onedoor 0.3.0__tar.gz

onedoor-0.3.0/LICENSE

@@ -0,0 +1,202 @@
+
+                                 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
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   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.

onedoor-0.3.0/PKG-INFO

@@ -0,0 +1,201 @@
+Metadata-Version: 2.4
+Name: onedoor
+Version: 0.3.0
+Summary: A tiered guardrail/policy engine for agentic systems: default-deny, bounds, caps, approvals, dry-run, kill switch, and reversibility as a precondition for autonomy.
+Author: Shamik Saha
+License: Apache-2.0
+Requires-Python: >=3.12
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: pydantic>=2.10
+Requires-Dist: pydantic-settings>=2.7
+Requires-Dist: pyyaml>=6.0
+Provides-Extra: service
+Requires-Dist: fastapi>=0.115; extra == "service"
+Requires-Dist: uvicorn>=0.34; extra == "service"
+Provides-Extra: otel
+Requires-Dist: opentelemetry-api>=1.27; extra == "otel"
+Provides-Extra: dev
+Requires-Dist: pytest>=8.3; extra == "dev"
+Requires-Dist: mypy>=1.14; extra == "dev"
+Requires-Dist: ruff>=0.8; extra == "dev"
+Requires-Dist: fastapi>=0.115; extra == "dev"
+Requires-Dist: httpx>=0.27; extra == "dev"
+Dynamic: license-file
+
+# onedoor
+
+**A tiered guardrail engine for agentic systems.**
+The model proposes; the policy layer disposes.
+
+Every action in an agentic system — scheduled, rule-fired, LLM-proposed, or
+human-clicked — is a structured `ActionRequest` evaluated by one executor
+against a policy table before anything touches the world. There is one door.
+Nothing else is allowed to call a connector.
+
+```
+kill switch → policy lookup / default-deny → tier-1 integrity (no undo, no
+autonomy) → bounds → dry-run → caps → two-phase execute → append-only audit
+```
+
+## Why another guardrail project?
+
+Most "guardrails" govern what a model may *say*. This engine governs what an
+agent may *do* — and it takes positions most frameworks leave as wishes:
+
+- **Default-deny.** An unlisted action type is not an error and not a pass:
+  it resolves to propose-and-confirm, with the reason recorded.
+- **Reversibility is a precondition for autonomy.** An auto-tier action whose
+  policy declares no compensating command is demoted to human approval at
+  runtime — and the policy loader refuses to boot if a Tier-1 entry lacks one.
+  Undo is not a feature; it is the admission ticket to auto-execution.
+- **The kill switch outranks everything, including prior consent.** Checked
+  before policy lookup; an already-approved action arriving while the switch
+  is engaged is blocked (without spawning an approval loop). Reads stay exempt
+  — you want visibility *during* the incident.
+- **Bounds are validated before a human ever sees a proposal**, so the
+  approval screen can only contain physically sane requests. The human decides
+  *whether*, never has to catch *whether it's insane*.
+- **Rehearsal must not spend a real budget.** Dry-run is resolved before cap
+  accounting; new action types start in dry-run and log "would have executed".
+- **Caps are reserved race-free** inside the deciding transaction
+  (`BEGIN IMMEDIATE`), so two concurrent requests cannot share the last slot.
+- **Two-phase execution.** Tx A decides, reserves caps, and records intent;
+  the connector call runs outside any DB lock under a hard timeout; Tx B
+  appends the result. A hung smart-plug API cannot hold the engine hostage,
+  and a crash leaves an honest "intended, unconfirmed" trail.
+- **The audit log is append-only** — decisions, results, denials, dry-runs,
+  and kill-switch blocks, all with typed reason codes, never updated in place.
+- **Policies are data, not code** (`config/policies.yaml`): tiers, bounds,
+  caps, undo windows, dry-run flags. Changing what's allowed never means
+  changing the engine.
+
+## Tiers
+
+| Tier | Meaning | Example policy |
+|------|---------|----------------|
+| 0 | observe only | reads (exempt from the kill switch) |
+| 1 | auto-execute, reversible, in-bounds | toggle with `compensating_command` + 15-min undo |
+| 2 | auto-execute under cumulative caps | rate + €/day + €/month budgets |
+| 3 | propose-and-confirm (TTL'd approval) | anything irreversible, unlisted, or over cap |
+
+## Documentation
+
+Developer guides live in [`docs/`](docs/index.md): the three-minute mental
+model, an integration guide per surface — [library](docs/integration-library.md),
+[HTTP decision service](docs/integration-service.md),
+[MCP proxy](docs/integration-mcp.md),
+[LiteLLM adapter](docs/integration-litellm.md) — and the full
+[policy reference](docs/policy-reference.md).
+
+## Quickstart
+
+Requires Python ≥ 3.12.
+
+```bash
+pip install -e ".[dev]"
+pytest                    # 63 tests — the guardrail suite is the release blocker
