stackone-defender 0.7.1__tar.gz → 0.7.3__tar.gz

stackone_defender-0.7.3/.release-please-manifest.json

@@ -0,0 +1 @@
+{".":"0.7.3"}

{stackone_defender-0.7.1 → stackone_defender-0.7.3}/CHANGELOG.md

@@ -1,5 +1,26 @@
 # Changelog
 
+## [0.7.3](https://github.com/StackOneHQ/defender-py/compare/stackone-defender-v0.7.2...stackone-defender-v0.7.3) (2026-07-22)
+
+
+### Features
+
+* **ENG-1279:** port Tier 3 score-threshold (decide on score, not the model's word) ([77c8a6a](https://github.com/StackOneHQ/defender-py/commit/77c8a6a38336ddd797f5691cedf4670f15f575fb))
+
+
+### Bug Fixes
+
+* **ENG-1279:** Tier 3 score threshold + ENG-1296 Tier 2/base64/SFE fixes ([3f2b9cd](https://github.com/StackOneHQ/defender-py/commit/3f2b9cd1f7dd82f19e468076bcb3196ad22bf83e))
+* **ENG-1296:** port TS parity fixes for Tier 2 truncation, base64, SFE recursion ([4fa8cda](https://github.com/StackOneHQ/defender-py/commit/4fa8cda445d8c08e30e74d6053e2eaf26026fed2))
+
+## [0.7.2](https://github.com/StackOneHQ/stackone-defender/compare/stackone-defender-v0.7.1...stackone-defender-v0.7.2) (2026-06-30)
+
+
+### Features
+
+* add Tier3Verdict.usage for TS 0.7.2 parity ([2e18269](https://github.com/StackOneHQ/stackone-defender/commit/2e182695c20e9b087041e55f7466cca9b659521a))
+* Tier3Verdict.usage parity with @stackone/defender 0.7.2 ([1c9c0c8](https://github.com/StackOneHQ/stackone-defender/commit/1c9c0c8026fd8c4dda2f78c31a1f500c01045d43))
+
 ## [0.7.1](https://github.com/StackOneHQ/stackone-defender/compare/stackone-defender-v0.7.0...stackone-defender-v0.7.1) (2026-06-16)
 
 

stackone_defender-0.7.3/LICENSE

@@ -0,0 +1,190 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but not
+      limited to compiled object code, generated documentation, and
+      conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to the Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by the Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding any notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   Copyright 2024 StackOne
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

{stackone_defender-0.7.1 → stackone_defender-0.7.3}/PKG-INFO

@@ -1,14 +1,14 @@
 Metadata-Version: 2.4
 Name: stackone-defender
-Version: 0.7.1
+Version: 0.7.3
 Summary: Indirect prompt injection defense for AI agents using tool calls
 Project-URL: Homepage, https://github.com/StackOneHQ/stackone-defender
 Project-URL: Repository, https://github.com/StackOneHQ/stackone-defender
 Author-email: StackOne <support@stackone.com>
-License: Apache-2.0
+License-Expression: Apache-2.0
+License-File: LICENSE
 Classifier: Development Status :: 4 - Beta
 Classifier: Intended Audience :: Developers
-Classifier: License :: OSI Approved :: Apache Software License
 Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
@@ -146,6 +146,47 @@ Packed-chunk MiniLM classifier (int8 ONNX ~22 MB, bundled):
 | jayavibhav (adversarial) | 0.9717 | ~1k |
 | **Average** | **0.9079** | ~25k |
 
