fix: restore cluster_pods network policy for nemoclaw and openclaw#25
Merged
fix: restore cluster_pods network policy for nemoclaw and openclaw#25
Conversation
The cluster_pods allowed_ips policy was accidentally removed in #24. This policy allows sandbox binaries to reach services on the k3s cluster pod network (10.42.0.0/16), which is required for internal service communication.
factory-octavian
pushed a commit
to factory-octavian/OpenShell-Community
that referenced
this pull request
Apr 1, 2026
## Summary Implements L7 (application-layer) protocol-aware policy enforcement for the sandbox proxy, enabling per-request allow/deny decisions based on HTTP method and path — not just host:port. ### Phase 1: HTTP/REST L7 Inspection (plaintext) - New `l7/` module with provider trait, REST parser, and relay loop - Parses HTTP/1.1 requests inside CONNECT tunnels, evaluates method+path against OPA/Rego policy - Supports `access` presets (`full`, `read-only`, `read-write`) and explicit `rules` with glob path matching - `enforcement` modes: `enforce` (deny with 403 JSON) vs `audit` (log only, forward traffic) - Structured logging for every L7 decision (`L7_REQUEST` with protocol, action, target, decision) ### Phase 2: MITM TLS Termination for HTTPS - Ephemeral CA generated at sandbox startup (`rcgen`) - Per-hostname leaf certificate cache (256 cap) for dynamic cert presentation - Client-side TLS termination (sandbox trusts ephemeral CA via `SSL_CERT_FILE`, `NODE_EXTRA_CA_CERTS`, etc.) - Upstream TLS connection with `webpki-roots` verification - Endpoints with `tls: terminate` get full L7 inspection; others remain passthrough ### Additional - Benign TLS/connection errors (close_notify, handshake EOF, reset) downgraded from WARN to DEBUG - Generic `AsyncRead + AsyncWrite` bounds throughout L7 module (works with both `TcpStream` and `TlsStream`) - Proto schema extended with `protocol`, `tls`, `enforcement`, `access`, `rules` fields on `NetworkEndpoint` - CLI `nav sandbox run` wired with `--rego-policy`/`--rego-data` flags and L7 policy display ### E2E Test Coverage - 7 L4 tests: no-matching-policy deny, wildcard binary, binary-restricted, wrong port, cross-policy isolation, non-CONNECT 405, structured log fields - 7 L7 TLS tests: full access allow, read-only deny POST, audit mode, explicit path rules, CA trust store injection, deny response JSON format, structured log fields ## Key Files | Area | Files | |------|-------| | L7 core | `l7/mod.rs`, `l7/provider.rs`, `l7/relay.rs`, `l7/rest.rs` | | TLS termination | `l7/tls.rs` | | Proxy integration | `proxy.rs`, `lib.rs`, `main.rs` | | Env/process | `process.rs`, `ssh.rs` | | OPA policy | `opa.rs`, `dev-sandbox-policy.rego` | | Proto schema | `proto/sandbox.proto` | | CLI | `navigator-cli/src/run.rs` | | E2E tests | `e2e/python/test_sandbox_policy.py` | ## Test plan - [x] 67 unit tests pass (`cargo test --workspace`) - [x] `cargo clippy --workspace --all-targets` clean - [x] 16 e2e policy tests pass (`mise run test:e2e:sandbox`) - [x] Manual smoke test: sandbox with `tls: terminate` on `api.anthropic.com:443` Closes NVIDIA#25
factory-octavian
pushed a commit
to factory-octavian/OpenShell-Community
that referenced
this pull request
Apr 1, 2026
…Policy, and nonce replay detection (#127) * fix(security): harden sandbox SSH with mandatory HMAC secret, NetworkPolicy, and nonce replay detection Closes NVIDIA#25 - Make NEMOCLAW_SSH_HANDSHAKE_SECRET mandatory: server and sandbox both refuse to start if the secret is empty/unset. Cluster deployments auto-generate it via openssl rand in the entrypoint script. - Add Kubernetes NetworkPolicy restricting sandbox port 2222 ingress to the gateway pod only, preventing lateral movement from other cluster workloads. - Add NSSH1 nonce replay detection with a TTL-bounded cache, rejecting replayed handshakes within the timestamp validity window. - Add unit tests for verify_preface (valid, replay, expired, bad HMAC, malformed) and env injection. * fix(deploy): pass sshHandshakeSecret in fast deploy helm upgrade --------- Co-authored-by: John Myers <johntmyers@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
cluster_podsallowed_ips network policy that was accidentally removed from nemoclaw and openclaw in chore: align sandbox tooling and policies with upstream OpenShell #24.10.42.0.0/16on port 8080), which is required for internal service communication.