+python -m scripts.demo    # one of everything, end to end, zero external deps
+```
+
+The demo walks the whole surface: auto-execution and undo, default-deny into a
+real approval that then executes, a bounds rejection, cap exhaustion, dry-run,
+and the kill switch clamping an auto action to propose-and-confirm.
+
+## A policy, concretely
+
+```yaml
+- action_type: ha.set_climate
+  tier: 1
+  dry_run: true                      # new action types rehearse first
+  compensating_command: ha.restore_climate
+  bounds:
+    numeric:
+      temperature: { min: 17, max: 23 }
+    required: [entity_id, temperature]
+    strict_params: true
+```
+
+## v0.2 — the decision/enforcement split, and the engine on other people's doors
+
+v0.2 separates the engine into the classic authorization pair — a **Policy
+Decision Point** and **Policy Enforcement Points** — without changing a single
+decision's semantics (the v0.1 suite passes unchanged):
+
+- `decision.decide_and_reserve(request, ...)` — Tx A: the full ordered check
+  pipeline, cap reservation, and the intent row in the audit log. Returns
+  either a terminal result (denied / proposed / dry-run) or a
+  `PermittedIntent`: an obligation the caller must enforce.
+- `decision.report_result(intent, ok, ...)` — Tx B: the linked, append-only
+  execution receipt, whatever happened.
+
+The in-process executor is now literally these two phases composed around a
+connector call. Any other enforcement point — a gateway filter, a tool
+wrapper — composes them around its own act.
+
+**The first external enforcement point ships with it: an MCP proxy.**
+`onedoor.mcp.proxy` speaks MCP's stdio transport on both sides: an agent host
+connects to it as if it were the tool server; it spawns the real server as a
+subprocess and forwards everything except `tools/call`, which becomes an
+`ActionRequest` (`mcp.<tool>`) through the full pipeline — unknown tools
+default-deny to a human, bounds are checked before the tool ever sees the
+call, money waits for approval, and the kill switch clamps everything at once.
+
+```bash
+python -m scripts.demo_mcp   # an agent's-eye view: 7 calls, every mechanism
+```
+
+This makes the engine usable with agents you don't control: point any MCP
+host at the proxy instead of the tool server, write a policy file, done.
+(The proxy's `onedoor/approve` and `onedoor/kill` JSON-RPC methods are demo
+conveniences, not part of MCP.)
+
+## Using it from an AI gateway (LiteLLM example)
+
+`examples/litellm_guardrail.py` is an experimental adapter showing the engine
+as a LiteLLM custom guardrail: `async_pre_call_hook` governs completions
+(model allow-list as *value* bounds, daily caps) and — because LiteLLM routes
+its MCP gateway's tool calls through the same hook (`call_type="call_mcp_tool"`)
+— every MCP tool call, with default-deny, bounds, tier-3 approval and the kill
+switch. Run `python -m examples.litellm_guardrail` for a proxy-free self-test.
+What this adds over the gateway's built-in MCP ACLs: decisions beyond
+allow/deny (defer with an approval id, dry-run), value-level bounds rather
+than parameter-name lists, race-free caps, and an audit row with a reason for
+every decision. `litellm` is not a dependency of this package — the example
+imports it only if you have it.
+
+## The decision service (v0.3)
+
+The PDP over HTTP, so any enforcement point in any language can consult the
+engine:
+
+```bash
+pip install "onedoor[service]"
+ONEDOOR_DECIDE_KEYS=dev ONEDOOR_ADMIN_KEYS=root \
+ONEDOOR_POLICIES=config/policies.yaml \
+uvicorn onedoor.service.app:create_app --factory --port 8470
+```
+
+`POST /v1/decide` returns the decision; a permitted one carries an
+`intent_audit_id` — enforce, then `POST /v1/report` the outcome. Approvals,
+denial and the kill switch live under admin-role keys (`ONEDOOR_ADMIN_KEYS`),
+separate from decide-role keys by design: the process that asks for permission
+should not be the process that grants it. Tier-3 proposals can notify a
+webhook (`ONEDOOR_APPROVAL_WEBHOOK`, Slack-compatible payload), and installing
+`onedoor[otel]` lights up OpenTelemetry spans and decision counters with no
+code changes. See `ROADMAP.md` for where this is going (tenancy, Postgres,
+OIDC, audit hardening).
+
+## Origin & status
+
+Extracted from a personal single-user control plane (home/energy/money with an
+LLM agent layer), where this engine has governed every action since July 2026 —
+the domain modules stayed home; the engine, its mock connector, its demo action
+types, and its full test suite are what you see here. v0.2: SQLite-backed,
+single-process, synchronous; PDP/PEP split with an MCP proxy as the first
+external enforcement point. Deliberately boring technology; the design is the
+contribution.
+
+## License
+
+Apache-2.0.

