coretrace-python-analyzer 0.6.0__tar.gz → 0.8.0__tar.gz

{coretrace_python_analyzer-0.6.0 → coretrace_python_analyzer-0.8.0}/CHANGELOG.md

@@ -1,5 +1,26 @@
 # Changelog
 
+## 0.8.0 (2026-09-25)
+
+### Precision
+
+- A `Validator` proves its argument safe only for the kinds it declares. A validator declared for `REDIRECT` used to refute a command injection through the same value; the flow is now a hotspot, and it is refuted only when the validator covers every kind reaching the sink (#143).
+- A sanitizer called inside a project function protects the flows through that function, for every caller. That covers a helper that escapes before answering, a view redirecting to `reverse(...)`, and a helper rendering a template the analyzer shows escaping. Function summaries keep, for each parameter, the taint kinds cleared on every path from it to an external call or to the return value. A path that skips the sanitizer, or a kind it does not clear, still reaches the sink. The cache format is bumped (#134).
+
+### Plugins
+
+- `Plugin.project_models(root)` lets a plugin read models from the project under analysis, such as validators the project declares in a file. It is called once per directory check, in every process, and its models are part of the cache key. A model declared wrongly raises `ModelError`, which the command line now reports as an error (exit status 2) instead of a traceback (#140).
+- A `Validator` model may name a function of the project by its project symbol (`python.hc.accounts.views._allow_redirect`), and the refutation recognises it in its own module too, where a call to it has no imported symbol. A project can thus declare its own validation helpers, and a flow they guard is refuted (#134).
+
+## 0.7.0 (2026-09-25)
+
+### Precision
+
+- A dominating guard makes a flow a hotspot only when it examines what the sink receives. `if request.method == 'POST':` before `request.POST['cmd']` no longer lowers a vulnerability to a hotspot. A test of the value itself, of the same attribute, of a method called on it (`form.is_valid()`), or of any part of an object the flow passes whole still does. The regression snapshots record the verdict of each taint finding (#135).
+- The SSRF sinks of Requests and httpx read the destination only: the URL (the first argument, the second after the method for `request` and `stream`, or `url=`) or the prepared request `send` takes. Attacker data in the body, the JSON, the headers or the query parameters is no longer a server-side request forgery. `Sink` gains `keywords` next to `positions`, and function summaries keep the name of each keyword argument, so a project function forwarding `url=` still reaches the sink; the cache format is bumped (#128).
+- `render_to_string` escapes what it renders where the analyzer can establish it: on a directory check, it reads the project's Django templates, and a template escapes when neither it nor what it extends or includes marks output safe, uses a tag, filter or library beyond Django's own, or prints a variable where HTML escaping does not protect it, and no settings turn autoescaping off. Data reaching a response through such a template is no longer `xss`; any other template keeps the flow reported. `TemplateRender` declares such rendering calls, and the escaped templates are part of the cache key (#130).
+- Django's `csrf.get_token` returns a token of letters and digits whatever the request holds, and `reverse`/`reverse_lazy` build a local path that attacker data in their arguments or query cannot turn into another host: the first is no longer an injection, the second no longer an open redirect. `reverse` still carries other kinds, since it leaves `'` unquoted (#132).
+
 ## 0.6.0 (2026-09-24)
 
 ### Reports

{coretrace_python_analyzer-0.6.0 → coretrace_python_analyzer-0.8.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: coretrace-python-analyzer
-Version: 0.6.0
+Version: 0.8.0
 Summary: Static security analysis infrastructure for Python code
 Project-URL: Homepage, https://github.com/CoreTrace/coretrace-python-analyzer
 Project-URL: Repository, https://github.com/CoreTrace/coretrace-python-analyzer

{coretrace_python_analyzer-0.6.0 → coretrace_python_analyzer-0.8.0}/docs/plugins.md

