* fix(providers): allow git clone/fetch via default GitHub provider
The github.com:443 git-transport endpoint used the read-only access
preset, which expands to GET/HEAD/OPTIONS only. Git smart HTTP requires
a POST to */git-upload-pack for clone and fetch, so the L7 proxy denied
those operations and `gh repo clone` / `git clone https://...` failed.
Replace the preset with explicit rules that permit the read-only methods
plus POST */git-upload-pack, so clone/fetch work while push
(git-receive-pack) stays blocked. Enabling push still requires an
explicit policy proposal.
Why allowing this POST is still read-only: in git's smart HTTP protocol
POST is an RPC transport, not a write. A clone/fetch does GET
*/info/refs (ref discovery) followed by POST */git-upload-pack, whose
body is only the client's want/have negotiation; the server responds
with a packfile and nothing on the server is modified (data flows
server -> client). The service names are from the server's perspective:
git-upload-pack = the server uploads a pack to the client (a read/
download), while git-receive-pack = the server receives a pack from the
client (the actual write/push). The new rule is scoped to
*/git-upload-pack only, so push (git-receive-pack) and arbitrary POSTs
to github.com remain denied.
Add a provider-profile regression test and a rego enforcement test
covering ref discovery, upload-pack (allowed), and receive-pack (denied).
Closes#1769
Signed-off-by: Russell Bryant <rbryant@redhat.com>
* test(providers): strengthen git-transport regression and add clone e2e
Pin the exact allowed rule set for the built-in github git-transport
endpoint in both the provider-profile and composed-policy tests, so a
broader or additional POST rule (e.g. POST **) that could enable push
via git-receive-pack fails the test instead of passing a substring
check. Add an e2e test that attaches the built-in github provider and
clones a public repo over HTTPS, exercising provider attachment,
effective-policy composition, TLS interception, and real git behavior.
Update the Providers V2 docs so the github.com git-transport endpoint
shows explicit clone/fetch rules instead of the stale read-only preset.
Refs #1769
Signed-off-by: Russell Bryant <rbryant@redhat.com>
* test(providers): isolate providers_v2 mutation in clone e2e
The clone e2e enables the gateway-global providers_v2_enabled setting.
Restore its exact prior value (or absence) captured via GetGatewayConfig
instead of unconditionally deleting it, and serialize the mutation
across xdist workers with an exclusive file lock on the run's shared
base temp dir, so a shared or pre-configured gateway is left untouched
and parallel workers cannot race the read-modify-restore.
Refs #1769
Signed-off-by: Russell Bryant <rbryant@redhat.com>
* test(providers): serialize providers_v2 mutation with a suite-wide guard
The clone e2e's per-fixture lock only coordinated fixtures that acquired
it; other xdist workers hit the same gateway without it and could
observe the transiently-enabled providers_v2_enabled global during their
own sandbox creation (CWE-362).
Add an autouse readers-writer guard in conftest: every test holds a
shared lock on the gateway config, and a test marked
exclusive_gateway_config holds an exclusive lock. Mark the clone test
exclusive so no other worker is mid-test while it enables and restores
the gateway-global setting. Exact prior-value restoration is retained.
Refs #1769
Signed-off-by: Russell Bryant <rbryant@redhat.com>
---------
Signed-off-by: Russell Bryant <rbryant@redhat.com>
* fix(helm): build chart dependencies before lint
Signed-off-by: Evan Lezar <elezar@nvidia.com>
* test(e2e): remove python gpu smoke test
Remove the Python GPU smoke test and its fixture. The e2e:k3s:gpu task only depended on e2e:python:gpu and did not have a separate k3s implementation, so remove that stale alias with the task it pointed at.
Signed-off-by: Evan Lezar <elezar@nvidia.com>
(cherry picked from commit 221a10378e188656c710560740cbc9463c002db6)
---------
Signed-off-by: Evan Lezar <elezar@nvidia.com>
Pass CLUSTER_GPU=1 inline in e2e:python:gpu's depends so that the
cluster is bootstrapped with --gpu when GPU e2e tests are run.
