santati 0.2.0__tar.gz

santati-0.2.0/.gitignore

@@ -0,0 +1,19 @@
+.DS_Store
+__pycache__/
+.venv/
+.pytest_cache/
+.mypy_cache/
+.ruff_cache/
+python/dist/
+typescript/node_modules/
+typescript/dist/
+rust/target/
+elixir/_build/
+elixir/deps/
+elixir/doc/
+ruby/vendor/
+ruby/*.gem
+php/vendor/
+php/.phpunit.cache/
+erl_crash.dump
+chaos/.results/

santati-0.2.0/.python-version

@@ -0,0 +1 @@
+3.14

santati-0.2.0/CHANGELOG.md

@@ -0,0 +1,51 @@
+# Changelog
+
+All notable changes to this package are documented here. The format is based
+on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), and this project
+adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [Unreleased]
+
+## [0.2.0] - 2026-10-06
+
+### Added
+
+- An opt-in outbox: with `outbox=` (`MemoryOutbox`, or Redis Streams via
+  `santati.outbox.redis.RedisOutbox`, `santati[redis]`) `events.emit` stores the
+  event and returns at once with `EmitResult.queued` true; a background worker
+  sends the outbox in batches of `batch_size` every `flush_interval_ms`
+  through `emit_batch`, with `pre_send`/`post_send` hooks, `Santati.flush()`,
+  and `OutboxError`. `close()` now stops the worker and flushes the outbox
+  first. A queued `emit` stores a copy of the event; a batch that fails with
+  anything but a `SantatiError` (e.g. a malformed `pre_send` result) is
+  reported to `post_send` as `OutboxError` `hook_failed` and dropped, never
+  raised. Each batch is sent once, without `max_retries`: a retryable failure releases it for the next pass.
+
+### Changed
+
+- `EmitResult` has a new `queued` field, and its `event` is `None` for a
+  queued emit.
+
+### Fixed
+
+- `events.emit_batch` raises `santati.ValidationError` (with `field`) for an
+  item the request model rejects, such as an actor without `type`, instead of
+  a raw `pydantic.ValidationError`.
+
+## [0.1.0] - 2026-10-01
+
+### Added
+
+- Framework integrations under `santati.integrations`, each behind an extra:
+  Django auth signals (`santati[django]`), LangChain/LangGraph tool calls
+  (`santati[langchain]`), OpenAI Agents SDK tool calls
+  (`santati[openai-agents]`), Pydantic AI tool executions
+  (`santati[pydantic-ai]`) and Claude Agent SDK tool calls
+  (`santati[claude-agent-sdk]`). Each sends events inline through `client` or
+  hands them to a `dispatch` callable, and never raises into the framework.
+- Emit one audit event or a batch, with generated idempotency keys and retries.
+- List audit events with filters and cursor pagination, and iterate across pages.
+
+[Unreleased]: https://github.com/jamescarr/santati-sdks/compare/python-v0.2.0...HEAD
+[0.2.0]: https://github.com/jamescarr/santati-sdks/compare/python-v0.1.0...python-v0.2.0
+[0.1.0]: https://github.com/jamescarr/santati-sdks/releases/tag/python-v0.1.0

santati-0.2.0/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 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 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 those 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 2026 The Ankusa Contributors
+
+   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.