@@ -141,22 +141,40 @@ class BottleModels(ModelPlugin):
 | Model | Declares |
 |---|---|
 | `Source(symbol, label, kinds=ALL)` | A call or attribute whose value is attacker-controlled; `label` names it in messages (`http`, `stdin`, `argv`, `http-response`). `kinds` restricts what it can inject. |
-| `Sink(symbol, kinds, positions=())` | A call dangerous for `kinds`; `positions` limits a kind to given argument indexes, as SQL sinks read the statement only. |
-| `Sanitizer(symbol, kinds)` | A call whose result is safe for `kinds`. |
+| `Sink(symbol, kinds, positions=(), keywords=())` | A call dangerous for `kinds`. `positions` limits a kind to the given argument indexes, and `keywords` to the given keyword names; a restricted kind reaches only the arguments listed there. SQL sinks read the statement only, and the SSRF sinks of HTTP clients read the URL, by position or as `url=`. A starred argument or `**kwargs` reaches no restricted kind. |
+| `Sanitizer(symbol, kinds)` | A call whose result is safe for `kinds`, including when the call is inside a project function the data goes through. |
 | `EntryPoint(symbol, label, kinds=ALL)` | A decorator or a class base whose functions receive their parameters as `label` input, such as `flask.Flask.route`. |
 | `TypedParameter(symbol, label, kinds=ALL)` | A class whose annotated parameters carry `label` input, such as Django's `HttpRequest`. |
 | `NamedParameter(pattern, label, kinds)` | A regular expression on parameter names; matching parameters carry `kinds`, as `password` carries `CREDENTIAL`. |
 | `RouteRegistrar(symbol, argument, label, kinds=ALL, keyword=None)` | A call registering a view at argument `argument` (or `keyword`), such as Django's `path`; the view receives `label` input. |
 | `SuffixSink(suffix, kinds, positions=())` | A sink matched by the end of the symbol, for methods of any class such as `objects.raw`. |
 | `SafeArgument(symbol, argument, values, position=None, kinds=ALL)` | A call to the sink `symbol` whose `argument` (by keyword, or at `position`) denotes one of `values` is not a sink for `kinds`, such as `yaml.load` with `Loader=yaml.SafeLoader`. Values are written as the analyzer records arguments: a symbol by its canonical name, a constant as Python writes it (`True`). Only a value given explicitly makes the call safe. |
-| `Validator(symbol, kinds=ALL, argument=0)` | A callable whose truth proves its argument safe, such as `re.fullmatch`; a flow guarded by it is refuted. |
+| `TemplateRender(symbol, position=0, keyword="template_name")` | A call rendering, with autoescaping, the template named by its argument at `position` or `keyword`, such as `render_to_string`. What it returns carries no `HTML` when the name is written in the call and the project's template escapes everything it prints (see the usage guide). |
+| `Validator(symbol, kinds=ALL, argument=0)` | A callable whose truth proves its argument safe for `kinds`, such as `re.fullmatch`. A flow guarded by it is refuted when it carries no other kind into the sink, and is a hotspot otherwise: a redirect validator does not make a shell command safe. `symbol` may be a function of the project, by its project symbol (`python.hc.accounts.views._allow_redirect`), which then counts wherever it is called, in its own module too. |
 | `AuthorizationGuard(symbol, label)` | A decorator or a condition restricting who reaches the code, such as `login_required`; a flow behind it is a hotspot. |
 
 Two plugins may describe the same symbol. An identical model is registered once; two
-`Sink` models merge their kinds and positions, so a plugin can add `SQL` to a sink
+`Sink` models merge their kinds, positions and keywords, so a plugin can add `SQL` to a sink
 another plugin declared for `COMMAND`; any other difference is a conflict that stops
 the analysis with a `ModelError` naming both plugins.
 
+A plugin may also read models from the project it analyses: `project_models(self, root)`
+returns them, from a file at `root` for instance. It is called once per directory check,
+in every process analysing it, before any module. Its models join the plugin's own and
+are part of what cached results depend on. A single-file check reads no project. A
+model declared wrongly raises `ModelError`, which stops the check with its message
+(exit status 2). This is how a project declares validators of its own:
+
+```python
+class DeclaredValidators(ModelPlugin):
+    name: ClassVar[str] = "declared-validators"
+
+    def project_models(self, root: Path) -> Sequence[Model]:
+        path = root / "validators.txt"
+        lines = path.read_text(encoding="utf-8").splitlines() if path.is_file() else []
+        return tuple(Validator(SymbolId(line)) for line in lines if line)
+```
+
 A model plugin may also carry `advisories`, a tuple of `Advisory` values
 (`coretrace_python.dependency`) with the package, the vulnerable range and the affected
 symbols, as the shipped `sample-advisories` plugin does. Requirements matching them are