+### Optional Tier 3 — LLM adjudication (consumer-supplied)
+
+Authoritative LLM-based classification for the cases Tier 2 finds ambiguous. The package ships **only** the orchestration and the `Tier3Provider` interface — the actual model endpoint (a hosted LLM, OpenAI, an internal inference service) lives in your code, keeping proprietary models and credentials out of the package.
+
+Two modes, selected via `defender_mode`:
+- **`"cascade"`** (default): Tier 1 → Tier 2 → Tier 3, with Tier 3 invoked only when the Tier 2 effective score falls in the gray band (default `[0.3, 0.85)`). The Tier 3 verdict overrides Tier 2 on the escalated chunk — a `block` forces a block, an `allow` rescues it. Outside the band defender skips the round trip.
+- **`"tier3_only"`**: skip Tier 2; the block/allow decision is the Tier 3 verdict alone. Tier 1 still sanitizes the returned `sanitized` payload.
+
+Register a provider once at startup, then opt in per instance:
+
+```python
+from typing import Any
+
+from stackone_defender import create_prompt_defense, set_default_tier3_provider
+from stackone_defender.types import Tier3Verdict
+
+class MyProvider:
+    def classify(self, text: str, *, ctx: dict[str, Any] | None = None) -> Tier3Verdict:
+        # Call your LLM endpoint here (sync or awaitable). ctx["toolName"] is available.
+        verdict = call_my_llm(text, tool_name=(ctx or {}).get("toolName"))
+        return Tier3Verdict(decision="block" if verdict.block else "allow", score=verdict.p_block)
+
+set_default_tier3_provider(MyProvider())
+
+defense = create_prompt_defense(
+    block_high_risk=True,
+    enable_tier3=True,
+    defender_mode="cascade",  # or "tier3_only"
+    tier3={
+        "provider": MyProvider(),                          # optional: overrides the registry for this instance
+        "escalation_band": {"lower": 0.3, "upper": 0.85},  # cascade gray band; [lower, upper), defaults shown
+        "max_text_length": 10000,                          # caps text passed to the provider
+        "block_threshold": 0.622,                          # optional; decide on score, not the model's word
+    },
+)
+```
+
+**Choosing the operating point (`block_threshold`).** By default the model's `decision` word is authoritative — but that word is its argmax, an implicit 0.5 cut that moves on its own whenever the model is retrained. Set `tier3.block_threshold` to decide on `verdict.score` (P(block)) instead: the cut becomes an explicit config value — raise it to trade recall for fewer false positives, lower it for the reverse; `0.5` reproduces argmax. It requires a provider that reports `score` as P(block) (not "confidence in whichever decision I made" — those invert on allows). If `score` is missing or outside `[0, 1]` the verdict's `decision` is used instead and defender warns once, so a provider that cannot report a score degrades to the default rather than failing.
+
+**Fail-open semantics.** A provider error or timeout records a `skip_reason` on `result.tier3`; in cascade defender falls back to the Tier 2 decision, in `tier3_only` it allows the request. `enable_tier3=True` with no registered provider falls back to the standard Tier 1 + Tier 2 cascade and logs one warning per instance — Tier 3 misconfiguration never silently disables defense. `result.tier3` carries the verdict (a `Tier3Verdict`, or a `Tier3Skip` when the provider ran but returned nothing usable) when Tier 3 runs, and is `None` when it doesn't.
+
 ### `allowed` vs `risk_level`
 
 - Use **`allowed`** for gating when `block_high_risk=True`: `False` means do not pass `sanitized` to the model as-is.

{stackone_defender-0.7.1 → stackone_defender-0.7.3}/README.md

@@ -120,6 +120,47 @@ Packed-chunk MiniLM classifier (int8 ONNX ~22 MB, bundled):
 | jayavibhav (adversarial) | 0.9717 | ~1k |
 | **Average** | **0.9079** | ~25k |
 