santati-0.2.0/PKG-INFO

@@ -0,0 +1,213 @@
+Metadata-Version: 2.5
+Name: santati
+Version: 0.2.0
+Summary: Official Python SDK for the Santati audit-log API: emit audit events (one or a batch) and list or iterate over them.
+Project-URL: Repository, https://github.com/jamescarr/santati-sdks
+Project-URL: Homepage, https://github.com/jamescarr/santati-sdks/tree/main/python
+Project-URL: Issues, https://github.com/jamescarr/santati-sdks/issues
+License-Expression: Apache-2.0
+License-File: LICENSE
+Keywords: audit,audit-log,events,logging,santati
+Requires-Python: >=3.10
+Requires-Dist: pydantic>=2.11
+Requires-Dist: python-dateutil>=2.8.2
+Requires-Dist: typing-extensions>=4.7.1
+Requires-Dist: urllib3<3,>=2.1.0
+Provides-Extra: claude-agent-sdk
+Requires-Dist: anyio>=4; extra == 'claude-agent-sdk'
+Requires-Dist: claude-agent-sdk>=0.1.46; extra == 'claude-agent-sdk'
+Provides-Extra: django
+Requires-Dist: django>=5.2; extra == 'django'
+Provides-Extra: langchain
+Requires-Dist: langchain-core>=1.2; extra == 'langchain'
+Provides-Extra: openai-agents
+Requires-Dist: anyio>=4; extra == 'openai-agents'
+Requires-Dist: openai-agents>=0.22.3; extra == 'openai-agents'
+Provides-Extra: pydantic-ai
+Requires-Dist: anyio>=4; extra == 'pydantic-ai'
+Requires-Dist: pydantic-ai-slim>=2.52; extra == 'pydantic-ai'
+Provides-Extra: redis
+Requires-Dist: redis>=5; extra == 'redis'
+Description-Content-Type: text/markdown
+
+# santati (Python)
+
+Official Python SDK for the [Santati](https://santati.io) audit-log API: emit
+audit events (one or a batch) and read them back with cursor pagination.
+
+## Install
+
+```sh
+pip install santati
+```
+
+## Quickstart
+
+```python
+import santati
+
+with santati.Santati("sat_sk_...", trail="billing") as client:
+    # Emit one event. A missing idempotency key is generated for you, and the
+    # request is retried on transport failures, 500/502/503/504 and 429.
+    result = client.events.emit(
+        "invoice.voided",
+        organization_id="org_acme",
+        actor={"type": "user", "id": "usr_123", "name": "Dana Ortiz"},
+        targets=[{"type": "invoice", "id": "inv_555"}],
+        data={"amount_cents": 4200, "currency": "usd"},
+    )
+    print(result.event.id, result.duplicate)
+
+    # Emit a batch: one generated key per item, per-item results on 202/207.
+    batch = client.events.emit_batch([{"event": "invoice.paid"}, {"event": "invoice.voided"}])
+    print(batch.accepted, batch.rejected)
+
+    # List a page, then iterate across every page.
+    page = client.events.list(trail="billing", limit=50)
+    for event in page.results:
+        print(event.id, event.event)
+
+    for event in client.events.iterate(trail="billing", limit=100):
+        print(event.id)
+```
+
+## Client options
+
+`Santati(api_key, *, base_url="https://api.santati.io", trail=None,
+timeout_ms=10000, max_retries=2, initial_backoff_ms=250, max_backoff_ms=8000,
+headers=None, outbox=None, batch_size=100, flush_interval_ms=1000,
+pre_send=None, post_send=None)`. `trail` is the default trail for emits and is
+never applied to reads; the last four only matter with an `outbox` (below).
+
+## Outbox
+
+With no `outbox`, `events.emit` sends the event and returns the stored one.
+Pass `outbox=` (a `santati.MemoryOutbox()` or any `OutboxStore`) and `emit`
+validates the event the same way, stores the envelope in the outbox and returns
+at once with `queued=True`, `event=None` and the idempotency key, without making
+a request. A background thread sends the outbox every `flush_interval_ms` in
+batches of `batch_size` through `emit_batch`; `client.flush()` sends it now and
+`client.close()` (or leaving the `with` block) stops the thread and flushes
+once.
+
+```python
+import redis
+import santati
+from santati.outbox.redis import RedisOutbox
+
+
+def report(event: santati.EventInput, outcome: santati.SendOutcome) -> None:
+    if outcome.status in ("rejected", "failed"):
+        print(event["idempotency_key"], outcome.error)
+
+
+with santati.Santati("sat_sk_...", trail="billing", outbox=santati.MemoryOutbox(), post_send=report) as client:
+    result = client.events.emit("invoice.voided", organization_id="org_acme")
+    assert result.queued
+
+# Survive restarts: keep the outbox in Redis (pip install "santati[redis]").
+client = santati.Santati("sat_sk_...", trail="billing", outbox=RedisOutbox(redis.Redis()))
+```
+
+`santati.MemoryOutbox(max_pending=10000)` is the in-process store; a full one
+makes a queued `emit` raise `OutboxError` (`code="outbox_full"`). `close()` sends each
+batch once; if the endpoint is down, what it could not send stays in the store and, in
+memory, is lost with the process (use `RedisOutbox` above to keep it). `pre_send(event)` may
+return a modified event or `None` to drop it; `post_send(event, outcome)`
+receives each original event with a `SendOutcome` (`accepted`, `duplicate`,
+`rejected` or `failed`). A batch that fails with `TransportError`,
+`ServerError` or `RateLimitedError` stays in the outbox for the next pass; any
+other failure drops it. Implement `santati.OutboxStore` (`enqueue`, `claim`,
