Skip to content

fix(auth): honor NOUS_INFERENCE_BASE_URL env override for Nous OAuth sessions - #52270

Merged
teknium1 merged 1 commit into
mainfrom
fix/nous-inference-env-override
Jun 25, 2026
Merged

fix(auth): honor NOUS_INFERENCE_BASE_URL env override for Nous OAuth sessions#52270
teknium1 merged 1 commit into
mainfrom
fix/nous-inference-env-override

Conversation

@benbarclay

Copy link
Copy Markdown
Collaborator

Summary

The NOUS_INFERENCE_BASE_URL dev/staging escape hatch is unreachable for OAuth (Nous Portal login) sessions, even though three separate code comments assert it still works. A user pointing Hermes at staging via the env var silently gets production (or 401s against a dead endpoint). This restores the documented behavior.

Root cause

The host-allowlist hardening (#30611) plus the refresh heal (#49735) interact with the resolution precedence in resolve_nous_runtime_credentials:

inference_base_url = (
    _optional_base_url(state.get("inference_base_url"))   # stored — wins
    or os.getenv("NOUS_INFERENCE_BASE_URL")               # env — never reached
    or DEFAULT_NOUS_INFERENCE_URL
)

A staging OAuth login persists its inference_base_url, but:

  1. the allowlist (#30611) rejects the staging host, and
  2. the refresh heal (#49735) rewrites the stored value to the production default.

The stored (now prod) value is read before the env var, so the override never takes effect. The comments in #30611/#49735 claim the env path is the trusted escape hatch — but precedence makes it dead for any already-authenticated OAuth session.

Fix

The user-set env override is the most-trusted source, so consult it first for the URL used to build the client / returned to callers — while keeping the persisted value the validated, network-provenance one. The override is a runtime overlay: it is never written to auth.json, so unsetting it cleanly reverts to the stored/default production URL (threat model intact).

Applied at both chokepoints (this is the whole bug class, not just the one site):

  • resolve_nous_runtime_credentials — both the no-refresh read path and the refresh path. Persistence keeps the validated network value; the returned/client URL is env-first.
  • hermes_cli/proxy/adapters/nous_portal.py — the proxy adapter re-validates the resolver's returned base_url against the prod allowlist as defense-in-depth, and would otherwise reject a legitimate staging override at the forward boundary. Same env-first overlay applied; the validator stays in place.

New _nous_inference_env_override() helper centralizes the (deliberately ungated) env read. Portal-returned URLs are still allowlist-validated at every network site.

Relationship to open PRs

Audit of sibling env vars (requested)

Env var Verdict
NOUS_INFERENCE_BASE_URL Was broken — fixed here.
HERMES_PORTAL_BASE_URL Honored at login (env-first). Stored-first at runtime-refresh, but that's intentional — a runtime portal override must match the token issuer (see _resolve_portal_base_url docstring), so overriding mid-session would break auth. Not a bug.
TOOL_GATEWAY_DOMAIN Read directly via os.getenv, no stored value, always honored. No shadowing bug. (Separate UX gap: not derived from the portal URL — out of scope.)
HERMES_DASHBOARD_PORTAL_URL Override-first (_resolve_portal_base_url). Correct — the model this fix follows.

Tests

  • TestEnvOverrideWins — env wins over a prod-pinned stored value on both the no-refresh and refresh paths; override never persisted to state; poisoned stored host still heals when no env is set.
  • TestProxyAdapterEnvOverride — forward-boundary consults the override before the validator.
  • Verified the 4 behavioral tests fail against pre-fix code and pass with the fix.
  • Full test_nous_inference_url_validation.py + test_auth_nous_provider.py: 85 passed. (One existing idempotency fixture used a non-allowlisted host incidentally; updated it to a realistic allowlisted host so the new read-path heal doesn't trip its "no write on unchanged" assertion.)
  • E2E against a real temp HERMES_HOME exercising the real resolver + real proxy adapter: resolver→staging, persisted→prod, proxy→staging, unset→reverts to prod.

How to test

pytest tests/hermes_cli/test_nous_inference_url_validation.py -v
# Manual: log into Nous (prod), then
NOUS_INFERENCE_BASE_URL=https://stg-inference-api.nousresearch.com/v1 hermes status
# inference URL should now reflect staging; auth.json keeps the prod host.