onedoor-0.3.0/README.md

@@ -0,0 +1,176 @@
+# onedoor
+
+**A tiered guardrail engine for agentic systems.**
+The model proposes; the policy layer disposes.
+
+Every action in an agentic system — scheduled, rule-fired, LLM-proposed, or
+human-clicked — is a structured `ActionRequest` evaluated by one executor
+against a policy table before anything touches the world. There is one door.
+Nothing else is allowed to call a connector.
+
+```
+kill switch → policy lookup / default-deny → tier-1 integrity (no undo, no
+autonomy) → bounds → dry-run → caps → two-phase execute → append-only audit
+```
+
+## Why another guardrail project?
+
+Most "guardrails" govern what a model may *say*. This engine governs what an
+agent may *do* — and it takes positions most frameworks leave as wishes:
+
+- **Default-deny.** An unlisted action type is not an error and not a pass:
+  it resolves to propose-and-confirm, with the reason recorded.
+- **Reversibility is a precondition for autonomy.** An auto-tier action whose
+  policy declares no compensating command is demoted to human approval at
+  runtime — and the policy loader refuses to boot if a Tier-1 entry lacks one.
+  Undo is not a feature; it is the admission ticket to auto-execution.
+- **The kill switch outranks everything, including prior consent.** Checked
+  before policy lookup; an already-approved action arriving while the switch
+  is engaged is blocked (without spawning an approval loop). Reads stay exempt
+  — you want visibility *during* the incident.
+- **Bounds are validated before a human ever sees a proposal**, so the
+  approval screen can only contain physically sane requests. The human decides
+  *whether*, never has to catch *whether it's insane*.
+- **Rehearsal must not spend a real budget.** Dry-run is resolved before cap
+  accounting; new action types start in dry-run and log "would have executed".
+- **Caps are reserved race-free** inside the deciding transaction
+  (`BEGIN IMMEDIATE`), so two concurrent requests cannot share the last slot.
+- **Two-phase execution.** Tx A decides, reserves caps, and records intent;
+  the connector call runs outside any DB lock under a hard timeout; Tx B
+  appends the result. A hung smart-plug API cannot hold the engine hostage,
+  and a crash leaves an honest "intended, unconfirmed" trail.
+- **The audit log is append-only** — decisions, results, denials, dry-runs,
+  and kill-switch blocks, all with typed reason codes, never updated in place.
+- **Policies are data, not code** (`config/policies.yaml`): tiers, bounds,
+  caps, undo windows, dry-run flags. Changing what's allowed never means
+  changing the engine.
+
+## Tiers
+
+| Tier | Meaning | Example policy |
+|------|---------|----------------|
+| 0 | observe only | reads (exempt from the kill switch) |
+| 1 | auto-execute, reversible, in-bounds | toggle with `compensating_command` + 15-min undo |
+| 2 | auto-execute under cumulative caps | rate + €/day + €/month budgets |
+| 3 | propose-and-confirm (TTL'd approval) | anything irreversible, unlisted, or over cap |
+
+## Documentation
+
+Developer guides live in [`docs/`](docs/index.md): the three-minute mental
+model, an integration guide per surface — [library](docs/integration-library.md),
+[HTTP decision service](docs/integration-service.md),
+[MCP proxy](docs/integration-mcp.md),
+[LiteLLM adapter](docs/integration-litellm.md) — and the full
+[policy reference](docs/policy-reference.md).
+
+## Quickstart
+
+Requires Python ≥ 3.12.
+
+```bash
+pip install -e ".[dev]"
+pytest                    # 63 tests — the guardrail suite is the release blocker
+python -m scripts.demo    # one of everything, end to end, zero external deps
+```
+
+The demo walks the whole surface: auto-execution and undo, default-deny into a
+real approval that then executes, a bounds rejection, cap exhaustion, dry-run,
+and the kill switch clamping an auto action to propose-and-confirm.
+
+## A policy, concretely
+
+```yaml
+- action_type: ha.set_climate
+  tier: 1
+  dry_run: true                      # new action types rehearse first
+  compensating_command: ha.restore_climate
+  bounds:
+    numeric:
+      temperature: { min: 17, max: 23 }