+### Optional Tier 3 — LLM adjudication (consumer-supplied)
+
+Authoritative LLM-based classification for the cases Tier 2 finds ambiguous. The package ships **only** the orchestration and the `Tier3Provider` interface — the actual model endpoint (a hosted LLM, OpenAI, an internal inference service) lives in your code, keeping proprietary models and credentials out of the package.
+
+Two modes, selected via `defender_mode`:
+- **`"cascade"`** (default): Tier 1 → Tier 2 → Tier 3, with Tier 3 invoked only when the Tier 2 effective score falls in the gray band (default `[0.3, 0.85)`). The Tier 3 verdict overrides Tier 2 on the escalated chunk — a `block` forces a block, an `allow` rescues it. Outside the band defender skips the round trip.
+- **`"tier3_only"`**: skip Tier 2; the block/allow decision is the Tier 3 verdict alone. Tier 1 still sanitizes the returned `sanitized` payload.
+
+Register a provider once at startup, then opt in per instance:
+
+```python
+from typing import Any
+
+from stackone_defender import create_prompt_defense, set_default_tier3_provider
+from stackone_defender.types import Tier3Verdict
+
+class MyProvider:
+    def classify(self, text: str, *, ctx: dict[str, Any] | None = None) -> Tier3Verdict:
+        # Call your LLM endpoint here (sync or awaitable). ctx["toolName"] is available.
+        verdict = call_my_llm(text, tool_name=(ctx or {}).get("toolName"))
+        return Tier3Verdict(decision="block" if verdict.block else "allow", score=verdict.p_block)
+
+set_default_tier3_provider(MyProvider())
+
+defense = create_prompt_defense(
+    block_high_risk=True,
+    enable_tier3=True,
+    defender_mode="cascade",  # or "tier3_only"
+    tier3={
+        "provider": MyProvider(),                          # optional: overrides the registry for this instance
+        "escalation_band": {"lower": 0.3, "upper": 0.85},  # cascade gray band; [lower, upper), defaults shown
+        "max_text_length": 10000,                          # caps text passed to the provider
+        "block_threshold": 0.622,                          # optional; decide on score, not the model's word
+    },
+)
+```
+
+**Choosing the operating point (`block_threshold`).** By default the model's `decision` word is authoritative — but that word is its argmax, an implicit 0.5 cut that moves on its own whenever the model is retrained. Set `tier3.block_threshold` to decide on `verdict.score` (P(block)) instead: the cut becomes an explicit config value — raise it to trade recall for fewer false positives, lower it for the reverse; `0.5` reproduces argmax. It requires a provider that reports `score` as P(block) (not "confidence in whichever decision I made" — those invert on allows). If `score` is missing or outside `[0, 1]` the verdict's `decision` is used instead and defender warns once, so a provider that cannot report a score degrades to the default rather than failing.
+
+**Fail-open semantics.** A provider error or timeout records a `skip_reason` on `result.tier3`; in cascade defender falls back to the Tier 2 decision, in `tier3_only` it allows the request. `enable_tier3=True` with no registered provider falls back to the standard Tier 1 + Tier 2 cascade and logs one warning per instance — Tier 3 misconfiguration never silently disables defense. `result.tier3` carries the verdict (a `Tier3Verdict`, or a `Tier3Skip` when the provider ran but returned nothing usable) when Tier 3 runs, and is `None` when it doesn't.
+
 ### `allowed` vs `risk_level`
 
 - Use **`allowed`** for gating when `block_high_risk=True`: `False` means do not pass `sanitized` to the model as-is.

{stackone_defender-0.7.1 → stackone_defender-0.7.3}/pyproject.toml

@@ -1,15 +1,15 @@
 [project]
 name = "stackone-defender"
-version = "0.7.1"
+version = "0.7.3"
 description = "Indirect prompt injection defense for AI agents using tool calls"
 readme = "README.md"
 requires-python = ">=3.11"
-license = { text = "Apache-2.0" }
+license = "Apache-2.0"
+license-files = ["LICENSE"]
 authors = [{ name = "StackOne", email = "support@stackone.com" }]
 classifiers = [
   "Development Status :: 4 - Beta",
   "Intended Audience :: Developers",
-  "License :: OSI Approved :: Apache Software License",
   "Programming Language :: Python :: 3",
   "Programming Language :: Python :: 3.11",
   "Programming Language :: Python :: 3.12",
@@ -32,6 +32,8 @@ sfe = ["fasttext-ng>=0.9.3"]
 [dependency-groups]
 dev = [
   "pytest>=8.0",
+  "pytest-cov>=5.0",
+  "ruff>=0.9",
   "onnxruntime>=1.16.0",
   "tokenizers>=0.15.0",
   "numpy>=1.24.0",
@@ -45,5 +47,26 @@ build-backend = "hatchling.build"
 [tool.hatch.build.targets.wheel]
 packages = ["src/stackone_defender"]
 
+[tool.hatch.build.targets.sdist]
+exclude = ["/models", "/.github"]  # legacy model + CI, dev-only
+
+[tool.ruff]
+target-version = "py311"
+line-length = 120
+src = ["src", "tests"]
+
+[tool.ruff.lint]
+select = ["E", "W", "F", "I", "N", "UP", "B", "C4", "SIM", "RUF"]
+# E501/RUF003 by design; SIM*/UP007 deferred to cleanup.
+ignore = ["E501", "RUF003", "SIM102", "SIM103", "SIM108", "SIM110", "SIM118", "UP007"]
+
+[tool.ruff.lint.per-file-ignores]
+"tests/*" = ["RUF012", "N814", "F841", "F811"]
+
 [tool.pytest.ini_options]
 testpaths = ["tests"]
+addopts = "--strict-markers --strict-config -ra"
+markers = [
+  "model: requires the bundled ONNX model + onnxruntime extra",
+  "sfe: requires the fasttext extra",
+]