{coretrace_python_analyzer-0.6.0 → coretrace_python_analyzer-0.8.0}/docs/usage.md

@@ -101,15 +101,19 @@ confidence (`high`, `medium`, `low`).
 ### Taint flows
 
 Reported when attacker-controlled data reaches a sensitive call without passing through
-a sanitizer for that kind of sink.
+a sanitizer for that kind of sink. The sanitizer may also sit inside a project function
+the data goes through, like a helper that escapes before it answers, or a view that
+redirects to `reverse(...)`. It then protects every caller, on every path through the
+function where it is called. What a function stores into an object or a `nonlocal`
+variable is not treated that way yet.
 
 | Rule | Reached sink |
 |---|---|
 | `command-injection` | Shell or process execution (`os.system`, `subprocess` with a string, …). |
 | `sql-injection` | A database statement (`cursor.execute`, SQLAlchemy `text`, Django `raw`, …). Parameters of a parameterised query are not statements. |
 | `path-traversal` | A file system path (`open`, `send_file`, `os.remove`, …). |
-| `ssrf` | The URL of an HTTP client request (Requests, httpx, `urllib`). |
-| `xss` | An HTTP response body built without escaping. |
+| `ssrf` | The URL of an HTTP client request (Requests, httpx, `urllib`), by position or as `url=`; the body, headers and query parameters are not the destination. |
+| `xss` | An HTTP response body built without escaping. A Django template rendered by `render_to_string` escapes, where the analyzer can establish it (below). |
 | `insecure-deserialization` | `pickle`, `marshal`, `dill`, `jsonpickle`, `shelve`, and the YAML loaders (`yaml.load`, `load_all`, `full_load`, `unsafe_load` and their several-document versions); `yaml.load` and `load_all` given a safe loader (`SafeLoader`, `BaseLoader`) are not. |
 | `open-redirect` | An HTTP redirect target. |
 | `plaintext-credential-storage` | A parameter named like a password stored in a database without hashing. Medium confidence, since a name is a hint. |
@@ -118,9 +122,42 @@ a sanitizer for that kind of sink.
 Each flow is judged before it is reported. A dominating guard that proves the value
 safe (`isdigit()`, membership in a constant collection, equality with a constant, a
 numeric value proven by `int()`, `len()` or bounded arithmetic, a validator such as
-`re.fullmatch`) refutes it and nothing is reported. A guard that only mentions the value,
-or an authorization decorator such as `login_required`, makes it a hotspot reported at
-medium confidence. The verdict and its evidence are in the finding's metadata.
+`re.fullmatch`) refutes it and nothing is reported. A hotspot is reported at medium
+confidence. Two things make a flow a hotspot:
+
+- A guard that examines the data without proving it safe. That covers a test of the value
+  itself (`if cmd:`, `len(cmd) < 10`), of the same attribute, or of a method called on
+  it (`form.is_valid()`).
+- An authorization decorator such as `login_required`.
+
+A guard that reads another attribute of the object the data comes from is no guard of
+that data. `if request.method == 'POST':` before `os.system(request.POST['cmd'])` leaves
+a vulnerability. When the flow passes the object whole, to the sink or to a function,
+a guard on any of its attributes still counts. The verdict and its evidence are in the
+finding's metadata.
+
+Django's `render_to_string(name, context)` returns HTML in which autoescaping escaped
+every variable, so data reaching a response through it is no `xss`, if the template really
+escapes. On a directory check, the analyzer reads the project's templates to decide it.
+
+It establishes escaping only when all of the following hold:
+
+- It finds every template of that name under a `templates` directory, by the path Django
+  uses (`app/page.html`), and everything those templates extend or include.
+- None of them marks output safe: `|safe`, `|safeseq` or `{% autoescape off %}`.
+- None of them uses a tag or filter beyond Django's own, or loads a library beyond its
+  built-ins (`static`, `i18n`, `l10n`, `tz`, `cache`, `humanize`).
+- None of them places a variable where HTML escaping does not protect it: a `<script>`
+  or `<style>` block, an event handler (`onclick="…"`), a `style` attribute, a URL
+  attribute not fixed by a relative or `http(s)://host/` prefix, an unquoted attribute,
+  or an attribute name.
+- No Python file of the project sets `autoescape` to `False`.
+
+The analyzer never assumes escaping where it cannot read the template. That covers a
+template named by an expression, a template found through a `DIRS` entry not named
+`templates`, one shipped by an installed package, one rendered through `get_template`,
+and Flask's Jinja2 templates. The flow is then reported as before. A single-file check
+reads no template.
 
 ### Dangerous API usage
 
@@ -202,7 +239,7 @@ the directory of the checked file; a path outside it is printed as it is.
 ```json
 {
   "schema_version": 1,
-  "tool": {"name": "coretrace-python-analyzer", "version": "0.6.0"},
+  "tool": {"name": "coretrace-python-analyzer", "version": "0.8.0"},
   "root": "/home/me/project",
   "findings": [
     {

{coretrace_python_analyzer-0.6.0 → coretrace_python_analyzer-0.8.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "coretrace-python-analyzer"
-version = "0.6.0"
+version = "0.8.0"
 description = "Static security analysis infrastructure for Python code"
 readme = "README.md"
 requires-python = ">=3.11"

{coretrace_python_analyzer-0.6.0 → coretrace_python_analyzer-0.8.0}/src/coretrace_python/__init__.py

@@ -1,4 +1,4 @@
 """CoreTrace's Python static analysis frontend."""
 
-__version__ = "0.6.0"
+__version__ = "0.8.0"
 

{coretrace_python_analyzer-0.6.0 → coretrace_python_analyzer-0.8.0}/src/coretrace_python/bundled/models/django/django_models.py

@@ -23,6 +23,7 @@ from coretrace_python.taint import (
     Source,
     SuffixSink,
     TaintKind,
+    TemplateRender,
     TypedParameter,
 )
 
@@ -119,7 +120,15 @@ class DjangoModels(ModelPlugin):
         Sink(_sym("django.http.HttpResponseRedirect"), TaintKind.REDIRECT, _TARGET_ONLY),
         Sink(_sym("django.http.HttpResponsePermanentRedirect"), TaintKind.REDIRECT, _TARGET_ONLY),
         Sanitizer(_sym("django.utils.html.escape"), TaintKind.HTML),
+        TemplateRender(_sym("django.template.loader.render_to_string")),
         Sanitizer(_sym("django.utils.html.conditional_escape"), TaintKind.HTML),
+        # The masked CSRF secret: ASCII letters and digits only, a malformed cookie is
+        # replaced before it is used.
+        Sanitizer(_sym("django.middleware.csrf.get_token"), TaintKind.ALL),
+        # A path of the project's URL configuration, never scheme-relative: attacker data
+        # in its arguments or query reaches no other host through a redirect.
+        *(Sanitizer(_sym(f"django.urls.{name}"), TaintKind.REDIRECT) for name in ("reverse", "reverse_lazy")),
+        *(Sanitizer(_sym(f"django.urls.base.{name}"), TaintKind.REDIRECT) for name in ("reverse", "reverse_lazy")),
         *(AuthorizationGuard(_sym(decorator), label) for decorator, label in _AUTHORIZATION_DECORATORS),
         # ``urlpatterns = [path('login/', views.log_in)]``: the referenced view is an
         # entry point wherever it is defined; routers register viewsets.

coretrace_python_analyzer-0.8.0/src/coretrace_python/bundled/models/http_clients/http_client_models.py

@@ -0,0 +1,52 @@
+"""Requests and httpx models: every request function is a SSRF sink for the argument that
+chooses the destination, and what it returns is an ``http-response`` source, so data
+fetched from a remote server is untrusted. The destination is the URL — first, or second
+after the method, or ``url=`` — or the prepared request a ``send`` takes; the body, the
+headers and the query parameters do not choose the host."""
+
+from __future__ import annotations
+
+from typing import ClassVar
+
+from coretrace_python.plugins import ModelPlugin
+from coretrace_python.semantic.symbols import SymbolId
+from coretrace_python.taint import Model, Sink, Source, TaintKind
+
+_METHODS = ("get", "post", "put", "patch", "delete", "head", "options", "request")
+
+_CALLERS = (
+    "requests",
+    "requests.Session",
+    "requests.api",
+    "httpx",
+    "httpx.Client",
+    "httpx.AsyncClient",
+)
+
+# Where each function takes its destination, by position (the receiver of a method
+# excluded) and by keyword, as its signature declares it.
+_DESTINATIONS = {
+    **{f"{caller}.{method}": (0, "url") for caller in _CALLERS for method in _METHODS if method != "request"},
+    **{f"{caller}.request": (1, "url") for caller in _CALLERS},
+    "httpx.stream": (1, "url"),
+    "httpx.Client.stream": (1, "url"),
+    "httpx.AsyncClient.stream": (1, "url"),
+    "requests.Session.send": (0, "request"),
+    "httpx.Client.send": (0, "request"),
+    "httpx.AsyncClient.send": (0, "request"),
+}
+
+
+def _sym(path: str) -> SymbolId:
+    return SymbolId(f"python.{path}")
+
+
+class HttpClientModels(ModelPlugin):
+    name: ClassVar[str] = "http-client-models"
+    models: ClassVar[tuple[Model, ...]] = (
+        *(
+            Sink(_sym(function), TaintKind.SSRF, ((TaintKind.SSRF, (position,)),), ((TaintKind.SSRF, (keyword,)),))
+            for function, (position, keyword) in _DESTINATIONS.items()
+        ),
+        *(Source(_sym(function), "http-response") for function in _DESTINATIONS),
+    )

{coretrace_python_analyzer-0.6.0 → coretrace_python_analyzer-0.8.0}/src/coretrace_python/cache.py

@@ -25,6 +25,7 @@ from coretrace_python.findings import Confidence, Finding, Severity
 from coretrace_python.interprocedural import (
     Arguments,
     CallSite,
+    Cleared,
     ExternalCall,
     ExternalSymbol,
     FunctionSummary,
@@ -41,7 +42,7 @@ from coretrace_python.interprocedural import (
 from coretrace_python.semantic.symbols import SymbolId
 from coretrace_python.source import SourceId, SourceSpan
 
-CACHE_FORMAT = 8
+CACHE_FORMAT = 10
 
 
 @dataclass(frozen=True)
@@ -233,10 +234,12 @@ def _encode_call(call: ExternalCall) -> dict[str, Any]:
     return {
         "symbol": str(call.symbol),
         "arguments": [sorted(deps) for deps in call.argument_dependencies],
-        "keywords": sorted(call.keyword_dependencies),
+        "keywords": [[name, sorted(deps)] for name, deps in call.keyword_dependencies],
         "location": _encode_span(call.location),
         "call_site": None if call.call_site is None else _encode_span(call.call_site),
         "given": _encode_arguments(call.arguments),
+        "argument_cleared": [_encode_cleared(c) for c in call.argument_cleared],
+        "keyword_cleared": [_encode_cleared(c) for c in call.keyword_cleared],
     }
 
 
@@ -245,13 +248,23 @@ def _decode_call(data: Mapping[str, Any]) -> ExternalCall:
     return ExternalCall(
         SymbolId(_string(data["symbol"])),
         tuple(_indices(deps) for deps in data["arguments"]),
-        _indices(data["keywords"]),
+        tuple((None if name is None else _string(name), _indices(deps)) for name, deps in data["keywords"]),
         _decode_span(data["location"]),
         None if site is None else _decode_span(site),
         _decode_arguments(data["given"]),
+        tuple(_decode_cleared(c) for c in data["argument_cleared"]),
+        tuple(_decode_cleared(c) for c in data["keyword_cleared"]),
     )
 
 
+def _encode_cleared(cleared: Cleared) -> list[list[int]]:
+    return [[parameter, bits] for parameter, bits in cleared]
+
+
+def _decode_cleared(data: Any) -> Cleared:
+    return tuple((_integer(parameter), _integer(bits)) for parameter, bits in data)
+
+
 def _encode_summary(summary: FunctionSummary) -> dict[str, Any]:
     return {
         "name": summary.name,
@@ -275,6 +288,7 @@ def _encode_summary(summary: FunctionSummary) -> dict[str, Any]:
             for w in summary.nonlocal_writes
         ],
         "static": summary.static,
+        "return_cleared": _encode_cleared(summary.return_cleared),
     }
 
 
@@ -303,6 +317,7 @@ def _decode_summary(data: Mapping[str, Any]) -> FunctionSummary:
             for w in data["nonlocal_writes"]
         ),
         bool(data["static"]),
+        _decode_cleared(data["return_cleared"]),
     )
 
 