+    required: [entity_id, temperature]
+    strict_params: true
+```
+
+## v0.2 — the decision/enforcement split, and the engine on other people's doors
+
+v0.2 separates the engine into the classic authorization pair — a **Policy
+Decision Point** and **Policy Enforcement Points** — without changing a single
+decision's semantics (the v0.1 suite passes unchanged):
+
+- `decision.decide_and_reserve(request, ...)` — Tx A: the full ordered check
+  pipeline, cap reservation, and the intent row in the audit log. Returns
+  either a terminal result (denied / proposed / dry-run) or a
+  `PermittedIntent`: an obligation the caller must enforce.
+- `decision.report_result(intent, ok, ...)` — Tx B: the linked, append-only
+  execution receipt, whatever happened.
+
+The in-process executor is now literally these two phases composed around a
+connector call. Any other enforcement point — a gateway filter, a tool
+wrapper — composes them around its own act.
+
+**The first external enforcement point ships with it: an MCP proxy.**
+`onedoor.mcp.proxy` speaks MCP's stdio transport on both sides: an agent host
+connects to it as if it were the tool server; it spawns the real server as a
+subprocess and forwards everything except `tools/call`, which becomes an
+`ActionRequest` (`mcp.<tool>`) through the full pipeline — unknown tools
+default-deny to a human, bounds are checked before the tool ever sees the
+call, money waits for approval, and the kill switch clamps everything at once.
+
+```bash
+python -m scripts.demo_mcp   # an agent's-eye view: 7 calls, every mechanism
+```
+
+This makes the engine usable with agents you don't control: point any MCP
+host at the proxy instead of the tool server, write a policy file, done.
+(The proxy's `onedoor/approve` and `onedoor/kill` JSON-RPC methods are demo
+conveniences, not part of MCP.)
+
+## Using it from an AI gateway (LiteLLM example)
+
+`examples/litellm_guardrail.py` is an experimental adapter showing the engine
+as a LiteLLM custom guardrail: `async_pre_call_hook` governs completions
+(model allow-list as *value* bounds, daily caps) and — because LiteLLM routes
+its MCP gateway's tool calls through the same hook (`call_type="call_mcp_tool"`)
+— every MCP tool call, with default-deny, bounds, tier-3 approval and the kill
+switch. Run `python -m examples.litellm_guardrail` for a proxy-free self-test.
+What this adds over the gateway's built-in MCP ACLs: decisions beyond
+allow/deny (defer with an approval id, dry-run), value-level bounds rather
+than parameter-name lists, race-free caps, and an audit row with a reason for
+every decision. `litellm` is not a dependency of this package — the example
+imports it only if you have it.
+
+## The decision service (v0.3)
+
+The PDP over HTTP, so any enforcement point in any language can consult the
+engine:
+
+```bash
+pip install "onedoor[service]"
+ONEDOOR_DECIDE_KEYS=dev ONEDOOR_ADMIN_KEYS=root \
+ONEDOOR_POLICIES=config/policies.yaml \
+uvicorn onedoor.service.app:create_app --factory --port 8470
+```
+
+`POST /v1/decide` returns the decision; a permitted one carries an
+`intent_audit_id` — enforce, then `POST /v1/report` the outcome. Approvals,
+denial and the kill switch live under admin-role keys (`ONEDOOR_ADMIN_KEYS`),
+separate from decide-role keys by design: the process that asks for permission
+should not be the process that grants it. Tier-3 proposals can notify a
+webhook (`ONEDOOR_APPROVAL_WEBHOOK`, Slack-compatible payload), and installing
+`onedoor[otel]` lights up OpenTelemetry spans and decision counters with no
+code changes. See `ROADMAP.md` for where this is going (tenancy, Postgres,
+OIDC, audit hardening).
+
+## Origin & status
+
+Extracted from a personal single-user control plane (home/energy/money with an
+LLM agent layer), where this engine has governed every action since July 2026 —
+the domain modules stayed home; the engine, its mock connector, its demo action
+types, and its full test suite are what you see here. v0.2: SQLite-backed,
+single-process, synchronous; PDP/PEP split with an MCP proxy as the first
+external enforcement point. Deliberately boring technology; the design is the
+contribution.
+
+## License
+
+Apache-2.0.