{stackone_defender-0.7.1 → stackone_defender-0.7.3}/src/stackone_defender/__init__.py

@@ -29,6 +29,7 @@ from .types import (
     RiskLevel,
     Tier1Result,
     Tier3Provider,
+    Tier3TokenUsage,
     Tier3Verdict,
 )
 from .utils.boundary import contains_boundary_patterns, generate_boundary_instructions
@@ -44,6 +45,7 @@ __all__ = [
     "SfePreprocessResult",
     "Tier1Result",
     "Tier3Provider",
+    "Tier3TokenUsage",
     "Tier3Verdict",
     "contains_boundary_patterns",
     "create_prompt_defense",

{stackone_defender-0.7.1 → stackone_defender-0.7.3}/src/stackone_defender/classifiers/onnx_classifier.py

@@ -17,7 +17,9 @@ from typing import Literal
 _logger = logging.getLogger(__name__)
 
 # Shared across all OnnxClassifier instances (keyed by resolved model dir path).
-_session_cache: dict[str, tuple[object, object]] = {}
+# Tuple is (session, tokenizer, count_tokenizer); the count tokenizer is
+# non-truncating so count_tokens reports true length (see _load_model).
+_session_cache: dict[str, tuple[object, object, object]] = {}
 _registry_lock = threading.Lock()
 _load_locks: dict[str, threading.Lock] = {}
 
@@ -64,6 +66,8 @@ class OnnxClassifier:
         self._model_path = model_path or get_default_model_path()
         self._session = None
         self._tokenizer = None
+        # Non-truncating tokenizer used only by count_tokens (see _load_model).
+        self._count_tokenizer = None
         self._max_length = 256
         self._load_failed = False
         # Output mode is detected lazily from the logits shape on the first
@@ -114,13 +118,13 @@ class OnnxClassifier:
         cache_key = str(Path(self._model_path).resolve())
         cached = _session_cache.get(cache_key)
         if cached:
-            self._session, self._tokenizer = cached
+            self._session, self._tokenizer, self._count_tokenizer = cached
             return
 
         with _lock_for_cache_key(cache_key):
             cached = _session_cache.get(cache_key)
             if cached:
-                self._session, self._tokenizer = cached
+                self._session, self._tokenizer, self._count_tokenizer = cached
                 return
 
             try:
@@ -140,13 +144,21 @@ class OnnxClassifier:
                 self._tokenizer.enable_truncation(max_length=self._max_length)
                 self._tokenizer.enable_padding(length=self._max_length)
 
+                # tokenizer.json bakes in truncation+padding, so disable both
+                # here to count true length rather than a value capped at
+                # max_length (which would make every long payload look like a
+                # single chunk and drop everything past the cap).
+                self._count_tokenizer = Tokenizer.from_file(tokenizer_path)
+                self._count_tokenizer.no_truncation()
+                self._count_tokenizer.no_padding()
+
                 onnx_path = str(Path(self._model_path) / "model_quantized.onnx")
                 self._session = ort.InferenceSession(onnx_path)
             except Exception as e:
                 _logger.warning("[defender] ONNX model failed to load: %s", e)
                 raise
 
-            _session_cache[cache_key] = (self._session, self._tokenizer)
+            _session_cache[cache_key] = (self._session, self._tokenizer, self._count_tokenizer)
 
     # ------------------------------------------------------------------
     # Inference
@@ -258,9 +270,9 @@ class OnnxClassifier:
 
     def count_tokens(self, text: str) -> int:
         self._ensure_loaded()
-        encoding = self._tokenizer.encode(text)
-        # Padding is enabled at a fixed length; count only real (attended) tokens.
-        return int(sum(encoding.attention_mask))
+        # Non-truncating count, including special tokens ([CLS]/[SEP]) to match
+        # the TS countTokens; the inference tokenizer would cap this at max_length.
+        return len(self._count_tokenizer.encode(text).ids)
 
     def get_max_length(self) -> int:
         return self._max_length

{stackone_defender-0.7.1 → stackone_defender-0.7.3}/src/stackone_defender/core/prompt_defense.py

@@ -11,7 +11,7 @@ import inspect
 import logging
 import math
 import time
-from dataclasses import dataclass, field
+from dataclasses import dataclass, field, replace
 from typing import Any
 
 from ..classifiers.pattern_detector import PatternDetector, create_pattern_detector
@@ -30,6 +30,7 @@ from ..types import (
     Tier3Provider,
     Tier3Result,
     Tier3Skip,
+    Tier3TokenUsage,
     Tier3Verdict,
 )
 from .tool_result_sanitizer import ToolResultSanitizer, create_tool_result_sanitizer
@@ -241,6 +242,8 @@ class PromptDefense:
         self._tier3_band = _DEFAULT_TIER3_BAND
         self._tier3_max_text_length = _DEFAULT_TIER3_MAX_TEXT_LENGTH
         self._tier3_missing_provider_warned = False
+        self._tier3_block_threshold: float | None = None
+        self._tier3_missing_score_warned = False
         tier3_opts = tier3 or {}
         if tier3_opts.get("provider") is not None:
             self._tier3_custom_provider = tier3_opts["provider"]
@@ -274,6 +277,21 @@ class PromptDefense:
                     lower,
                     upper,
                 )
+        block_threshold = tier3_opts.get("block_threshold", tier3_opts.get("blockThreshold"))
+        if block_threshold is not None:
+            if (
+                isinstance(block_threshold, (int, float))
+                and not isinstance(block_threshold, bool)
+                and math.isfinite(block_threshold)
+                and 0 <= block_threshold <= 1
+            ):
+                self._tier3_block_threshold = float(block_threshold)
+            else:
+                _logger.warning(
+                    "[defender] invalid tier3.block_threshold %s — must be a finite number in [0, 1]. "
+                    "Falling back to the provider's decision.",
+                    block_threshold,
+                )
 
     def warmup_tier2(self) -> None:
         if self._tier2:
@@ -292,11 +310,37 @@ class PromptDefense:
     def _resolve_tier3_provider(self) -> Tier3Provider | None:
         return self._tier3_custom_provider or get_default_tier3_provider()
 
+    @staticmethod
+    def _parse_tier3_usage(usage: Any) -> Tier3TokenUsage | None:
+        if usage is None or not isinstance(usage, dict):
+            return None
+        prompt_tokens = usage.get("prompt_tokens", usage.get("promptTokens"))
+        completion_tokens = usage.get("completion_tokens", usage.get("completionTokens"))
+        total_tokens = usage.get("total_tokens", usage.get("totalTokens"))
+        if not all(isinstance(value, int) for value in (prompt_tokens, completion_tokens, total_tokens)):
+            return None
+        return Tier3TokenUsage(
+            prompt_tokens=prompt_tokens,
+            completion_tokens=completion_tokens,
+            total_tokens=total_tokens,
+        )
+
+    @staticmethod
+    def _normalize_tier3_score(score: Any) -> float | None:
+        # Public contract is score in [0, 1]; drop anything else (wrong type,
+        # non-finite, out of range) rather than leak it to consumers or the
+        # block gate. bool is excluded since it is not a numeric score.
+        if isinstance(score, bool) or not isinstance(score, (int, float)):
+            return None
+        if not math.isfinite(score) or score < 0 or score > 1:
+            return None
+        return float(score)
+
     @staticmethod
     def _validate_tier3_verdict(verdict: Any) -> Tier3Verdict | Tier3Skip:
         if isinstance(verdict, Tier3Verdict):
             if verdict.decision in ("block", "allow"):
-                return verdict
+                return replace(verdict, score=PromptDefense._normalize_tier3_score(verdict.score))
             return Tier3Skip(
                 skip_reason=(
                     f'Tier 3 provider returned invalid decision: {verdict.decision!r} '
@@ -314,11 +358,37 @@ class PromptDefense:
             )
         return Tier3Verdict(
             decision=decision,
-            score=verdict.get("score"),
+            score=PromptDefense._normalize_tier3_score(verdict.get("score")),
             raw=verdict.get("raw"),
             latency_ms=verdict.get("latency_ms", verdict.get("latencyMs")),
+            usage=PromptDefense._parse_tier3_usage(verdict.get("usage")),
         )
 