Type of Change

  • 🐛 Bug fix (non-breaking)
  • ✅ Tests

NOUS_INFERENCE_BASE_URL escape hatch restored

…sessions

The host-allowlist hardening (#30611) plus the refresh heal (#49735) left
the documented NOUS_INFERENCE_BASE_URL dev/staging escape hatch unreachable
for OAuth sessions, despite three code comments asserting it still works.

Root cause — resolution precedence in resolve_nous_runtime_credentials:

    inference_base_url = (
        _optional_base_url(state.get("inference_base_url"))  # stored — wins
        or os.getenv("NOUS_INFERENCE_BASE_URL")              # env — unreachable
        or DEFAULT_NOUS_INFERENCE_URL
    )

A staging OAuth login persists its inference_base_url, but the allowlist
rejects the staging host and the refresh heal rewrites the stored value to
the production default. The stored (now prod) value is then read BEFORE the
env var, so the override never takes effect — every request 401s against
prod or is pinned to prod, and setting the env var does nothing.

Fix: the user-set env override is the most-trusted source, so consult it
FIRST for the URL used to build the client / returned to callers — while
keeping the PERSISTED value the validated, network-provenance one (the
override is a runtime overlay, never written to auth.json, so unsetting it
cleanly reverts to prod). Applied at both chokepoints:

- resolve_nous_runtime_credentials (no-refresh read path AND refresh path)
- the nous_portal proxy adapter, which re-validates the resolver's returned
  base_url against the prod allowlist as defense-in-depth and would
  otherwise reject a legitimate staging override at the forward boundary.

New _nous_inference_env_override() / split of stored-vs-effective URL keep
the threat model intact: Portal-returned URLs are still allowlist-validated
at every network site, and the env path stays ungated (trusted OS user).

Also folds in the no-refresh read-path heal (supersedes the approach in
the open #50265): a poisoned stored staging host now heals to the prod
default on read even when no refresh fires.

Tests: TestEnvOverrideWins (env wins on read + refresh paths; override never
persisted; poisoned stored heals) and TestProxyAdapterEnvOverride. Verified
the 4 behavioral tests fail against pre-fix code and pass with the fix; full
inference-validation + nous-provider suites green (85 passed). E2E-validated
against a real temp HERMES_HOME exercising the real resolver + proxy adapter:
resolver→staging, persisted→prod, proxy→staging, unset→reverts to prod.
@benbarclay
benbarclay requested a review from teknium1 June 25, 2026 02:47
@github-actions

Copy link
Copy Markdown
Contributor

🔎 Lint report: fix/nous-inference-env-override vs origin/main

ruff

Total: 0 on HEAD, 0 on base (➖ 0)

🆕 New issues: none

✅ Fixed issues: none

Unchanged: 0 pre-existing issues carried over.

ty (type checker)

Total: 11258 on HEAD, 11260 on base (✅ -2)

🆕 New issues (1):

Rule Count
invalid-assignment 1
First entries
tests/run_agent/test_credits_notices_toggle.py:76: [invalid-assignment] invalid-assignment: Object of type `None` is not assignable to attribute `_credits_session_start_micros` of type `int`

✅ Fixed issues (2):

Rule Count
unresolved-attribute 2
First entries
tests/run_agent/test_credits_notices_toggle.py:76: [unresolved-attribute] unresolved-attribute: Unresolved attribute `_credits_session_start_micros` on type `AIAgent`
run_agent.py:2984: [unresolved-attribute] unresolved-attribute: Object of type `Self@get_credits_spent_micros` has no attribute `_credits_session_start_micros`

Unchanged: 5939 pre-existing issues carried over.

Diagnostics are surfaced as warnings — this check never fails the build.

@alt-glitch alt-glitch added type/security Security vulnerability or hardening comp/cli CLI entry point, hermes_cli/, setup wizard area/auth Authentication, OAuth, credential pools P2 Medium — degraded but workaround exists sweeper:risk-security-boundary Sweeper risk: may affect sandboxing, auth, credentials, or sensitive data labels Jun 25, 2026
@teknium1
teknium1 merged commit 736e981 into main Jun 25, 2026
27 checks passed
@teknium1
teknium1 deleted the fix/nous-inference-env-override branch June 25, 2026 07:11
pai-scaffolde pushed a commit to pai-scaffolde/hermes-agent that referenced this pull request Jun 28, 2026
…sessions (NousResearch#52270)

The host-allowlist hardening (NousResearch#30611) plus the refresh heal (NousResearch#49735) left
the documented NOUS_INFERENCE_BASE_URL dev/staging escape hatch unreachable
for OAuth sessions, despite three code comments asserting it still works.

Root cause — resolution precedence in resolve_nous_runtime_credentials:

    inference_base_url = (
        _optional_base_url(state.get("inference_base_url"))  # stored — wins
        or os.getenv("NOUS_INFERENCE_BASE_URL")              # env — unreachable
        or DEFAULT_NOUS_INFERENCE_URL
    )

A staging OAuth login persists its inference_base_url, but the allowlist
rejects the staging host and the refresh heal rewrites the stored value to
the production default. The stored (now prod) value is then read BEFORE the
env var, so the override never takes effect — every request 401s against
prod or is pinned to prod, and setting the env var does nothing.

Fix: the user-set env override is the most-trusted source, so consult it
FIRST for the URL used to build the client / returned to callers — while
keeping the PERSISTED value the validated, network-provenance one (the
override is a runtime overlay, never written to auth.json, so unsetting it
cleanly reverts to prod). Applied at both chokepoints:

- resolve_nous_runtime_credentials (no-refresh read path AND refresh path)
- the nous_portal proxy adapter, which re-validates the resolver's returned
  base_url against the prod allowlist as defense-in-depth and would
  otherwise reject a legitimate staging override at the forward boundary.

New _nous_inference_env_override() / split of stored-vs-effective URL keep
the threat model intact: Portal-returned URLs are still allowlist-validated
at every network site, and the env path stays ungated (trusted OS user).

Also folds in the no-refresh read-path heal (supersedes the approach in
the open NousResearch#50265): a poisoned stored staging host now heals to the prod
default on read even when no refresh fires.

Tests: TestEnvOverrideWins (env wins on read + refresh paths; override never
persisted; poisoned stored heals) and TestProxyAdapterEnvOverride. Verified
the 4 behavioral tests fail against pre-fix code and pass with the fix; full
inference-validation + nous-provider suites green (85 passed). E2E-validated
against a real temp HERMES_HOME exercising the real resolver + proxy adapter:
resolver→staging, persisted→prod, proxy→staging, unset→reverts to prod.
teknium1 pushed a commit that referenced this pull request Jul 1, 2026
…paths

Upstream #52270 added `_nous_inference_env_override()` but wired it into
only `resolve_nous_runtime_credentials`. Three sibling resolution paths
still ignored the override, so a self-hosted Nous inference endpoint set
via `NOUS_INFERENCE_BASE_URL` was silently dropped whenever credentials
arrived through any of them:

- the credential-pool path (`_resolve_runtime_from_pool_entry`)
- the explicit-provider path (`_resolve_explicit_runtime`)
- the auxiliary side-LLM client (`_pool_runtime_base_url`)

Route all three through the same auth-layer reader so every
`NOUS_INFERENCE_BASE_URL` read shares one normalization path
(trailing-slash stripping, blank -> empty) and the documented
trusted-bypass intent stays in one place. The override is live-only: it
wins for the base URL returned this run but is never persisted to
auth.json or the credential pool, so an ephemeral dev/staging value
cannot poison durable auth state.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
waefrebeorn pushed a commit to waefrebeorn/slermes that referenced this pull request Jul 2, 2026
…sessions (NousResearch#52270)

The host-allowlist hardening (NousResearch#30611) plus the refresh heal (NousResearch#49735) left
the documented NOUS_INFERENCE_BASE_URL dev/staging escape hatch unreachable
for OAuth sessions, despite three code comments asserting it still works.

Root cause — resolution precedence in resolve_nous_runtime_credentials:

    inference_base_url = (
        _optional_base_url(state.get("inference_base_url"))  # stored — wins
        or os.getenv("NOUS_INFERENCE_BASE_URL")              # env — unreachable
        or DEFAULT_NOUS_INFERENCE_URL
    )

A staging OAuth login persists its inference_base_url, but the allowlist
rejects the staging host and the refresh heal rewrites the stored value to
the production default. The stored (now prod) value is then read BEFORE the
env var, so the override never takes effect — every request 401s against
prod or is pinned to prod, and setting the env var does nothing.

