The gateway half of relay Phase 3. On a MANAGED boot with relay configured and
no secret pinned, the runtime self-provisions its relay credentials IN-PROCESS:
resolve the agent's own Nous access token (resolve_nous_access_token) -> POST
the connector's /relay/provision asserting its own endpoint + route keys ->
set GATEWAY_RELAY_ID/SECRET/DELIVERY_KEY into os.environ so the immediately-
following register_relay_adapter() reads them and dials out authenticated.
No human, no enrollment token, no disk write — the creds live only in process
memory (save_env_value refuses under managed anyway, and keeping the secret off
any volume is the stronger posture). Stateless: process-env creds don't survive
a restart, so a managed container re-provisions every boot; the connector's
rotation window covers a still-connected prior instance. An explicitly-pinned
GATEWAY_RELAY_SECRET is respected (skip). Self-hosted is unchanged: humans keep
using `hermes gateway enroll`.
Endpoint provenance is gateway-asserted (GATEWAY_RELAY_ENDPOINT +
GATEWAY_RELAY_ROUTE_KEYS, env or gateway.relay_* config) — uniform code path
whether the operator sets it (self-hosted) or NAS stamps it (hosted, the only
case NAS knows the public URL). Both absent -> outbound-only provisioning
(credentials, no inbound routes). The connector scopes the asserted endpoint to
the verified tenant, so it stays within the security model.
- gateway/relay/__init__.py: relay_endpoint(), relay_route_keys(),
_provision_url(), _post_provision(), self_provision_if_managed() (never
raises — a provision failure logs and boots without relay auth).
- gateway/run.py: call self_provision_if_managed() immediately before
register_relay_adapter() in the startup path.
Tests: 12 unit (trigger logic, respect-pinned-secret, in-process env wiring,
endpoint+routes vs outbound-only, fail-soft on token/connector failure);
mutation-checked (drop is_managed guard / pinned-secret guard -> tests fail).
Cross-repo live E2E driver lands on the connector side (depends on this).
EXPERIMENTAL: relay auth scheme may change until >=2 Class-1 platforms validate.
* feat(relay): authenticate the connector⇄gateway WS channel
The relay gateway may be customer-managed and internet-exposed, so the
connector⇄gateway channel is itself authenticated (distinct from the
platform crypto the relay path sheds). Add gateway/relay/auth.py — a
Python port of the connector's HMAC token + delivery-signature schemes
(relayAuthToken.ts / deliverySigning.ts), verified byte-for-byte against
the connector's compiled TypeScript via cross-language test vectors.
Present an Authorization bearer on the /relay WS upgrade keyed by the
per-gateway secret (resolved from GATEWAY_RELAY_ID / GATEWAY_RELAY_SECRET
in env or config). The connector rejects an unauthenticated/invalid/
revoked upgrade with close 4401.
* feat(relay): signed-HTTP inbound delivery receiver
The connector delivers normalized inbound events to a tenant's gateway
over a signed HTTP POST, not the outbound /relay WS: the connector
instance owning a platform socket is generally not the instance a given
gateway dialed out to, so inbound targets a tenant endpoint that may
load-balance across gateway instances.
Add gateway/relay/inbound_receiver.py — verifies x-relay-signature /
x-relay-timestamp over the EXACT raw request bytes (re-serializing would
break the HMAC: JS JSON.stringify is compact, Python json.dumps spaces)
against the per-tenant delivery key verify list within a 300s replay
window, then dispatches messages to handle_message and interrupts to the
interrupt handler. Wire it into the adapter lifecycle (start in connect()
when a delivery key + bind port are configured, tear down in disconnect();
a purely-outbound dev gateway runs without it).
Refine test_relay_sheds_crypto to distinguish PLATFORM crypto (Discord
ed25519, Twilio/WeCom HMAC — still shed) from the connector⇄gateway
CHANNEL auth (intended): auth.py / inbound_receiver.py are exempt from
the platform-symbol scan but still banned from importing platform-crypto
modules, plus a positive guard that auth.py uses only stdlib hmac/hashlib.