+    def _is_tier3_block(self, verdict: Tier3Verdict) -> bool:
+        """Resolve a validated verdict to block/allow.
+
+        With ``block_threshold`` unset this is the model's ``decision`` (its
+        argmax, an implicit 0.5 cut). With a threshold set we decide on ``score``
+        (P(block)) instead, which is what makes any other operating point
+        reachable. A threshold with no usable score falls back to ``decision``
+        (warned once) rather than taking the tier offline.
+        """
+        if self._tier3_block_threshold is None:
+            return verdict.decision == "block"
+        # Validation already dropped any score outside [0, 1], so a non-None
+        # score here is usable as P(block).
+        if verdict.score is not None:
+            return verdict.score >= self._tier3_block_threshold
+        if not self._tier3_missing_score_warned:
+            self._tier3_missing_score_warned = True
+            _logger.warning(
+                "[defender] tier3.block_threshold=%s is set but the provider did not report a usable "
+                "score (missing, or not a number in [0, 1]). Falling back to the provider's decision — "
+                "the threshold is not being applied.",
+                self._tier3_block_threshold,
+            )
+        return verdict.decision == "block"
+
     @staticmethod
     async def _invoke_tier3_classify(provider: Tier3Provider, text: str, tool_name: str) -> Any:
         ctx = {"toolName": tool_name}
@@ -374,7 +444,7 @@ class PromptDefense:
             validated = self._validate_tier3_verdict(raw)
             if isinstance(validated, Tier3Skip):
                 return validated, None
-            return validated, validated.decision == "block"
+            return validated, self._is_tier3_block(validated)
         except Exception as e:
             return Tier3Skip(skip_reason=f"Tier 3 provider error: {e}"), None
 
@@ -440,7 +510,7 @@ class PromptDefense:
         sanitized = self._tool_sanitizer.sanitize(value, tool_name=tool_name)
         detections, fields_sanitized, prm = self._tier1_metadata(sanitized)
 
-        blocked = verdict is not None and verdict.decision == "block"
+        blocked = verdict is not None and self._is_tier3_block(verdict)
         risk_level: RiskLevel = "high" if blocked else "low"
         allowed = not self._config.block_high_risk or not blocked
         tier3_result: Tier3Result = (

{stackone_defender-0.7.1 → stackone_defender-0.7.3}/src/stackone_defender/core/tool_result_sanitizer.py

@@ -22,7 +22,6 @@ from ..types import (
     SanitizationContext,
     SanitizationMetadata,
     SanitizationResult,
-    SizeMetrics,
     TraversalConfig,
 )
 from ..utils.boundary import generate_data_boundary

{stackone_defender-0.7.1 → stackone_defender-0.7.3}/src/stackone_defender/sanitizers/__init__.py

@@ -1,6 +1,12 @@
 """Sanitizers for prompt injection mitigation."""
 
-from .encoding_detector import contains_encoded_content, contains_suspicious_encoding, decode_all_encoding, detect_encoding, redact_all_encoding
+from .encoding_detector import (
+    contains_encoded_content,
+    contains_suspicious_encoding,
+    decode_all_encoding,
+    detect_encoding,
+    redact_all_encoding,
+)
 from .normalizer import analyze_suspicious_unicode, contains_suspicious_unicode, normalize_unicode
 from .pattern_remover import remove_patterns
 from .role_stripper import contains_role_markers, strip_role_markers

{stackone_defender-0.7.1 → stackone_defender-0.7.3}/src/stackone_defender/sanitizers/encoding_detector.py

@@ -105,7 +105,10 @@ def _detect_base64(text: str, min_length: int) -> list[EncodingDetection]:
         if len(candidate) < min_length:
             continue
         try:
-            decoded_bytes = base64.b64decode(candidate, validate=False)
+            # Pad to a multiple of 4; b64decode rejects unpadded input, which
+            # the except below would silently drop (JS atob tolerates it).
+            padded = candidate + "=" * (-len(candidate) % 4)
+            decoded_bytes = base64.b64decode(padded, validate=False)
             try:
                 decoded = decoded_bytes.decode("ascii")
             except UnicodeDecodeError: