."""
+"""Delegate to ``download_files_async`` (deprecated alias)."""
+"""See ``PromptTarget.apply_capabilities`` for details."""
+```
+
+Roles to avoid include `:class:`, `:func:`, `:meth:`, `:mod:`, `:attr:`,
+`:data:`, `:exc:`, `:obj:`, `:ref:`, and any `:py:*:` variants
+(e.g. `:py:class:`, `:py:func:`).
+
+If you genuinely need a Sphinx cross-reference (rare in PyRIT — most
+docstrings just name the symbol in backticks), use the MyST role syntax
+`` {class}`Name` `` instead. The default, though, is plain double-backticks.
+
 ### Class-Level Constants
 - Define constants as class attributes, not module-level
 - Use UPPER_CASE naming for constants
@@ -454,6 +481,7 @@ Before committing code, ensure:
 - [ ] All functions have complete type annotations
 - [ ] Functions with >1 parameter use keyword-only arguments
 - [ ] Docstrings include parameter types
+- [ ] Docstrings use plain double-backtick code spans for symbol references (no reST roles)
 - [ ] Enums are used instead of Literals
 - [ ] Functions are focused and under 20 lines
 - [ ] Error messages are helpful and specific

doc/bibliography.md

@@ -5,6 +5,6 @@ All academic papers, research blogs, and technical reports referenced throughout
 :::{dropdown} Citation Keys
 :class: hidden-citations
 
-[@aakanksha2024multilingual; @adversaai2023universal; @andriushchenko2024tense; @anthropic2024manyshot; @aqrawi2024singleturncrescendo; @bethany2024mathprompt; @bhardwaj2023harmfulqa; @bhardwaj2024homer; @bryan2025agentictaxonomy; @bullwinkel2025airtlessons; @bullwinkel2025repeng; @bullwinkel2026trigger; @chao2023pair; @chao2024jailbreakbench; @cui2024orbench; @darkbench2025; @derczynski2024garak; @ding2023wolf; @embracethered2024unicode; @embracethered2025sneakybits; @ghosh2025aegis; @gupta2024walledeval; @haider2024phi3safety; @han2024medsafetybench; @hines2024spotlighting; @ji2023beavertails; @ji2024pkusaferlhf; @jiang2025sosbench; @jones2025computeruse; @kingma2014adam; @li2024saladbench; @li2024wmdp; @lin2023toxicchat; @liu2024flipattack; @lopez2024pyrit; @lv2024codechameleon; @mazeika2023tdc; @mazeika2024harmbench; @mckee2024transparency; @mehrotra2023tap; @microsoft2024skeletonkey; @palaskar2025vlsu; @pfohl2024equitymedqa; @promptfoo2025ccp; @robustintelligence2024bypass; @roccia2024promptintel; @rottger2023xstest; @rottger2025msts; @russinovich2024crescendo; @russinovich2025price; @scheuerman2025transphobia; @shaikh2022second; @shayegani2025computeruse; @shen2023donotanything; @sheshadri2024lat; @stok2023ansi; @tan2026comicjailbreak; @tang2025multilingual; @tedeschi2024alert; @vantaylor2024socialbias; @vidgen2023simplesafetytests; @vidgen2024ailuminate; @wang2023decodingtrust; @wang2023donotanswer; @wei2023jailbroken; @xie2024sorrybench; @yu2023gptfuzzer; @yuan2023cipherchat; @zeng2024persuasion; @zhang2024cbtbench; @zou2023gcg]
+[@aakanksha2024multilingual; @adversaai2023universal; @andriushchenko2024tense; @anthropic2024manyshot; @aqrawi2024singleturncrescendo; @bethany2024mathprompt; @bhardwaj2023harmfulqa; @bhardwaj2024homer; @brahman2024coconot; @bryan2025agentictaxonomy; @bullwinkel2025airtlessons; @bullwinkel2025repeng; @bullwinkel2026trigger; @chao2023pair; @chao2024jailbreakbench; @cui2024orbench; @darkbench2025; @derczynski2024garak; @ding2023wolf; @embracethered2024unicode; @embracethered2025sneakybits; @ghosh2025aegis; @gupta2024walledeval; @haider2024phi3safety; @han2024medsafetybench; @hines2024spotlighting; @ji2023beavertails; @ji2024pkusaferlhf; @jiang2025sosbench; @jones2025computeruse; @kingma2014adam; @li2024saladbench; @li2024wmdp; @lin2023toxicchat; @liu2024flipattack; @lopez2024pyrit; @lv2024codechameleon; @mazeika2023tdc; @mazeika2024harmbench; @mckee2024transparency; @mehrotra2023tap; @microsoft2024skeletonkey; @palaskar2025vlsu; @pfohl2024equitymedqa; @promptfoo2025ccp; @robustintelligence2024bypass; @roccia2024promptintel; @rottger2023xstest; @rottger2025msts; @russinovich2024crescendo; @russinovich2025price; @scheuerman2025transphobia; @shaikh2022second; @shayegani2025computeruse; @shen2023donotanything; @sheshadri2024lat; @stok2023ansi; @tan2026comicjailbreak; @tang2025multilingual; @tedeschi2024alert; @vantaylor2024socialbias; @vidgen2023simplesafetytests; @vidgen2024ailuminate; @wang2023decodingtrust; @wang2023donotanswer; @wei2023jailbroken; @xie2024sorrybench; @yu2023gptfuzzer; @yuan2023cipherchat; @zeng2024persuasion; @zhang2024cbtbench; @zou2023gcg]
 
 :::

doc/code/datasets/1_loading_datasets.ipynb

@@ -19,6 +19,7 @@
     "BeaverTails [@ji2023beavertails],\n",
     "CBT-Bench [@zhang2024cbtbench],\n",
     "CategoricalHarmfulQA (CatQA) [@bhardwaj2024homer],\n",
+    "CoCoNot [@brahman2024coconot],\n",
     "DarkBench [@darkbench2025],\n",
     "Do Anything Now [@shen2023donotanything],\n",
     "Do-Not-Answer [@wang2023donotanswer],\n",
@@ -79,6 +80,8 @@
        " 'categorical_harmful_qa',\n",
        " 'cbt_bench',\n",
        " 'ccp_sensitive_prompts',\n",
+       " 'coconot_contrast',\n",
+       " 'coconot_refusal',\n",
        " 'comic_jailbreak',\n",
        " 'dangerous_qa',\n",
        " 'dark_bench',\n",

doc/code/datasets/1_loading_datasets.py

@@ -23,6 +23,7 @@
 # BeaverTails [@ji2023beavertails],
 # CBT-Bench [@zhang2024cbtbench],
 # CategoricalHarmfulQA (CatQA) [@bhardwaj2024homer],
+# CoCoNot [@brahman2024coconot],
 # DarkBench [@darkbench2025],
 # Do Anything Now [@shen2023donotanything],
 # Do-Not-Answer [@wang2023donotanswer],

doc/references.bib

@@ -366,10 +366,10 @@ @misc{embracethered2024unicode
 
 @misc{robustintelligence2024bypass,
   title     = {Bypassing Meta's {LLaMA} Classifier: A Simple Jailbreak},
-  author    = {{Robust Intelligence}},
+  author    = {Aman Priyanshu},
   year      = {2024},
-  url       = {https://www.robustintelligence.com/blog-posts/bypassing-metas-llama-classifier-a-simple-jailbreak},
-  note      = {Robust Intelligence Blog},
+  url       = {https://blogs.cisco.com/security/bypassing-metas-llama-classifier-a-simple-jailbreak},
+  note      = {Robust Intelligence (now part of Cisco) Blog},
 }
 
 @misc{adversaai2023universal,
@@ -610,3 +610,11 @@ @misc{embracethered2025sneakybits
   url       = {https://embracethered.com/blog/posts/2025/sneaky-bits-and-ascii-smuggler/},
   note      = {Embrace The Red Blog},
 }
+
+@article{brahman2024coconot,
+  title     = {The Art of Saying No: Contextual Noncompliance in Language Models},
+  author    = {Faeze Brahman and Sachin Kumar and Vidhisha Balachandran and Pradeep Dasigi and Valentina Pyatkin and Abhilasha Ravichander and Sarah Wiegreffe and Nouha Dziri and Khyathi Chandu and Jack Hessel and Yulia Tsvetkov and Noah A. Smith and Yejin Choi and Hannaneh Hajishirzi},
+  journal   = {arXiv preprint arXiv:2407.12043},
+  year      = {2024},
+  url       = {https://arxiv.org/abs/2407.12043},
+}

pyrit/auth/azure_auth.py

@@ -443,10 +443,10 @@ async def get_speech_config_async(
     """
     Get the speech config, resolving a callable token provider if one is provided.
 
-    This is the async counterpart to :func:`get_speech_config`. When a callable
+    This is the async counterpart to ``get_speech_config``. When a callable
     ``token_provider`` is supplied, it is invoked (and awaited if async) to obtain
     a token, which is then used with the ``aad#{resource_id}#{token}`` auth format.
-    Otherwise, it delegates to the synchronous :func:`get_speech_config`.
+    Otherwise, it delegates to the synchronous ``get_speech_config``.
 
     Args:
         token_provider (Callable | None): An optional sync or async callable that returns a token string.

pyrit/common/data_url_converter.py

@@ -47,7 +47,7 @@ async def convert_local_image_to_data_url_async(image_path: str) -> str:
 
 async def convert_local_image_to_data_url(image_path: str) -> str:
     """
-    Delegate to :func:`convert_local_image_to_data_url_async` (deprecated alias).
+    Delegate to ``convert_local_image_to_data_url_async`` (deprecated alias).
 
     Returns:
         str: A string containing the MIME type and the base64-encoded data of the image, formatted as a data URL.

pyrit/common/display_response.py

@@ -58,7 +58,7 @@ async def display_image_response_async(response_piece: MessagePiece) -> None:
 
 
 async def display_image_response(response_piece: MessagePiece) -> None:
-    """Delegate to :func:`display_image_response_async` (deprecated alias)."""
+    """Delegate to ``display_image_response_async`` (deprecated alias)."""
     print_deprecation_message(
         old_item="pyrit.common.display_response.display_image_response",
         new_item="pyrit.common.display_response.display_image_response_async",

pyrit/common/download_hf_model.py

@@ -129,7 +129,7 @@ async def download_with_limit_async(url: str) -> None:
 
 
 async def download_specific_files(model_id: str, file_patterns: list[str] | None, token: str, cache_dir: Path) -> None:
-    """Delegate to :func:`download_specific_files_async` (deprecated alias)."""
+    """Delegate to ``download_specific_files_async`` (deprecated alias)."""
     print_deprecation_message(
         old_item="pyrit.common.download_hf_model.download_specific_files",
         new_item="pyrit.common.download_hf_model.download_specific_files_async",
@@ -140,7 +140,7 @@ async def download_specific_files(model_id: str, file_patterns: list[str] | None
 
 async def download_chunk(url: str, headers: dict[str, str], start: int, end: int, client: httpx.AsyncClient) -> bytes:
     """
-    Delegate to :func:`download_chunk_async` (deprecated alias).
+    Delegate to ``download_chunk_async`` (deprecated alias).
 
     Returns:
         The content of the downloaded chunk.
@@ -154,7 +154,7 @@ async def download_chunk(url: str, headers: dict[str, str], start: int, end: int
 
 
 async def download_file(url: str, token: str, download_dir: Path, num_splits: int) -> None:
-    """Delegate to :func:`download_file_async` (deprecated alias)."""
+    """Delegate to ``download_file_async`` (deprecated alias)."""
     print_deprecation_message(
         old_item="pyrit.common.download_hf_model.download_file",
         new_item="pyrit.common.download_hf_model.download_file_async",
@@ -166,7 +166,7 @@ async def download_file(url: str, token: str, download_dir: Path, num_splits: in
 async def download_files(
     urls: list[str], token: str, download_dir: Path, num_splits: int = 3, parallel_downloads: int = 4
 ) -> None:
-    """Delegate to :func:`download_files_async` (deprecated alias)."""
+    """Delegate to ``download_files_async`` (deprecated alias)."""
     print_deprecation_message(
         old_item="pyrit.common.download_hf_model.download_files",
         new_item="pyrit.common.download_hf_model.download_files_async",

pyrit/datasets/seed_datasets/remote/__init__.py

@@ -28,6 +28,12 @@
 from pyrit.datasets.seed_datasets.remote.ccp_sensitive_prompts_dataset import (
     _CCPSensitivePromptsDataset,
 )  # noqa: F401
+from pyrit.datasets.seed_datasets.remote.coconot_dataset import (
+    CoCoNotCategory,
+    CoCoNotSplit,
+    _CoCoNotContrastDataset,
+    _CoCoNotRefusalDataset,
+)  # noqa: F401
 from pyrit.datasets.seed_datasets.remote.comic_jailbreak_dataset import (
     COMIC_JAILBREAK_TEMPLATES,
     ComicJailbreakTemplateConfig,
@@ -142,6 +148,8 @@
 )  # noqa: F401
 
 __all__ = [
+    "CoCoNotCategory",
+    "CoCoNotSplit",
     "HiXSTestLanguage",
     "PromptIntelCategory",
     "PromptIntelSeverity",
@@ -156,6 +164,8 @@
     "_CBTBenchDataset",
     "_CCPSensitivePromptsDataset",
     "_CategoricalHarmfulQADataset",
+    "_CoCoNotContrastDataset",
+    "_CoCoNotRefusalDataset",
     "_ComicJailbreakDataset",
     "COMIC_JAILBREAK_TEMPLATES",
     "ComicJailbreakTemplateConfig",