{coretrace_python_analyzer-0.6.0 → coretrace_python_analyzer-0.8.0}/src/coretrace_python/cli.py

@@ -27,6 +27,7 @@ from coretrace_python.reporters import FORMATS, render
 from coretrace_python.semantic.imports import ImportResolutionError
 from coretrace_python.semantic.scopes import ScopeError
 from coretrace_python.source import SourceManager
+from coretrace_python.taint import ModelError
 
 EXIT_CLEAN = 0
 EXIT_FINDINGS = 1
@@ -45,6 +46,7 @@ _ANALYSIS_ERRORS = (
     AnalysisError,
     ManifestError,
     IncompatiblePluginError,
+    ModelError,
 )
 
 

{coretrace_python_analyzer-0.6.0 → coretrace_python_analyzer-0.8.0}/src/coretrace_python/engine.py

@@ -70,6 +70,7 @@ from coretrace_python.interprocedural import (
     CallGraph,
     CallGraphAnalysis,
     CallSite,
+    ClearingAnalysis,
     FunctionSummary,
     ModuleFunction,
     ModuleGraph,
@@ -109,6 +110,7 @@ from coretrace_python.source import SourceFile, SourceId, SourceManager, SourceS
 from coretrace_python.taint import (
     EntryPoint,
     EntryPointAnalysis,
+    EscapedTemplates,
     ModelTable,
     RegisteredRoutes,
     Routes,
@@ -116,6 +118,7 @@ from coretrace_python.taint import (
     SecurityModelRegistry,
     TaintAnalysis,
     TaintKind,
+    escaped_templates,
     registered_routes,
 )
 
@@ -146,6 +149,8 @@ ALL_ANALYSES: tuple[AnyAnalysis, ...] = (
     RefutationAnalysis,
     DependencyAnalysis,
     RegisteredRoutes,
+    EscapedTemplates,
+    ClearingAnalysis,
 )
 
 
@@ -178,7 +183,15 @@ class ResultsEvicted(TransformationPass):
 
     name: ClassVar[str] = "project.results-evicted"
     preserves: ClassVar[frozenset[AnyAnalysis]] = frozenset(
-        {*SEMANTIC_ANALYSES, SecurityModelAnalysis, ProjectSummaries, DependencyAnalysis, RegisteredRoutes}
+        {
+            *SEMANTIC_ANALYSES,
+            SecurityModelAnalysis,
+            ProjectSummaries,
+            DependencyAnalysis,
+            RegisteredRoutes,
+            EscapedTemplates,
+            ClearingAnalysis,
+        }
     )
 
     @classmethod
@@ -236,7 +249,9 @@ def build_manager(
     """A manager with every engine analysis registered and the engine inputs provided."""
 
     manager = _register_all(module)
-    manager.provide(SecurityModelAnalysis, (models or SecurityModelRegistry()).freeze())
+    table = (models or SecurityModelRegistry()).freeze()
+    manager.provide(SecurityModelAnalysis, table)
+    manager.provide(ClearingAnalysis, table.clearing())
     manager.provide(ProjectSummaries, SummaryIndex())
     manager.provide(RegisteredRoutes, _routes_of(manager))
     return manager
@@ -266,10 +281,14 @@ def load_plugins(plugin_roots: Sequence[Path], manager: AnalysisManager) -> Plug
     return registry
 
 
-def plugin_models(plugins: Iterable[Plugin]) -> ModelTable:
+def plugin_models(plugins: Iterable[Plugin], root: Path | None = None) -> ModelTable:
+    """The models of ``plugins``, with those they read from the project at ``root``."""
+
     models = SecurityModelRegistry()
     for plugin in plugins:
         models.register(*plugin.models, origin=plugin.name)
+        if root is not None:
+            models.register(*plugin.project_models(root), origin=plugin.name)
     return models.freeze()
 
 
@@ -284,7 +303,9 @@ def analyze_file(source: SourceFile, plugin_roots: Sequence[Path]) -> FileAnalys
 
     manager = _register_all(build_hir(source))
     registry = load_plugins(plugin_roots, manager)
-    manager.provide(SecurityModelAnalysis, plugin_models(loaded.plugin for loaded in registry))
+    table = plugin_models(loaded.plugin for loaded in registry)
+    manager.provide(SecurityModelAnalysis, table)
+    manager.provide(ClearingAnalysis, table.clearing())
     manager.provide(ProjectSummaries, SummaryIndex())
     manager.provide(RegisteredRoutes, _routes_of(manager))
     findings, supported = _check_module(manager, tuple(loaded.plugin for loaded in registry))
@@ -375,7 +396,7 @@ def analyze_project(
         (a for plugin in all_plugins for a in plugin.advisories), file_advisories
     )
     affected = affected_symbols(dependencies, advisories)
-    models = plugin_models(all_plugins).extended(*advisory_sinks(affected))
+    models = plugin_models(all_plugins, root).extended(*advisory_sinks(affected))
     for manager in managers.values():
         manager.provide(SecurityModelAnalysis, models)
         manager.provide(DependencyAnalysis, dependencies)
@@ -397,10 +418,14 @@ def analyze_project(
     for name in sorted(analysable):
         for symbol, registered in _routes_of(analysable[name]).items():
             routes.setdefault(symbol, registered)
+    escaped = escaped_templates(root)
+    clearing = models.clearing(escaped)
     for manager in analysable.values():
         manager.provide(RegisteredRoutes, routes)
+        manager.provide(EscapedTemplates, escaped)
+        manager.provide(ClearingAnalysis, clearing)
 
-    configuration = _configuration_key(registry, plugins, models, advisories, dependencies, routes)
+    configuration = _configuration_key(registry, plugins, models, advisories, dependencies, routes, escaped)
     keys = module_keys(
         graph,
         {name: fingerprint(configuration, str(files[name].source_id), name, files[name].text) for name in analysable},
@@ -441,6 +466,7 @@ def analyze_project(
                             encode_index(_seed(results, graph, component)),
                             advisory_paths,
                             _encode_routes(routes),
+                            tuple(sorted(escaped)),
                         ),
                     )
                     for component in pending
@@ -588,6 +614,7 @@ class _Batch:
     seed: Mapping[str, Any]
     advisory_paths: tuple[Path, ...] = ()
     routes: tuple[tuple[str, str, str, int], ...] = ()
+    escaped: tuple[str, ...] = ()
 
 
 def _analyse_batch(batch: _Batch) -> dict[str, dict[str, Any]]:
@@ -606,12 +633,15 @@ def _analyse_batch(batch: _Batch) -> dict[str, dict[str, Any]]:
         (a for plugin in all_plugins for a in plugin.advisories), file_advisories
     )
     affected = affected_symbols(dependencies, advisories)
-    models = plugin_models(all_plugins).extended(*advisory_sinks(affected))
+    models = plugin_models(all_plugins, batch.root).extended(*advisory_sinks(affected))
     routes = _decode_routes(batch.routes)
+    escaped = frozenset(batch.escaped)
     for manager in managers.values():
         manager.provide(SecurityModelAnalysis, models)
         manager.provide(DependencyAnalysis, dependencies)
         manager.provide(RegisteredRoutes, routes)
+        manager.provide(EscapedTemplates, escaped)
+        manager.provide(ClearingAnalysis, models.clearing(escaped))
     module_plugins = tuple(p for p in all_plugins if not isinstance(p, ProjectPlugin))
     results = _analyse_managers(managers, decode_index(batch.seed), module_plugins, affected)
     return {name: encode(entry) for name, entry in results.items()}
@@ -639,6 +669,7 @@ def _configuration_key(
     advisories: tuple[Advisory, ...],
     dependencies: DependencyGraph,
     routes: Routes | None = None,
+    escaped: frozenset[str] = frozenset(),
 ) -> str:
     """Everything a module's results depend on besides the project sources (§11)."""
 
@@ -658,6 +689,7 @@ def _configuration_key(
         repr(dependencies.requirements),
         repr(dependencies.errors),
         repr(_encode_routes(routes or {})),
+        repr(sorted(escaped)),
     )