* feat(relay): hermes gateway enroll CLI
Add the gateway half of zero-touch enrollment. `hermes gateway enroll`
resolves a fresh Nous Portal access token (the tenant-proving identity),
POSTs {enrollmentToken, gatewayId} to the connector's /relay/enroll, and
persists GATEWAY_RELAY_ID / GATEWAY_RELAY_SECRET / GATEWAY_RELAY_DELIVERY_KEY
to ~/.hermes/.env. The per-gateway secret authenticates the WS upgrade;
the per-tenant delivery key verifies signed inbound deliveries.
Refuses under is_managed() (hosted installs get the secret stamped in by
the orchestrator). Added as an 'enroll' subcommand on the existing
gateway subparser — not a new top-level command.
* docs(relay): inbound is signed HTTP, not WS; document channel auth
Fix the stale contract: §3/§5 said inbound rode the WS socket (single-
instance only, predates the multi-instance socket-ownership + channel-auth
model). Inbound + connector→gateway interrupt are signed HTTP POSTs to the
tenant endpoint. Add §6.1 documenting the two channel-auth schemes (per-
gateway WS-upgrade secret, per-tenant inbound delivery key) and how they
differ from the platform crypto the relay path sheds.
* test(relay): update build_gateway_parser callers for cmd_gateway_enroll
The enroll subcommand added cmd_gateway_enroll as a required keyword-only
arg to build_gateway_parser, but two existing parser-extraction tests still
called it with only cmd_gateway/cmd_proxy — failing CI with TypeError.
Thread the new handler through both call sites and add a test asserting
`gateway enroll` dispatches to cmd_gateway_enroll with its flags parsed.
- test_ws_transport.py: drives WebSocketRelayTransport against a REAL in-process
websockets server (not a mock socket): handshake (hello->descriptor), inbound
frame -> handler, outbound request/response correlation, follow_up routing,
and clean disconnect failing pending waiters. Skips if websockets is absent.
- test_relay_registration.py: rewritten for the config-driven gate — registers
when GATEWAY_RELAY_URL is set / an explicit url is passed / force=True; no-op
without a URL; trailing slash stripped; adapter constructs through the registry.
Full relay suite: 57 passed.
The relay outbound surface had send/edit/typing but no way to act on a
SHARED-identity capability (e.g. a Discord interaction follow-up token,
~15min) that the connector captured + stripped at the edge. Under A2 that
credential never reaches the gateway, so the gateway can't just 'send with
the token' — it needs a semantic op naming the session it's already in.
Adds the follow_up op end to end on the gateway side:
- RelayTransport.send_follow_up(action): protocol method. Action carries
op='follow_up' + session_key + kind + content (+ metadata) and NO token.
- RelayAdapter.send_follow_up(session_key, kind, content, metadata): builds
that action and returns a SendResult. The connector resolves the real
capability (its resolveOutboundCapability), enforces the tenant match so
tenant B can't wield tenant A's capability, and egresses; success=False
when the capability is absent/expired/mismatched (nothing to retry — a
leaked gateway holds zero capability material).
- StubConnector records follow_ups + a canned next_follow_up_result.
Tests: round-trips without a token; the wire action carries only session
refs (no credential value field — the 'kind' string is a type ref, not the
secret); failure surfaces when the connector can't resolve; no-transport
fails cleanly. 55 passed. §4 doc entry follows in the contract-rewrite commit.
Under the A2 trust model the connector is the SOLE crypto/identity
boundary: it verifies/decrypts every inbound platform payload at the edge
(it holds the tenant secrets), normalizes to a tenant-scoped MessageEvent,
and forwards only the sanitized event. The gateway re-validates nothing —
it cannot without being handed the shared signing secret, which on a
shared bot is itself the cross-tenant leak.