Add --gpu flag handling to cluster-bootstrap.sh and default
OPENSHELL_E2E_GPU_IMAGE to an empty string so the server resolves
the default sandbox image when no override is provided.
Signed-off-by: Evan Lezar <elezar@nvidia.com>
* feat(sandbox): add gpu sandbox scheduling support
Allow sandbox creation to request GPU resources explicitly or infer them from GPU image names. This wires GPU intent through bootstrap, validates gateway support, and adds dedicated GPU E2E coverage for follow-up cluster testing.
Closes#101
Add pytest-xdist for parallel e2e test execution with configurable
concurrency. Default to 5 workers; override via E2E_PARALLEL env var
(accepts a number or 'auto' for CPU-count matching). Make session-scoped
mock inference route fixtures worker-safe by incorporating the xdist
worker_id into route names and routing hints.
Co-authored-by: John Myers <johntmyers@users.noreply.github.com>
> **🏗️ build-from-issue-agent**
Closes#79
## Summary
Replaces the gateway-proxied inference model (`ProxyInference` RPC) with sandbox-local execution. The sandbox now resolves routes from either a standalone YAML route file or a cluster bundle fetched via the new `GetSandboxInferenceBundle` RPC, then forwards requests directly to inference backends using `navigator-router`.
### Benefits of removing the gRPC hop
Moving inference execution from the gateway to the sandbox eliminates the gRPC round-trip for every inference request:
- **No payload size limits**: Direct HTTP forwarding replaces protobuf-over-gRPC serialization.
- **Streaming-ready**: Preserves connection semantics for future SSE support (e.g., `stream: true`).
- **Lower latency**: Direct sandbox → backend path instead of sandbox → gateway → backend round-trip.
- **Reduced gateway load**: Gateway only serves the control-plane bundle delivery RPC.
- **Simpler error model**: HTTP status codes flow through directly without gRPC wrapping.
## Changes Made
- **Proto**: Removed `ProxyInference` RPC. Added `GetSandboxInferenceBundle` RPC for route bundle delivery.
- **Gateway**: Replaced `proxy_inference` with `get_sandbox_inference_bundle`. Removed `Router` and `navigator-router` dependency — gateway is now control-plane only for inference.
- **Router**: Switched config file format from TOML to YAML. Added `Debug` impl that redacts API keys.
- **Sandbox**: New `InferenceContext` with local `Router` + route cache. Routes loaded from `--inference-routes` file (standalone) or cluster bundle via gRPC, with 30s background refresh. New `--inference-routes` / `NAVIGATOR_INFERENCE_ROUTES` CLI arg.
- **Dev sandbox**: Added `inference-routes.yaml` at repo root with default NVIDIA NIM route. `mise run sandbox` now mounts it automatically and supports `-e` flag to forward env vars (e.g., `mise run sandbox`).
- **Example**: Added standalone example (`examples/inference/routes.yaml`) and rewrote README to cover both standalone and cluster workflows.
- **E2E tests**: Updated existing test names/docs. Added tests for Anthropic messages protocol and multi-route policy filtering.
## Tests Added
- **Unit (21 new):** `router_error_to_http` (6 variants), `load_from_file` YAML round-trip (5), `resolve_sandbox_inference_bundle` gRPC handler (6), `build_inference_context` route loading (4)
- **E2E (2 new):** Anthropic messages protocol routing, route filtering by `allowed_routes`
- **Total:** 188 unit tests passing across 3 crates
## Documentation Updated
- `architecture/gateway.md`: Removed router, updated Inference Service
- `architecture/sandbox.md`: Updated orchestration flow, InferenceContext, route loading
- `architecture/inference-routing.md`: Complete rewrite for sandbox-local architecture
## Verification
- [x] All unit tests passing (188 across 3 crates)
- [x] Pre-commit Rust checks passing (fmt, clippy, tests)
- [x] Architecture documentation updated
- [x] E2E tests updated with new test cases
- [x] Standalone example and documentation added
- [x] Default inference-routes.yaml with env passthrough for mise run sandbox
Closes#67
## Summary
Implements transparent inference interception and routing for sandboxed AI agents. The sandbox proxy intercepts outbound AI SDK calls (OpenAI, Anthropic) and reroutes them through the gateway to policy-controlled backends — enabling organizations to redirect inference traffic to local or self-hosted models without modifying agent code.