Fix: the user-set env override is the most-trusted source, so consult it
FIRST for the URL used to build the client / returned to callers — while
keeping the PERSISTED value the validated, network-provenance one (the
override is a runtime overlay, never written to auth.json, so unsetting it
cleanly reverts to prod). Applied at both chokepoints:

- resolve_nous_runtime_credentials (no-refresh read path AND refresh path)
- the nous_portal proxy adapter, which re-validates the resolver's returned
  base_url against the prod allowlist as defense-in-depth and would
  otherwise reject a legitimate staging override at the forward boundary.

New _nous_inference_env_override() / split of stored-vs-effective URL keep
the threat model intact: Portal-returned URLs are still allowlist-validated
at every network site, and the env path stays ungated (trusted OS user).

Also folds in the no-refresh read-path heal (supersedes the approach in
the open NousResearch#50265): a poisoned stored staging host now heals to the prod
default on read even when no refresh fires.

Tests: TestEnvOverrideWins (env wins on read + refresh paths; override never
persisted; poisoned stored heals) and TestProxyAdapterEnvOverride. Verified
the 4 behavioral tests fail against pre-fix code and pass with the fix; full
inference-validation + nous-provider suites green (85 passed). E2E-validated
against a real temp HERMES_HOME exercising the real resolver + proxy adapter:
resolver→staging, persisted→prod, proxy→staging, unset→reverts to prod.
waefrebeorn pushed a commit to waefrebeorn/slermes that referenced this pull request Jul 2, 2026
…paths

Upstream NousResearch#52270 added `_nous_inference_env_override()` but wired it into
only `resolve_nous_runtime_credentials`. Three sibling resolution paths
still ignored the override, so a self-hosted Nous inference endpoint set
via `NOUS_INFERENCE_BASE_URL` was silently dropped whenever credentials
arrived through any of them:

- the credential-pool path (`_resolve_runtime_from_pool_entry`)
- the explicit-provider path (`_resolve_explicit_runtime`)
- the auxiliary side-LLM client (`_pool_runtime_base_url`)

Route all three through the same auth-layer reader so every
`NOUS_INFERENCE_BASE_URL` read shares one normalization path
(trailing-slash stripping, blank -> empty) and the documented
trusted-bypass intent stays in one place. The override is live-only: it
wins for the base URL returned this run but is never persisted to
auth.json or the credential pool, so an ephemeral dev/staging value
cannot poison durable auth state.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Jasper6439 pushed a commit to Jasper6439/hermes-agent that referenced this pull request Jul 5, 2026
…paths

Upstream NousResearch#52270 added `_nous_inference_env_override()` but wired it into
only `resolve_nous_runtime_credentials`. Three sibling resolution paths
still ignored the override, so a self-hosted Nous inference endpoint set
via `NOUS_INFERENCE_BASE_URL` was silently dropped whenever credentials
arrived through any of them:

- the credential-pool path (`_resolve_runtime_from_pool_entry`)
- the explicit-provider path (`_resolve_explicit_runtime`)
- the auxiliary side-LLM client (`_pool_runtime_base_url`)

Route all three through the same auth-layer reader so every
`NOUS_INFERENCE_BASE_URL` read shares one normalization path
(trailing-slash stripping, blank -> empty) and the documented
trusted-bypass intent stays in one place. The override is live-only: it
wins for the base URL returned this run but is never persisted to
auth.json or the credential pool, so an ephemeral dev/staging value
cannot poison durable auth state.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
habarmc1223-sudo pushed a commit to habarmc1223-sudo/hermes-agent-fluxmem that referenced this pull request Jul 8, 2026
…sessions (NousResearch#52270)

The host-allowlist hardening (NousResearch#30611) plus the refresh heal (NousResearch#49735) left
the documented NOUS_INFERENCE_BASE_URL dev/staging escape hatch unreachable
for OAuth sessions, despite three code comments asserting it still works.

Root cause — resolution precedence in resolve_nous_runtime_credentials:

    inference_base_url = (
        _optional_base_url(state.get("inference_base_url"))  # stored — wins
        or os.getenv("NOUS_INFERENCE_BASE_URL")              # env — unreachable
        or DEFAULT_NOUS_INFERENCE_URL
    )

A staging OAuth login persists its inference_base_url, but the allowlist
rejects the staging host and the refresh heal rewrites the stored value to
the production default. The stored (now prod) value is then read BEFORE the
env var, so the override never takes effect — every request 401s against
prod or is pinned to prod, and setting the env var does nothing.

Fix: the user-set env override is the most-trusted source, so consult it
FIRST for the URL used to build the client / returned to callers — while
keeping the PERSISTED value the validated, network-provenance one (the
override is a runtime overlay, never written to auth.json, so unsetting it
cleanly reverts to prod). Applied at both chokepoints:

- resolve_nous_runtime_credentials (no-refresh read path AND refresh path)
- the nous_portal proxy adapter, which re-validates the resolver's returned
  base_url against the prod allowlist as defense-in-depth and would
  otherwise reject a legitimate staging override at the forward boundary.

New _nous_inference_env_override() / split of stored-vs-effective URL keep
the threat model intact: Portal-returned URLs are still allowlist-validated
at every network site, and the env path stays ungated (trusted OS user).

Also folds in the no-refresh read-path heal (supersedes the approach in
the open NousResearch#50265): a poisoned stored staging host now heals to the prod
default on read even when no refresh fires.

Tests: TestEnvOverrideWins (env wins on read + refresh paths; override never
persisted; poisoned stored heals) and TestProxyAdapterEnvOverride. Verified
the 4 behavioral tests fail against pre-fix code and pass with the fix; full
inference-validation + nous-provider suites green (85 passed). E2E-validated
against a real temp HERMES_HOME exercising the real resolver + proxy adapter:
resolver→staging, persisted→prod, proxy→staging, unset→reverts to prod.
habarmc1223-sudo pushed a commit to habarmc1223-sudo/hermes-agent-fluxmem that referenced this pull request Jul 8, 2026
…paths

Upstream NousResearch#52270 added `_nous_inference_env_override()` but wired it into
only `resolve_nous_runtime_credentials`. Three sibling resolution paths
still ignored the override, so a self-hosted Nous inference endpoint set
via `NOUS_INFERENCE_BASE_URL` was silently dropped whenever credentials
arrived through any of them:

- the credential-pool path (`_resolve_runtime_from_pool_entry`)
- the explicit-provider path (`_resolve_explicit_runtime`)
- the auxiliary side-LLM client (`_pool_runtime_base_url`)

Route all three through the same auth-layer reader so every
`NOUS_INFERENCE_BASE_URL` read shares one normalization path
(trailing-slash stripping, blank -> empty) and the documented
trusted-bypass intent stays in one place. The override is live-only: it
wins for the base URL returned this run but is never persisted to
auth.json or the credential pool, so an ephemeral dev/staging value
cannot poison durable auth state.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
santhreal pushed a commit to santhreal/hermes-agent that referenced this pull request Jul 13, 2026
…sessions (NousResearch#52270)

The host-allowlist hardening (NousResearch#30611) plus the refresh heal (NousResearch#49735) left
the documented NOUS_INFERENCE_BASE_URL dev/staging escape hatch unreachable
for OAuth sessions, despite three code comments asserting it still works.

Root cause — resolution precedence in resolve_nous_runtime_credentials:

    inference_base_url = (
        _optional_base_url(state.get("inference_base_url"))  # stored — wins
        or os.getenv("NOUS_INFERENCE_BASE_URL")              # env — unreachable
        or DEFAULT_NOUS_INFERENCE_URL
    )

A staging OAuth login persists its inference_base_url, but the allowlist
rejects the staging host and the refresh heal rewrites the stored value to
the production default. The stored (now prod) value is then read BEFORE the
env var, so the override never takes effect — every request 401s against
prod or is pinned to prod, and setting the env var does nothing.

Fix: the user-set env override is the most-trusted source, so consult it
FIRST for the URL used to build the client / returned to callers — while
keeping the PERSISTED value the validated, network-provenance one (the
override is a runtime overlay, never written to auth.json, so unsetting it
cleanly reverts to prod). Applied at both chokepoints:

- resolve_nous_runtime_credentials (no-refresh read path AND refresh path)
- the nous_portal proxy adapter, which re-validates the resolver's returned
  base_url against the prod allowlist as defense-in-depth and would
  otherwise reject a legitimate staging override at the forward boundary.

New _nous_inference_env_override() / split of stored-vs-effective URL keep
the threat model intact: Portal-returned URLs are still allowlist-validated
at every network site, and the env path stays ungated (trusted OS user).

Also folds in the no-refresh read-path heal (supersedes the approach in
the open NousResearch#50265): a poisoned stored staging host now heals to the prod
default on read even when no refresh fires.

Tests: TestEnvOverrideWins (env wins on read + refresh paths; override never
persisted; poisoned stored heals) and TestProxyAdapterEnvOverride. Verified
the 4 behavioral tests fail against pre-fix code and pass with the fix; full
inference-validation + nous-provider suites green (85 passed). E2E-validated
against a real temp HERMES_HOME exercising the real resolver + proxy adapter:
resolver→staging, persisted→prod, proxy→staging, unset→reverts to prod.
santhreal pushed a commit to santhreal/hermes-agent that referenced this pull request Jul 13, 2026
…paths

Upstream NousResearch#52270 added `_nous_inference_env_override()` but wired it into
only `resolve_nous_runtime_credentials`. Three sibling resolution paths
still ignored the override, so a self-hosted Nous inference endpoint set
via `NOUS_INFERENCE_BASE_URL` was silently dropped whenever credentials
arrived through any of them:

- the credential-pool path (`_resolve_runtime_from_pool_entry`)
- the explicit-provider path (`_resolve_explicit_runtime`)
- the auxiliary side-LLM client (`_pool_runtime_base_url`)

Route all three through the same auth-layer reader so every
`NOUS_INFERENCE_BASE_URL` read shares one normalization path
(trailing-slash stripping, blank -> empty) and the documented
trusted-bypass intent stays in one place. The override is live-only: it
wins for the base URL returned this run but is never persisted to
auth.json or the credential pool, so an ephemeral dev/staging value
cannot poison durable auth state.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Gravezzz pushed a commit to Gravezzz/hermes-agent that referenced this pull request Jul 21, 2026
…sessions (NousResearch#52270)

The host-allowlist hardening (NousResearch#30611) plus the refresh heal (NousResearch#49735) left
the documented NOUS_INFERENCE_BASE_URL dev/staging escape hatch unreachable
for OAuth sessions, despite three code comments asserting it still works.

Root cause — resolution precedence in resolve_nous_runtime_credentials:

    inference_base_url = (
        _optional_base_url(state.get("inference_base_url"))  # stored — wins
        or os.getenv("NOUS_INFERENCE_BASE_URL")              # env — unreachable
        or DEFAULT_NOUS_INFERENCE_URL
    )

A staging OAuth login persists its inference_base_url, but the allowlist
rejects the staging host and the refresh heal rewrites the stored value to
the production default. The stored (now prod) value is then read BEFORE the
env var, so the override never takes effect — every request 401s against
prod or is pinned to prod, and setting the env var does nothing.

Fix: the user-set env override is the most-trusted source, so consult it
FIRST for the URL used to build the client / returned to callers — while
keeping the PERSISTED value the validated, network-provenance one (the
override is a runtime overlay, never written to auth.json, so unsetting it
cleanly reverts to prod). Applied at both chokepoints:

- resolve_nous_runtime_credentials (no-refresh read path AND refresh path)
- the nous_portal proxy adapter, which re-validates the resolver's returned
  base_url against the prod allowlist as defense-in-depth and would
  otherwise reject a legitimate staging override at the forward boundary.

New _nous_inference_env_override() / split of stored-vs-effective URL keep
the threat model intact: Portal-returned URLs are still allowlist-validated
at every network site, and the env path stays ungated (trusted OS user).

Also folds in the no-refresh read-path heal (supersedes the approach in
the open NousResearch#50265): a poisoned stored staging host now heals to the prod
default on read even when no refresh fires.

Tests: TestEnvOverrideWins (env wins on read + refresh paths; override never
persisted; poisoned stored heals) and TestProxyAdapterEnvOverride. Verified
the 4 behavioral tests fail against pre-fix code and pass with the fix; full
inference-validation + nous-provider suites green (85 passed). E2E-validated
against a real temp HERMES_HOME exercising the real resolver + proxy adapter:
resolver→staging, persisted→prod, proxy→staging, unset→reverts to prod.
Gravezzz pushed a commit to Gravezzz/hermes-agent that referenced this pull request Jul 21, 2026
…paths

Upstream NousResearch#52270 added `_nous_inference_env_override()` but wired it into
only `resolve_nous_runtime_credentials`. Three sibling resolution paths
still ignored the override, so a self-hosted Nous inference endpoint set
via `NOUS_INFERENCE_BASE_URL` was silently dropped whenever credentials
arrived through any of them:

- the credential-pool path (`_resolve_runtime_from_pool_entry`)
- the explicit-provider path (`_resolve_explicit_runtime`)
- the auxiliary side-LLM client (`_pool_runtime_base_url`)

Route all three through the same auth-layer reader so every
`NOUS_INFERENCE_BASE_URL` read shares one normalization path
(trailing-slash stripping, blank -> empty) and the documented
trusted-bypass intent stays in one place. The override is live-only: it
wins for the base URL returned this run but is never persisted to
auth.json or the credential pool, so an ephemeral dev/staging value
cannot poison durable auth state.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
leewenjie pushed a commit to leewenjie/hermes-agent that referenced this pull request Aug 7, 2026
…sessions (NousResearch#52270)

The host-allowlist hardening (NousResearch#30611) plus the refresh heal (NousResearch#49735) left
the documented NOUS_INFERENCE_BASE_URL dev/staging escape hatch unreachable
for OAuth sessions, despite three code comments asserting it still works.

Root cause — resolution precedence in resolve_nous_runtime_credentials:

    inference_base_url = (
        _optional_base_url(state.get("inference_base_url"))  # stored — wins
        or os.getenv("NOUS_INFERENCE_BASE_URL")              # env — unreachable
        or DEFAULT_NOUS_INFERENCE_URL
    )

A staging OAuth login persists its inference_base_url, but the allowlist
rejects the staging host and the refresh heal rewrites the stored value to
the production default. The stored (now prod) value is then read BEFORE the
env var, so the override never takes effect — every request 401s against
prod or is pinned to prod, and setting the env var does nothing.

Fix: the user-set env override is the most-trusted source, so consult it
FIRST for the URL used to build the client / returned to callers — while
keeping the PERSISTED value the validated, network-provenance one (the
override is a runtime overlay, never written to auth.json, so unsetting it
cleanly reverts to prod). Applied at both chokepoints:

- resolve_nous_runtime_credentials (no-refresh read path AND refresh path)
- the nous_portal proxy adapter, which re-validates the resolver's returned
  base_url against the prod allowlist as defense-in-depth and would
  otherwise reject a legitimate staging override at the forward boundary.

New _nous_inference_env_override() / split of stored-vs-effective URL keep
the threat model intact: Portal-returned URLs are still allowlist-validated
at every network site, and the env path stays ungated (trusted OS user).

Also folds in the no-refresh read-path heal (supersedes the approach in
the open NousResearch#50265): a poisoned stored staging host now heals to the prod
default on read even when no refresh fires.

Tests: TestEnvOverrideWins (env wins on read + refresh paths; override never
persisted; poisoned stored heals) and TestProxyAdapterEnvOverride. Verified
the 4 behavioral tests fail against pre-fix code and pass with the fix; full
inference-validation + nous-provider suites green (85 passed). E2E-validated
against a real temp HERMES_HOME exercising the real resolver + proxy adapter:
resolver→staging, persisted→prod, proxy→staging, unset→reverts to prod.
leewenjie pushed a commit to leewenjie/hermes-agent that referenced this pull request Aug 7, 2026
…paths

Upstream NousResearch#52270 added `_nous_inference_env_override()` but wired it into
only `resolve_nous_runtime_credentials`. Three sibling resolution paths
still ignored the override, so a self-hosted Nous inference endpoint set
via `NOUS_INFERENCE_BASE_URL` was silently dropped whenever credentials
arrived through any of them:

- the credential-pool path (`_resolve_runtime_from_pool_entry`)
- the explicit-provider path (`_resolve_explicit_runtime`)
- the auxiliary side-LLM client (`_pool_runtime_base_url`)

Route all three through the same auth-layer reader so every
`NOUS_INFERENCE_BASE_URL` read shares one normalization path
(trailing-slash stripping, blank -> empty) and the documented
trusted-bypass intent stays in one place. The override is live-only: it
wins for the base URL returned this run but is never persisted to
auth.json or the credential pool, so an ephemeral dev/staging value
cannot poison durable auth state.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

area/auth Authentication, OAuth, credential pools comp/cli CLI entry point, hermes_cli/, setup wizard P2 Medium — degraded but workaround exists sweeper:risk-security-boundary Sweeper risk: may affect sandboxing, auth, credentials, or sensitive data type/security Security vulnerability or hardening

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants