* fix(agents): collapse repeated waits queued inside one model turn
* fix(agents): state that one wait is enough in every prompt variant
---------
Co-authored-by: Ahmed Allam <ahmed39652003@gmail.com>
* fix(llm): cap the tool calls one assistant response may queue
* fix(llm): cap the subscription backend's responses too
---------
Co-authored-by: Ahmed Allam <ahmed39652003@gmail.com>
An exception escaping a non-interactive cycle re-raised before the status
handling, so a dying child stayed 'running' and its parent waited out the
timeout on a completion report the child could no longer send. Set the
terminal status and wake the parent on the way out too.
A tool call for a name Strix does not register raised ModelBehaviorError
from the SDK turn resolver, which nothing retries: the root agent's raise
tore down the whole scan and a sub-agent died before its status was set.
Opt into the SDK's tool_not_found_behavior="return_error_to_model" so the
unknown call comes back as a tool result and the agent self-corrects.
The setting landed in openai-agents 0.19.0, which requires openai>=2.45,
so both pins move.
One tool was doing three jobs (wait on the user, wait on other agents, and
- wrongly - wait for a long-running command), so the driver had to guess which
one an agent meant and used parent_id as the proxy: the root waits for a human,
everyone else waits for agents. That proxy is wrong, since the user can message
any agent from the TUI's agent tree.
Tool identity now carries the intent, and the coordinator records it as a
wait_kind that survives snapshot/restore:
respond_to_user -> wait_kind="user", never auto-resumed (root or not)
wait_for_agents -> wait_kind="agents", auto-resumed on a 300s timer
recovery exhaust -> wait_kind="stalled"
respond_to_user fuses the message and the yield into one call, so there is no
way to answer and then forget to stop - the two-step that gpt-4o-mini skipped
2/2 in live testing. Plain text still renders as before.
Auto-resume is also bounded now: an agent that re-parks after every timeout
burned a model turn every 300s for the rest of the scan (and, since parked
children notify their parent, spammed the parent's inbox on the same cycle).
After _MAX_IDLE_AUTO_RESUMES it stays parked until a real message arrives.
The user can message any agent from the TUI, not only the root, so the
justification is that the parent is an agent with no other way to learn
the child parked - not that the child has no human resumer.
Parking is self-service only for the root, which the user is watching.
A parked child owes its parent a report it can no longer send, so the
parent would wait out its full timeout for nothing.
An exhausted agent parked in 'waiting' got a fresh nudge budget on every
600s auto-resume, so a wedged agent could nudge-park-nudge indefinitely.
Track the count on the coordinator, snapshot it, and reset it only on
real input or an explicit lifecycle tool.
Interactive turns ended by plain text left the agent parked in 'waiting'
forever. Require an explicit lifecycle tool in both modes and nudge a
text-only turn back into a tool call, bounded by a recovery limit.
LiteLLM treats provider-qualified metadata lookups as an auth path.
Use the underlying model slug so context sizing cannot block the scan
loop in a device-code poll.
Adds an integration test that drives Runner.run_streamed against a
non-streaming gateway through _NonStreamingModel: the synthetic terminal
event feeds the runner, which executes the tool call and continues to a
final answer over two non-streaming turns. Removes the README env-var note.
Some OpenAI-compatible gateways don't support Server-Sent Events (or
deliver them unreliably), but the SDK run loop Strix uses only issues
streamed requests, so such a gateway fails every turn. Add an opt-in
LLM_DISABLE_STREAMING setting that wraps the resolved model in
_NonStreamingModel: each turn makes one non-streaming get_response and
replays the completed result as a single terminal stream event, so tool
calls, usage, and the rest of the agent loop are unchanged. Subscription
(ChatGPT) models are always streamed and are not wrapped.
The predictable lock path was opened with Path.open("w"), following (and
truncating through) a pre-positioned symlink. Open it via os.open with
O_NOFOLLOW and no O_TRUNC, raising StoreLockError on a symlinked lock path.
Third review pass (security): the store temp file used a predictable
subscription-auth.json.tmp name, so a local attacker could pre-plant a symlink
there and divert the OAuth token write. Create it with tempfile.mkstemp
(random name, mode 0600, no symlink following) in the same directory, then
atomically rename over the target.
Third review pass: the shared credential store no longer proceeds with an
unlocked read-modify-write when fcntl is missing or flock fails. It now retries
on EINTR and otherwise raises StoreLockError, so concurrent provider
login/refresh/logout can never race by silently skipping the cross-process lock.
Second review pass:
- strix auth logout (all providers) now holds the shared store lock across every provider removal, so a concurrent save/refresh cannot leave one credential behind while reporting all removed.
- _persist_run_record writes subscription_provider alongside auth_mode, matching ReportState, so resumed runs keep their original provider label.
Addresses PR review:
- Extract the shared ~/.strix/subscription-auth.json handling into subscription_store, writing tokens owner-only (0600) from creation via os.open instead of chmod-after-write, closing the window where credentials were briefly world/group-readable.
- Serialize read-modify-write across providers and processes with a reentrant lock, so overlapping ChatGPT/Grok save/logout/refresh operations no longer clobber each other.
- Live/TUI stats label the subscription from the persisted run record (falling back to provider-aware settings), so resumed runs no longer mislabel the provider when STRIX_LLM changes.
read_run_summary backfills subscription_provider from the recorded provider/model slug (reusing subscription.provider_label) when the field is absent, so runs recorded before it existed still label correctly without a rescan. The viewer no longer defaults to "ChatGPT" when the provider is unknown. Rebuilds the committed viewer bundle.
The Run details panel hardcoded "ChatGPT subscription" for any subscription run. Emit subscription_provider (ChatGPT/Grok) in the run record and render it in the viewer, so Grok runs read "Grok subscription". Rebuilds the committed viewer bundle.
Add Sign in with Grok mirroring the merged ChatGPT/Codex integration: a strix/config/grok.py OAuth module (PKCE loopback flow against auth.x.ai, refresh with cross-process locking, secure ~/.strix/subscription-auth.json store) plus grok/<model> routing through the OpenAI-compatible api.x.ai/v1 endpoint via a bearer-stamping chat-completions client.
strix auth login grok / status / logout become provider-aware; subscription.py shares provider-agnostic auth-mode detection; Grok runs are marked zero-cost like other subscriptions.
Uses a consumer subscription outside xAI own products, which xAI does not officially support; opt-in and experimental.
yoni