+`ack`, `release`) for another backend.
+
+## Errors
+
+Every failure raises a subclass of `santati.SantatiError`: `ValidationError`,
+`AuthError`, `NotFoundError`, `RateLimitedError`, `ServerError`,
+`TransportError`, `ApiError` or `OutboxError`. Each carries `status`, `code`,
+`field`, `retry_after` and `message`; `status` is `None` for local validation
+failures, transport errors and outbox errors.
+
+`AuditEvent`, `EventActor` and `EventTarget` are the generated read models,
+re-exported from the package root. See `docs/sdk-surface.md` in the repository
+for the surface every Santati SDK implements.
+
+## Framework integrations
+
+`santati.integrations` turns the callbacks of the frameworks you already run
+into audit events. Each integration lives behind its own extra:
+
+| module | extra | records |
+|---|---|---|
+| `santati.integrations.django` | `santati[django]` | `user.logged_in`, `user.logged_out`, `user.login_failed` |
+| `santati.integrations.langchain` | `santati[langchain]` | `agent.tool_call.succeeded` / `.failed` |
+| `santati.integrations.openai_agents` | `santati[openai-agents]` | `agent.tool_call.succeeded` / `.failed` |
+| `santati.integrations.pydantic_ai` | `santati[pydantic-ai]` | `agent.tool_call.succeeded` / `.failed` |
+| `santati.integrations.claude_agent_sdk` | `santati[claude-agent-sdk]` | `agent.tool_call.succeeded` / `.failed` |
+
+```python
+from santati.integrations.django import instrument_auth
+
+
+class BillingConfig(AppConfig):
+    def ready(self) -> None:
+        instrument_auth(client, organization_id="org_acme")
+```
+
+Every integration takes exactly one of `client` — sent inline through the
+synchronous client — or `dispatch`, a callable that receives each built
+envelope (a Celery task's `delay`, an RQ queue, a thread pool):
+
+```python
+handler = SantatiCallbackHandler(dispatch=record_event.delay, organization_id="org_acme")
+agent.invoke({"messages": [...]}, config={"callbacks": [handler]})
+```
+
+The four agent adapters fix `organization_id` and `actor` per instance, so a
+multi-tenant app builds one per request — or passes it per run, which all four
+agent frameworks support. Their default actor is
+`{"type": "system", "id": <agent name or framework slug>}`; the agent name is
+also kept in `metadata.agent`. Every integration lets you override its event
+names, which teams with a declared action catalog need. Django is different:
+see its entry below.
+
+Nothing an integration does can raise into the framework that called it: a
+failed emit is logged to the `santati.integrations` logger at WARNING and
+dropped. Events are built with a fixed idempotency key, so a dispatcher's
+retry replays instead of double-recording.
+
+Per framework:
+
+- **Django** — `instrument_auth` connects to `user_logged_in`,
+  `user_logged_out` and `user_login_failed`. Call it once, from
+  `AppConfig.ready`, and never per request: it installs one process-wide set
+  of receivers and a repeated call replaces the previous instrumentation, so
+  concurrent requests would overwrite each other's settings and could record
+  one tenant's logins under another's `organization_id`. A multi-tenant app
+  passes `organization_id` as a callable `(request, user)` that returns the
+  tenant's organization, or `None`/`""` to skip the event; `user` is `None`
+  for a failed login, and `request` is `None` when `authenticate()` ran
+  without one. `actor` is likewise a callable `(request, user)` returning the
+  actor to record instead of the Django user, for logins and logouts (a
+  failed login is always `anonymous`). A failed login records the attempted
+  identifier only (`metadata.username`) and never any other credential value.
+  Pass `context` to replace the default IP/user-agent context, and set an
+  `*_event` name to `None` to leave that signal alone.
+- **LangChain / LangGraph** — `SantatiCallbackHandler` audits `ToolNode`,
+  `create_agent` and bare `tool.invoke` calls. It runs in an executor thread
+  during async runs, so emits never stall the event loop. A call LangGraph
+  pauses with `interrupt()` produces no event; the resumed run audits the call
+  that actually executes.
+- **OpenAI Agents SDK** — `SantatiRunHooks` audits local tool calls.
+  Failures are opt-in: the SDK's default formatter turns a tool exception into
+  a result nothing can observe, so a failing tool is recorded as succeeded
+  unless the tool is given `failure_error_function=record_tool_failure` (or
+  retrofitted with `set_function_tool_failure_error_function`). Hosted tools
+  and handoffs produce no events, and a nested `agent.as_tool(...)` run is
+  audited only when it is given the hooks too (`as_tool(..., hooks=hooks)`).
+- **Pydantic AI** — `SantatiCapability` audits every tool execution. A call
+  that never ran (deferred, awaiting approval, skipped) produces no event.
+- **Claude Agent SDK** — `santati_hooks(...)` returns the hook configuration for
+  `ClaudeAgentOptions(hooks=...)`: it audits `PostToolUse` and
+  `PostToolUseFailure`. Tools denied by a permission hook produce no events.
+

santati-0.2.0/README.md

@@ -0,0 +1,181 @@
+# santati (Python)
+
+Official Python SDK for the [Santati](https://santati.io) audit-log API: emit
+audit events (one or a batch) and read them back with cursor pagination.
+
+## Install
+
+```sh
+pip install santati
+```
+
+## Quickstart
+
+```python
+import santati
+
+with santati.Santati("sat_sk_...", trail="billing") as client:
+    # Emit one event. A missing idempotency key is generated for you, and the
+    # request is retried on transport failures, 500/502/503/504 and 429.
+    result = client.events.emit(
+        "invoice.voided",
+        organization_id="org_acme",
+        actor={"type": "user", "id": "usr_123", "name": "Dana Ortiz"},
+        targets=[{"type": "invoice", "id": "inv_555"}],
+        data={"amount_cents": 4200, "currency": "usd"},
+    )
+    print(result.event.id, result.duplicate)
+
+    # Emit a batch: one generated key per item, per-item results on 202/207.
+    batch = client.events.emit_batch([{"event": "invoice.paid"}, {"event": "invoice.voided"}])
+    print(batch.accepted, batch.rejected)
+
+    # List a page, then iterate across every page.
+    page = client.events.list(trail="billing", limit=50)
+    for event in page.results:
+        print(event.id, event.event)
+
+    for event in client.events.iterate(trail="billing", limit=100):
+        print(event.id)
+```
+
+## Client options
+
+`Santati(api_key, *, base_url="https://api.santati.io", trail=None,
+timeout_ms=10000, max_retries=2, initial_backoff_ms=250, max_backoff_ms=8000,
+headers=None, outbox=None, batch_size=100, flush_interval_ms=1000,
+pre_send=None, post_send=None)`. `trail` is the default trail for emits and is
+never applied to reads; the last four only matter with an `outbox` (below).
+
+## Outbox
+
+With no `outbox`, `events.emit` sends the event and returns the stored one.
+Pass `outbox=` (a `santati.MemoryOutbox()` or any `OutboxStore`) and `emit`
+validates the event the same way, stores the envelope in the outbox and returns
+at once with `queued=True`, `event=None` and the idempotency key, without making
+a request. A background thread sends the outbox every `flush_interval_ms` in
+batches of `batch_size` through `emit_batch`; `client.flush()` sends it now and
+`client.close()` (or leaving the `with` block) stops the thread and flushes
+once.
+
+```python
+import redis
+import santati
+from santati.outbox.redis import RedisOutbox
+
+
+def report(event: santati.EventInput, outcome: santati.SendOutcome) -> None:
+    if outcome.status in ("rejected", "failed"):
+        print(event["idempotency_key"], outcome.error)
+
+
+with santati.Santati("sat_sk_...", trail="billing", outbox=santati.MemoryOutbox(), post_send=report) as client:
+    result = client.events.emit("invoice.voided", organization_id="org_acme")
+    assert result.queued
+
+# Survive restarts: keep the outbox in Redis (pip install "santati[redis]").
+client = santati.Santati("sat_sk_...", trail="billing", outbox=RedisOutbox(redis.Redis()))
+```
+
+`santati.MemoryOutbox(max_pending=10000)` is the in-process store; a full one
+makes a queued `emit` raise `OutboxError` (`code="outbox_full"`). `close()` sends each
+batch once; if the endpoint is down, what it could not send stays in the store and, in
+memory, is lost with the process (use `RedisOutbox` above to keep it). `pre_send(event)` may
+return a modified event or `None` to drop it; `post_send(event, outcome)`
+receives each original event with a `SendOutcome` (`accepted`, `duplicate`,
+`rejected` or `failed`). A batch that fails with `TransportError`,
+`ServerError` or `RateLimitedError` stays in the outbox for the next pass; any
+other failure drops it. Implement `santati.OutboxStore` (`enqueue`, `claim`,
+`ack`, `release`) for another backend.
+
+## Errors
+
+Every failure raises a subclass of `santati.SantatiError`: `ValidationError`,
+`AuthError`, `NotFoundError`, `RateLimitedError`, `ServerError`,
+`TransportError`, `ApiError` or `OutboxError`. Each carries `status`, `code`,
+`field`, `retry_after` and `message`; `status` is `None` for local validation
+failures, transport errors and outbox errors.
+
+`AuditEvent`, `EventActor` and `EventTarget` are the generated read models,
+re-exported from the package root. See `docs/sdk-surface.md` in the repository
+for the surface every Santati SDK implements.
+
+## Framework integrations
+