The relay path already imports no platform-crypto today; this locks that
in as an enforced invariant so nobody bolts re-validation (Discord
ed25519, Twilio HMAC, WeCom BizMsgCrypt, generic webhook signature checks)
onto the relay later and silently re-couples the gateway to platform
secrets it must never hold. Verification stays in the direct platform
adapters (gateway/platforms/*) which serve non-relay deployments.
- test_relay_package_imports_no_platform_crypto: AST-walks gateway/relay/*
and fails on any import of a platform-crypto/verification module.
- test_relay_package_calls_no_signature_verification: fails on any
verification-symbol reference (ed25519/hmac/bizmsg/verify_*).
Invariants (assert the relation 'relay re-validates nothing'), not frozen
snapshots. Verified the guard bites: injecting a wecom_crypto import makes
it fail, removing it goes green. docs §6 rewrite follows in a later commit.
The Phase 1 exit gate requires BOTH Discord and Telegram to round-trip
through the relay stub, but test_relay_roundtrip.py only covered Discord.
Add the Telegram companion exercising its distinct discriminator profile:
- no guild_id — two chats isolate on chat_id alone
- forum topics share one chat_id and isolate by thread_id (the Telegram
analog of Discord per-guild isolation), shared across participants by
default (thread_sessions_per_user=False)
- DM isolation by chat_id
- utf16 len_unit + markdown_v2 dialect round-trip and configure the adapter
- outbound send round-trips through the stub
Proves the CapabilityDescriptor + build_session_key generalize beyond
Discord, not just the struct (which the descriptor unit tests already
covered).
Add an invariant test pinning docs/relay-connector-contract.md to the
Python source of truth so the doc (which the connector repo mirrors by
hand) cannot silently drift:
- CapabilityDescriptor §2 table ⟷ dataclass fields + required/optional
- SessionSource wire keys (to_dict output) ⟷ §3 documented fields
- per-platform discriminator columns exist as real SessionSource fields
- guard that is_bot stays off the wire until deliberately promoted
Writing the test surfaced a real gap: §3 only enumerated 5 discriminators
in its per-platform table while to_dict() emits 12 keys. Seven wire keys
the connector must populate (chat_name, chat_topic, user_id_alt,
chat_id_alt, parent_chat_id, message_id, user_name) were undocumented —
a connector author reading the doc would never know to set them. Added a
complete SessionSource wire-field table to §3. The connector's existing
contract.ts already carries all 12, so no connector change is needed; the
doc was the lagging artifact.
The platform-connected-checker invariant test requires every built-in
Platform enum member to have either a generic token path or a bespoke
entry in _PLATFORM_CONNECTED_CHECKERS. Platform.RELAY was added without
one, so test_all_builtins_have_checker_or_generic_token_path failed.
Relay dials OUT to a connector and is 'connected' once an endpoint URL
is configured (extra['relay_url'] or extra['url']); the capability
descriptor is negotiated at handshake time, so the URL is the only
config-level signal in the experimental phase. Add the checker plus a
synthetic-config case exercising its True path.
CI guard: fails if gateway/ or plugins/ ever imports the test-only stub
connector or defines StubConnector. Matches code leaks (imports / class defs),
not prose mentions, so the transport.py docstring reference to the stub's path
is allowed.
Phase 1 complete. Task 1.6 of the gateway-relay plan.
RelayAdapter.on_interrupt(session_key, chat_id) bridges a connector-delivered
mid-turn /stop into the existing interrupt_session_activity path, setting the
per-session _active_sessions Event and clearing typing — cancelling exactly the
targeted session's turn without touching siblings (mirrors test_stop_thread_
sibling isolation). Transport.send_interrupt carries the gateway-side egress to
the connector for socket-owner routing.
Phase 1, Task 1.4 of the gateway-relay plan.
register_relay_adapter() registers the generic 'relay' platform via the same
PlatformRegistry path as plugin adapters — no core dispatch changes. OFF by
default (dark-launch): only registers when HERMES_GATEWAY_RELAY is truthy (or
force=True for tests), so existing single-tenant/direct deployments are
unaffected. Factory builds a transport-less RelayAdapter with a placeholder
descriptor; the real descriptor is negotiated at handshake.
Phase 1, Task 1.3 of the gateway-relay plan.
Defines RelayTransport (lifecycle/handshake/inbound/outbound/interrupt) as the
gateway<->connector wire contract; RelayAdapter.connect now registers an inbound
handler that bridges connector-delivered MessageEvents into handle_message.
Adds an in-memory StubConnector under tests/ and an E2E round-trip proving:
connect registers the handler, inbound events reach the adapter, guild_id drives
build_session_key isolation (two guilds -> two keys; same guild/channel/user ->
one), outbound send round-trips, get_chat_info is proxied.
Phase 1, Task 1.2 of the gateway-relay plan.
One BasePlatformAdapter subclass that reads its capability profile from a
CapabilityDescriptor: MAX_MESSAGE_LENGTH attribute, message_len_fn (table-driven
by len_unit: chars=len, utf16=Telegram-style code units), supports_draft_streaming.
Implements the four abstract methods (connect/disconnect/send/get_chat_info) by
delegating to an injected RelayTransport (full protocol lands in Task 1.2). Adds
Platform.RELAY enum member. No per-platform gateway code.
Phase 1, Task 1.1 of the gateway-relay plan.
CapabilityDescriptor.from_platform_entry() projects an existing PlatformEntry
(label, max_message_length, emoji, platform_hint, pii_safe, name) into a
descriptor, proving the descriptor is a projection of existing config rather
than a parallel concept. Runtime-only capabilities (len_unit, draft/edit/
thread/markdown) are caller-supplied. max_message_length==0 ('no limit') maps
to the stream_consumer 4096 default.
Phase 0 complete. Task 0.3 of the gateway-relay plan.
Behavioral regression harness locking the capability surface that the future
RelayAdapter must reproduce: the abstract-method set (connect/disconnect/send/
get_chat_info), message_len_fn default, supports_draft_streaming default, and
the stream_consumer MAX_MESSAGE_LENGTH attribute read. Passes on main before
any RelayAdapter exists.
Phase 0, Task 0.1 of the gateway-relay plan.
Follow-up to salvaged PR #41633: the timestamp prefix injection was
unconditional. Gate the in-context render behind
gateway.message_timestamps.enabled (default false) at both the live-message
and history-replay sites; timestamp metadata is still captured + persisted
regardless so the toggle can be flipped on later. Add DEFAULT_CONFIG entry,
docs, and gate tests.
Consolidates these related Amy fork patches:
- 429830f39 feat(gateway): inject message timestamps into user messages for LLM context
- 3c3d6fac0 fix: handle both ISO string and epoch float timestamps in history replay
- 2874f7725 feat: human-friendly timestamp format with weekday and timezone name
- 3735f4c8b fix: render gateway message timestamps once
Cover #47375 fix: record-on-rich-send + lookup-on-reply round trip,
lookup miss leaving reply_to_text None, and precedence (native quote
and echoed caption both win over the index fallback).
PROBLEM: Mattermost threads can become invalid or enormous, exposing two failure modes: internal scratch/reasoning/commentary displays could leak into persistent Mattermost threads via global display toggles, while rejected threaded user-visible replies could disappear unless every failed send fell back flat. A broad flat fallback would pollute channels with tool/status/progress noise.
SOLUTION: Require explicit Mattermost platform opt-in for scratch displays, keep using the existing notify=True metadata marker for user-visible final text/media/file replies, and allow the Mattermost plugin adapter to flat-fallback only notify-worthy sends whose threaded POST failure looks like a broken root/thread. Keep tool/status/progress and other non-notify sends thread-strict. Add regression tests for display opt-in, notify-only broken-thread fallback, generic API failure suppression, and stream notify metadata.
Verification: tests/gateway/test_mattermost.py tests/gateway/test_stream_consumer.py tests/gateway/test_stream_consumer_thread_routing.py tests/gateway/test_stream_consumer_fresh_final.py tests/gateway/test_stream_consumer_draft.py; tests/gateway/test_session_api.py tests/gateway/test_status_command.py tests/gateway/test_resume_command.py tests/hermes_cli/test_commands.py; py_compile touched gateway files; git diff --check.
Session: Mattermost thread 6qg8e9dd1pd9pkhi74xyaa1mry, 2026-06-01.
`BasePlatformAdapter.send_multiple_images` passes `metadata=metadata` to
`send_image` / `send_image_file` / `send_animation` on every send. The
WhatsApp and email `send_image` overrides stopped their signature at
`reply_to`, so any image delivered as a URL (the common case — image-gen
backends return URLs) raised:
TypeError: send_image() got an unexpected keyword argument "metadata"
and the image silently failed to send. Their sibling overrides
(`send_image_file` / `send_video` / `send_voice` / `send_document`)
already absorb it via **kwargs, which is why only plain image-URL sends
broke.
- whatsapp/email `send_image`: accept `metadata` (matches the base
signature); WhatsApp forwards it to the super() text fallback.
- Add `tests/gateway/test_media_metadata_contract.py`: asserts WhatsApp +
email accept it, plus a best-effort sweep over every adapter so the next
slip fails at test time instead of in production.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Streamed Telegram replies that finalize through editMessageText were
converted to MarkdownV2, which has no table syntax and rewrites pipe
tables into bullet lists — users saw a table while streaming that
collapsed to a list at the last moment.
Finalize now edits the existing preview IN PLACE via Bot API 10.1's
editMessageText rich_message parameter when the content has constructs
the legacy path degrades (tables, task lists, <details>, block math).
No fresh send + delete, so no duplicate-preview flicker — the reason
#46206 reverted the fresh-final re-send path. prefers_fresh_final_streaming
stays False; the in-place edit replaces it.
- _needs_rich_rendering(): rich reserved for table/task-list/details/math
(adapted from #45995, @YonganZhang); plain replies stay on MarkdownV2.
- _try_edit_rich(): editMessageText + rich_message via do_api_request,
mirroring _try_send_rich's fallback/latch/transient contract.
- edit_message finalize tries rich in place before the 4,096 overflow
pre-flight (rich cap is 32,768), falling back to legacy on rejection.
- rich_messages default flipped back to True (DEFAULT_CONFIG + adapter).
- docs (en + zh-Hans) + cli-config example updated to default-on.
Closes the root cause behind #45911 / #46009.
* fix(teams): package Microsoft Teams SDK as an installable extra
The Teams adapter imports the microsoft-teams-apps SDK, but it was never
declared as a dependency, so source/local installs hit ImportError and the
adapter silently reported the SDK as unavailable. Add a 'teams' extra
(microsoft-teams-apps==2.0.13.4 + aiohttp) and document 'uv sync --extra teams'.
Per the 2026-05-12 [all] policy, opt-in messaging-platform SDKs are NOT added
to [all] (they would break every fresh install on a quarantined release); the
teams extra is installed on demand like the other platform backends.
Co-authored-by: rio-jeong <rio.jeong@thebytesize.ai>
* chore: map rio-jeong contributor email for attribution (#43945)
* feat(teams): lazy-install the Teams SDK on demand (parity with other channels)
The teams extra alone left Teams as the only messaging platform that wouldn't
auto-install its SDK — every other channel (telegram, discord, slack, matrix,
dingtalk, feishu) lazy-installs via tools.lazy_deps on first connect. Bring
Teams to parity:
- Add 'platform.teams' to LAZY_DEPS (microsoft-teams-apps + aiohttp).
- Replace the passive 'check_teams_requirements = check_requirements' alias with
a real lazy-installer that calls ensure_and_bind('platform.teams', ...),
rebinding all Teams SDK globals on success (mirrors check_slack_requirements).
- Call check_teams_requirements() at the top of TeamsAdapter.connect() so
enabling Teams installs the SDK on demand.
- Keep the passive check_requirements() as the registry check_fn so 'gateway
status' probes never trigger a pip install.
The 'teams' extra remains for packagers / explicit 'uv sync --extra teams'.
Tests: rework the alias test into shortcircuit + lazy-install assertions, and
update test_connect_fails_without_sdk to simulate an uninstallable SDK.
---------
Co-authored-by: rio-jeong <rio.jeong@thebytesize.ai>
Co-authored-by: Teknium <127238744+teknium1@users.noreply.github.com>
PROBLEM: The old public /status PR drifted out of the current Amy patch stack, leaving /status without the model/provider, context window, or explicit cumulative token label that Wolfram uses to monitor context pressure from chat.
SOLUTION: Re-port the feature onto the current gateway status handler. Prefer live/cached agent runtime metadata, fall back to SessionDB + SessionStore state between turns, add localized status model/context lines, and keep token totals explicitly labeled cumulative.
Verification: tests/gateway/test_status_command.py, tests/hermes_cli/test_commands.py
Add a parser-only routing regression that proves raw WhatsApp group JIDs bypass channel-directory resolution and home-channel fallback, include channel_aliases.json in quick state snapshots, harden malformed alias handling, and map Keiron McCammon for release attribution.
send_message(target="whatsapp:<group-jid>") silently delivered to the
configured home DM instead of the requested group. Two gaps:
1. _parse_target_ref had no WhatsApp branch. Group JIDs (<id>@g.us),
user JIDs (<id>@s.whatsapp.net), linked-identity JIDs (<id>@lid), and
broadcast/newsletter JIDs matched no pattern and fell through to
`return None, None, False`, so the caller treated them as
unresolvable and used the home channel. The bridge's /send endpoint
accepts any chatId, so only the tool-side target parsing was at fault.
Add a whatsapp branch that recognizes native JIDs as explicit targets.
The pre-existing '+'-prefixed E.164 path is preserved.
2. WhatsApp groups have no human-friendly name — the channel directory
is regenerated from session data on a timer, so a group shows up as
its raw 18-digit JID and any hand-edit to channel_directory.json is
clobbered on the next rebuild. Add a user-maintained alias overlay
(~/.hermes/channel_aliases.json) re-applied on every build AND every
load, giving durable friendly names and letting a freshly-created
group be pre-named before its first message.
Tests: TestParseTargetRefWhatsAppJID (7 cases) for the parser;
TestChannelAliases (7 cases) for the overlay, plus an autouse fixture
isolating CHANNEL_ALIASES_PATH so a real alias file can't leak into the
existing directory tests.
On Linux, systemd spawns core services (cron, nginx, sshd) with
deterministic PIDs and jiffy start_times across reboots. A service can
land on the exact same PID and start_time as a previous gateway, causing
acquire_scoped_lock to mistake it for a live gateway and block startup.
The existing stale-detection paths only covered:
- start_times both non-None and different (clear mismatch)
- start_times both None (macOS/Windows fallback to cmdline check)
The boot-time collision falls through both: times are non-None and
equal, so neither branch fired.
Add a third check: when both start_times are known and match but the
live process fails _looks_like_gateway_process, read its cmdline. If
the cmdline is readable (non-None), we have positive evidence of an
impostor and mark the lock stale. Requiring a readable cmdline keeps the
check conservative — if cmdline is unreadable we do not evict.
The #45966 cross-process coherence guard snapshots a session's on-disk
message_count next to the cached agent and rebuilds the agent when the
count changes. But the snapshot is taken at agent-BUILD time — before
the turn writes its own user + assistant (+ tool) rows — and the cache
entry is never rewritten on a reuse. So this process's OWN turn grows
message_count, and the very next turn sees a mismatch and rebuilds the
agent. That happens every turn, for every conversation, silently
destroying the per-conversation prompt caching the cache exists to
protect (AGENTS.md: prompt caching is sacred).
Add _refresh_agent_cache_message_count(): after a turn completes and the
agent has flushed its rows to the SessionDB, re-baseline the stored count
to the now-current value. The guard then fires ONLY when a DIFFERENT
process changes the transcript — preserving the #45966 fix while keeping
the cache warm for normal single-process operation.
Tests drive the real SessionDB + the real guard condition: 5 consecutive
same-process turns now all REUSE the cached agent (0 before the fix); a
cross-process append still invalidates; and the re-baseline is fail-safe
(no DB, falsy session_id, raising probe, legacy 2-tuple, pending sentinel
all no-op).
Port 465 expects implicit TLS (SMTP_SSL) from the first byte. The email
adapter always used SMTP() + starttls(), which is correct for port 587
but hangs/fails on port 465 providers (e.g., Swiss ISPs).
Additionally, when the SMTP host has AAAA DNS records but IPv6 is
unreachable, socket.create_connection() tries IPv6 first and hangs
until timeout. Add an IPv4 fallback via AF_INET socket.
Extract _connect_smtp() helper to consolidate the 4 duplicate SMTP
connection sites into a single method with correct protocol selection
and IPv6 fallback logic.
Gateway startup now queues real inbound messages until restart-interrupted auto-resume turns have completed, preventing duplicate agents for the same session after a restart.
Registers 200 plugin commands on top of the native + COMMAND_REGISTRY set
and asserts the tree never exceeds Discord's 100-command limit, that native
high-priority commands survive the cap, and that overflow is actually
dropped. Regression guard for the recurring error 30032
("Maximum number of application commands reached") sync failures.
Carry forward focused follow-ups from PR #45741: treat PTB's raw Bot API 10.1 response shapes safely, recognize real missing-endpoint errors, preserve link preview settings on rich sends, and lock the rich limit to Telegram's character-based cap.
The salvaged fix's two regression tests mock adapter.handle_message, so
they only assert the pre-claimed sentinel is set/cleaned around a stub —
they never drive the real dispatch chain. Add a full-path test that
exercises _schedule_resume_pending_sessions -> _guarded_handle_message ->
adapter.handle_message -> _process_message_background -> _handle_message
and asserts the resumed session's agent runs EXACTLY ONCE: not zero (the
pre-claim must not self-bounce the resume into a queued no-op) and not
twice (the duplicate-agent bug #45456 the fix targets). Also assert no
leaked sentinel and no orphaned pending event after the drain settles.
Tighten the _guarded_handle_message docstring: on current main the real
sentinel is taken over inside _handle_message (not _process_message_background),
and note the `is _AGENT_PENDING_SENTINEL` guard only releases the slot we
ourselves placed, never one a live run owns.
When the gateway restarts and auto-resumes an interrupted session, an
inbound message arriving in the window between `asyncio.create_task()`
and the task's first await could spin up a second AIAgent for the same
session. Both agents would then process messages concurrently,
producing interleaved duplicate responses (#45456).
Fix: set `_AGENT_PENDING_SENTINEL` in `_running_agents` immediately
after the "already running" check, before creating the task. This
closes the race window — any inbound message sees the slot as occupied
and queues behind the auto-resume.
A `_guarded_handle_message` wrapper ensures the pre-claimed sentinel is
always released, even if `handle_message` raises before reaching
`_process_message_background` (whose `finally` block handles normal
cleanup).
(cherry picked from commit 85150c976b)