**Decision model** — a tri-state OPA evaluation for every CONNECT request:
1. Binary + endpoint explicitly allowed in `network_policies` → **allow** (pass through)
2. Not explicitly allowed + `inference.allowed_routes` configured → **inspect for inference** (TLS intercept, detect API patterns, route through gateway)
3. Otherwise → **deny**
No endpoint declarations or binary lists needed for inference routing. Just configure `inference.allowed_routes`.
## Key Changes
### Sandbox (interception)
- **OPA policy**: New `network_action` Rego rule with three outcomes (`allow`, `inspect_for_inference`, `deny`). New `NetworkAction` enum replaces `PolicyDecision.allowed` bool for the proxy's main decision path.
- **Proxy**: New `InspectForInference` path — TLS-terminates client, parses HTTP, detects inference API patterns (`POST /v1/chat/completions`, `/v1/completions`, `/v1/messages`), strips auth headers, forwards via gRPC.
- **New module**: `l7/inference.rs` — `InferenceApiPattern`, `detect_inference_pattern()`, HTTP request/response parsing.
- **gRPC client**: New `proxy_inference()` for sandbox→gateway forwarding.
- **Sandbox init**: Creates OPA engine when inference is configured, even without `network_policies`.
### Gateway (dispatch)
- **InferenceService**: `ProxyInference` RPC loads sandbox policy, resolves allowed routes, dispatches to router. Full CRUD for inference routes.
- **Proto**: `InferenceRoute`, `InferenceRouteSpec`, `ProxyInferenceRequest/Response`, Inference gRPC service.
### Router (backend proxying)
- **New crate**: `navigator-router` with `Router`, `proxy_with_candidates()`, protocol-based route selection, backend HTTP proxying with auth header rewriting.
- **Mock support** for testing (`mock://` scheme).
### CLI
- `nav inference create/update/delete/list` commands for route management.
### Python SDK
- Updated protobuf bindings. Removed old `inference.py` client (replaced by transparent interception).
### Documentation
- New `architecture/inference-routing.md` — end-to-end system documentation.
- Updated `architecture/sandbox.md` — proxy, OPA, and source index sections.
- Updated `architecture/README.md` — new subsystem overview and diagram.
## Addendum: Chunked Transfer Compatibility
This branch now also fixes intercepted SDK requests that send chunked request bodies:
- `inspect_for_inference` now accepts `Transfer-Encoding: chunked` and decodes chunked request bodies before forwarding to the gateway
- Removed the prior `411 Length Required` response for chunked intercepted requests
- Added request/response header sanitization for framing and hop-by-hop headers (`content-length`, `transfer-encoding`, `connection`, etc.) to keep forwarded requests and returned responses valid
- Added unit tests for chunked parsing and header sanitization
Note: this improves compatibility for streaming-style SDK request patterns; true token-by-token passthrough response streaming is still a separate follow-up.
## Minimal Policy for Inference Routing
```yaml
inference:
allowed_routes:
- local
```
Any outgoing connection from a binary not explicitly allowed in `network_policies` will be intercepted and checked for inference API patterns.
## Test Plan
- [x] `cargo test --workspace` — all tests pass
- [x] `mise run pre-commit` — all checks pass
- [x] E2E: OpenAI chat completions routed through gateway
- [x] E2E: Anthropic messages routed through gateway
- [x] E2E: Python OpenAI SDK from sandbox (`examples/inference/inference.py`)
- [x] E2E test: `e2e/python/test_inference_routing.py`
Closes#13
## Summary
- add Python sandbox execution APIs for command and callable workflows
- consolidate sandbox policy fixtures and expand e2e test coverage for policy and Python exec paths
- update CI/build config and images for sandbox e2e execution dependencies
## Test Plan
- mise run pre-commit
Russell Bryant