Repository navigation
feat(audit): audit credential-disclosure reads (provider environment, inference bundle) #3017
Description
User Story
As a security or compliance owner running OpenShell, I want the gateway RPCs that hand out secret-bearing material — a sandbox's provider environment, an inference bundle — to emit a structured audit event recording who requested the disclosure and when, so that credential access is as accountable as credential change, even though the read alters no state.
Problem Statement
OpenShell is gaining a structured audit trail for state-changing gateway operations (#2911) and for authentication and authorization outcomes (#2912). Neither covers credential-disclosure reads: RPCs that return secret-bearing material to a caller without changing any state. GetSandboxProviderEnvironment returns a sandbox's resolved provider credentials, and GetInferenceBundle returns inference credential material. Today these produce no audit record. A read is not a mutation, so it sits outside #2911's mutations-only scope; a successful, authorized read is not an auth failure, so it sits outside #2912. The disclosure is therefore invisible in the audit trail.
Impact / Why This Matters
After a suspected credential leak, or during a routine access review, an operator needs to answer "who pulled this sandbox's provider credentials, and when?" and "which principal fetched the inference bundle?" Those questions have no data behind them today. Mutation is accounted for and authentication is accounted for, but the moment secret material actually leaves the gateway to a caller is not. The current workaround — correlating diagnostic logs with request traces — is not a stable audit contract and may not reliably identify the principal or the resource. The audit record must name the principal and the resource disclosed, and must never contain the disclosed secret itself.
Proposed Design
Each gateway RPC that returns secret-bearing material emits one structured OCSF audit event per attempt — on success and on failure — naming the authenticated principal (taken from the session, not the request), the resource disclosed (provider or profile id, sandbox id, inference route), the outcome, and a correlation id. The event never carries credential values, tokens, or bundle contents. It reuses the same audit helper, actor model, and master toggle that govern the control-plane audit events, so disclosure-read auditing turns on and off with the rest of the trail. The OCSF class is left to the implementer — an Entity Management (3004) access/read activity or a dedicated access event both fit. Because disclosure reads can be higher-volume than mutations, a finer toggle to gate their capture is a reasonable option, but the master audit toggle governs by default.
Acceptance Criteria
- Each gateway RPC that discloses secret-bearing material (
GetSandboxProviderEnvironment,GetInferenceBundle, and any peer added later) emits one audit record per attempt, on both success and failure. - The record names the authenticated principal, the disclosed resource (stable identifier), the outcome, and a correlation id.
- No credential value, token, or bundle content appears in any record; automated tests cover representative secret canaries.
- Emission is governed by the existing
[openshell.gateway.audit]master toggle. - Contributor documentation notes that a new secret-disclosing read RPC must emit a disclosure-audit event.
- The event, its fields, and redaction behavior are documented for operators.
Alternatives Considered
- Fold into feat(audit): emit structured gateway control-plane mutation events #2911. Rejected: reads are not mutations. Adding them to the mutation catalog blurs "what changed" with "what was accessed" and gives the two very different volume and retention characteristics one owner.
- Rely on network or proxy telemetry. A proxy can see a request but not the authenticated principal established at the gateway boundary, nor the store outcome, so it cannot attribute the disclosure.
- Accept the gap. Leaves credential access unaccountable, which is precisely the case an operator most needs to reconstruct.
Agent Investigation
- The disclosing RPCs are
GetSandboxProviderEnvironmentandGetInferenceBundlein the gateway (crates/openshell-server/src/grpc/), with the authenticated principal already available at the boundary. - The audit helper, principal→actor mapping, and master toggle being contributed for feat(audit): emit structured gateway control-plane mutation events #2911/feat(auth): emit structured gateway authentication and authorization events #2912 are directly reusable here — this is incremental instrumentation, not new infrastructure.
- This scope was explicitly deferred in the original audit-taxonomy design (recorded as out of the mutation scope). It has no home in the current decomposition, which is why it is filed as its own request.
Checklist
- I've reviewed existing issues and the architecture docs
- This is a design proposal, not a "please build this" request
Activity
krishicks commentedon Sep 3, 2026
GetInferenceBundlewill go away as part of #3172, so it should be considered out of scope for this.kvnloo commentedon Sep 27, 2026
Given the note that
GetInferenceBundleis going away in #3172, I would narrow this implementation to provider-environment credential disclosure only.For each credential read, I think the audit event should identify:
But it should never contain:
I would add a regression using a unique canary secret and assert that the canary appears nowhere in the emitted OCSF event, logs, or serialized audit payload.
That gives the audit event a useful provenance trail while explicitly proving that observing credential access cannot itself disclose the credential.
If #3654 lands first, the running-sandbox binding would also be useful context for the event.
mattj-monad commentedon Sep 30, 2026
Thanks @kvnloo, agreed on all of it: scope to provider-environment disclosure (
GetInferenceBundleis out per #3172), those fields, never the secret value, and a canary-secret test. We're planning to implement this on top of the gateway audit foundation (#2911), so i'll pick it up once that lands.sunilteja93 commentedon Oct 9, 2026
Hi @mattj-monad
I had opened a vouch request (#3827) to contribute to #3017.
I saw the updated scope around provider environment credential disclosure and the dependency on item #2911.
I’d be interested in helping with this, particularly the audit instrumentation and regression tests to ensure that credential values never appear in emitted events.
I also noticed #1955 plans to retire the legacy GetSandboxProviderEnvironment rpc. Would it make more sense to target the shared credential disclosure path instead?
Happy to take on a specific part of the implementation or testing if that would be helpful.