+`santati.integrations` turns the callbacks of the frameworks you already run
+into audit events. Each integration lives behind its own extra:
+
+| module | extra | records |
+|---|---|---|
+| `santati.integrations.django` | `santati[django]` | `user.logged_in`, `user.logged_out`, `user.login_failed` |
+| `santati.integrations.langchain` | `santati[langchain]` | `agent.tool_call.succeeded` / `.failed` |
+| `santati.integrations.openai_agents` | `santati[openai-agents]` | `agent.tool_call.succeeded` / `.failed` |
+| `santati.integrations.pydantic_ai` | `santati[pydantic-ai]` | `agent.tool_call.succeeded` / `.failed` |
+| `santati.integrations.claude_agent_sdk` | `santati[claude-agent-sdk]` | `agent.tool_call.succeeded` / `.failed` |
+
+```python
+from santati.integrations.django import instrument_auth
+
+
+class BillingConfig(AppConfig):
+    def ready(self) -> None:
+        instrument_auth(client, organization_id="org_acme")
+```
+
+Every integration takes exactly one of `client` — sent inline through the
+synchronous client — or `dispatch`, a callable that receives each built
+envelope (a Celery task's `delay`, an RQ queue, a thread pool):
+
+```python
+handler = SantatiCallbackHandler(dispatch=record_event.delay, organization_id="org_acme")
+agent.invoke({"messages": [...]}, config={"callbacks": [handler]})
+```
+
+The four agent adapters fix `organization_id` and `actor` per instance, so a
+multi-tenant app builds one per request — or passes it per run, which all four
+agent frameworks support. Their default actor is
+`{"type": "system", "id": <agent name or framework slug>}`; the agent name is
+also kept in `metadata.agent`. Every integration lets you override its event
+names, which teams with a declared action catalog need. Django is different:
+see its entry below.
+
+Nothing an integration does can raise into the framework that called it: a
+failed emit is logged to the `santati.integrations` logger at WARNING and
+dropped. Events are built with a fixed idempotency key, so a dispatcher's
+retry replays instead of double-recording.
+
+Per framework:
+
+- **Django** — `instrument_auth` connects to `user_logged_in`,
+  `user_logged_out` and `user_login_failed`. Call it once, from
+  `AppConfig.ready`, and never per request: it installs one process-wide set
+  of receivers and a repeated call replaces the previous instrumentation, so
+  concurrent requests would overwrite each other's settings and could record
+  one tenant's logins under another's `organization_id`. A multi-tenant app
+  passes `organization_id` as a callable `(request, user)` that returns the
+  tenant's organization, or `None`/`""` to skip the event; `user` is `None`
+  for a failed login, and `request` is `None` when `authenticate()` ran
+  without one. `actor` is likewise a callable `(request, user)` returning the
+  actor to record instead of the Django user, for logins and logouts (a
+  failed login is always `anonymous`). A failed login records the attempted
+  identifier only (`metadata.username`) and never any other credential value.
+  Pass `context` to replace the default IP/user-agent context, and set an
+  `*_event` name to `None` to leave that signal alone.
+- **LangChain / LangGraph** — `SantatiCallbackHandler` audits `ToolNode`,
+  `create_agent` and bare `tool.invoke` calls. It runs in an executor thread
+  during async runs, so emits never stall the event loop. A call LangGraph
+  pauses with `interrupt()` produces no event; the resumed run audits the call
+  that actually executes.
+- **OpenAI Agents SDK** — `SantatiRunHooks` audits local tool calls.
+  Failures are opt-in: the SDK's default formatter turns a tool exception into
+  a result nothing can observe, so a failing tool is recorded as succeeded
+  unless the tool is given `failure_error_function=record_tool_failure` (or
+  retrofitted with `set_function_tool_failure_error_function`). Hosted tools
+  and handoffs produce no events, and a nested `agent.as_tool(...)` run is
+  audited only when it is given the hooks too (`as_tool(..., hooks=hooks)`).
+- **Pydantic AI** — `SantatiCapability` audits every tool execution. A call
+  that never ran (deferred, awaiting approval, skipped) produces no event.
+- **Claude Agent SDK** — `santati_hooks(...)` returns the hook configuration for
+  `ClaudeAgentOptions(hooks=...)`: it audits `PostToolUse` and
+  `PostToolUseFailure`. Tools denied by a permission hook produce no events.
+