onedoor-0.3.0/onedoor/__init__.py

@@ -0,0 +1 @@
+"""OneDoor — a tiered guardrail engine for agentic systems. The model proposes; the policy layer disposes."""

onedoor-0.3.0/onedoor/config.py

@@ -0,0 +1,19 @@
+"""Engine settings — the minimal surface the guardrail engine consumes.
+
+The parent project carries a much larger Settings object; the engine only ever
+reads these fields (via ``EngineConfig.from_settings``). Override with
+environment variables prefixed ``ONEDOOR_`` or construct directly in code.
+"""
+
+from __future__ import annotations
+
+from pydantic_settings import BaseSettings, SettingsConfigDict
+
+
+class Settings(BaseSettings):
+    model_config = SettingsConfigDict(env_prefix="ONEDOOR_", extra="ignore")
+
+    db_path: str = "onedoor.db"
+    timezone: str = "Europe/Amsterdam"
+    approval_ttl_seconds: int = 3600
+    connector_timeout_seconds: float = 10.0

onedoor-0.3.0/onedoor/connectors/__init__.py

@@ -0,0 +1,6 @@
+"""Connectors expose read_* (pure reads) and act_* (side effects).
+
+act_* functions may only be invoked by the guardrail executor, via the injected
+ConnectorRegistry. In M0 the only connector is the mock used to exercise the
+engine and its tests — no real device/service connectors exist yet.
+"""

onedoor-0.3.0/onedoor/connectors/errors.py

@@ -0,0 +1,11 @@
+"""Connector exceptions. Connectors raise these; schedulers catch and fail soft."""
+
+from __future__ import annotations
+
+
+class ConnectorError(Exception):
+    """A connector read/act failed (HTTP, timeout, parse)."""
+
+
+class NotConfigured(ConnectorError):
+    """The connector's credentials/URL are not set — skip, don't crash."""

onedoor-0.3.0/onedoor/connectors/mock.py

@@ -0,0 +1,46 @@
+"""Mock connector — exercises the guardrail engine without any real integration.
+
+Registered into a :class:`ConnectorRegistry` for the M0 demo and the test suite.
+The ``act_flaky`` / ``act_slow`` functions exist purely to test fail-soft handling.
+"""
+
+from __future__ import annotations
+
+import time
+
+from onedoor.guardrail.models import JsonValue
+from onedoor.guardrail.registry import ConnectorRegistry
+
+
+def act_toggle(params: dict[str, JsonValue]) -> dict[str, JsonValue]:
+    return {"toggled": True, "target": params.get("target"), "state": params.get("state")}
+
+
+def act_restore(params: dict[str, JsonValue]) -> dict[str, JsonValue]:
+    return {"restored": True, "target": params.get("target")}
+
+
+def act_ok(params: dict[str, JsonValue]) -> dict[str, JsonValue]:
+    return {"ok": True}
+
+
+def act_flaky(params: dict[str, JsonValue]) -> dict[str, JsonValue]:
+    raise RuntimeError("simulated connector failure")
+
+
+def act_slow(params: dict[str, JsonValue]) -> dict[str, JsonValue]:
+    time.sleep(30)  # exceeds the connector timeout — used to test timeout handling
+    return {"ok": True}
+
+
+def build_registry() -> ConnectorRegistry:
+    """Registry for the M0 demo action types (mirrors config/policies.yaml)."""
+    registry = ConnectorRegistry()
+    registry.register("demo.toggle", act_toggle)
+    registry.register("demo.dry", act_toggle)
+    registry.register("demo.capped", act_ok)
+    registry.register("demo.restore", act_restore)
+    # An unlisted action type (default-deny -> Tier 3) that still has a connector,
+    # so approving it actually executes — demonstrating the full approval loop.
+    registry.register("demo.unlisted", act_ok)
+    return registry