The first checkpoint removed terminal-metrics discovery, but a normal
compact chat turn still ran two context systems: build_chat_context's
legacy untyped lookup (directly or inside maybe_compact) and the typed
resolver inside stream_preview.
Resolve the typed ContextResolution once, at the chat route, before
build_chat_context, using the session's provider credentials. The
predicate mirrors _clean_v3_preview; every input it needs is known at
that point and the native-workspace term cannot veto a requested clean
route. The same object then:
- sizes legacy history shaping in build_chat_context through a new
maybe_compact(context_length=...) override, so no legacy probe runs;
an unknown window still shapes with DEFAULT_CONTEXT but gains no
provenance;
- crosses stream_agent_loop (one new parameter, forwarded only at the
compact dispatch) into stream_preview, which reuses it and probes only
for callers that arrive without one or with one bound to another
route.
ContextResolution now records the endpoint and model it describes
(endpoint URL excluded from repr and metrics). The bare legacy
context_length is never converted into typed evidence.
Credential scoping: origins compare with default ports normalized, an
empty host is never trusted, and the probe client never follows
redirects. Tests cover the configured origin, the server-resolved
Tailscale form, scheme/port/lookalike/userinfo/path origins, redirects,
and secret-free errors, logs and metrics.
The conftest guard now replaces only the resolver's I/O edges (HTTP
client and DNS-capable URL building) instead of the whole probe, and
exposes a context_probe_ledger fixture, so route integration tests run
the real resolver offline and can count metadata requests.
* refactor(model-routing): centralize explicit foreground fallback policy
Make foreground fallback an explicit per-user, availability-only policy shared by streaming Chat, non-stream Chat, and Agent runs.
Preserve strict defaults, owner/model and credential boundaries, pinned Agent routes, and truthful per-round provenance/accounting. Carry provider-reported model identifiers through native streaming adapters, non-stream responses, and caches, and keep legacy default_model_fallbacks as tombstoned raw storage that generic settings APIs and agent tools cannot expose or mutate.
* fix(agent-loop): restore rebase-dropped qwen routing, workspace prompt, and temperature clamp
* fix(model-routing): thread selected endpoint identity, fix cost classification and fallback eligibility
* fix(chat): restore stream helpers and harden run stop lifecycle
* fix(model-routing): let numeric provider codes win over symbolic rate-limit statuses
* fix(agent-loop): apply qwen temperature and notes-tool clamps per fallback candidate
* fix(chat): honor queued stop across resend and reload canonical terminal on EOF
* fix(chat): track stop queue and cleanup ownership by per-send generation
* fix(agent-loop): preserve requested temperature for non-qwen fallback candidates
* fix(chat): reserve send ownership before any await and scope stop to the current send
* fix(chat): clear the previous run identity at send reservation
---------
Co-authored-by: RaresKeY <158580472+RaresKeY@users.noreply.github.com>
Co-authored-by: StressTestor <212606152+StressTestor@users.noreply.github.com>
* fix(research): preserve Discuss spin-off primer during context trimming
trim_for_context() kept only system_msgs[:1] as essential and dropped the
rest under budget pressure. A research "Discuss" spin-off seeds the report
as a system message that sits after the preface system messages, so it
landed in extra_system and was the first thing evicted once the chat grew
— the conversation then lost its grounding and drifted off task.
Treat any system message carrying research_spinoff_from metadata as
essential, alongside the leading system prompt, so the seeded report
survives trimming. maybe_compact already retains all system messages.
Tests: tests/test_context_compactor.py::TestResearchPrimerPreserved
* fix(research): ground Discuss spin-off chats on the seeded report
build_chat_context injected global memory (pinned + hybrid-retrieved) and
personal-doc RAG every turn, keyed off the user-level memory_enabled pref
and a request-scoped use_rag flag — never the session. A research spin-off,
whose primer declares the report the sole knowledge base, thus had
unrelated keyword-matched facts pulled in ("wrong data") competing with the
report; its rag=False flag was also ignored (use_rag defaulted on).
Add _session_is_research_spinoff(sess) (detects the primer research_spinoff_from
metadata; handles ChatMessage and dict forms) and, for such sessions,
disable memory injection and force RAG off.
Tests: tests/test_chat_helpers.py spin-off detection cases
---------
Co-authored-by: Dan (cirim) <claude@cirim.org>
* docs: add implementation plan for fixing chat context drifting (#135)
* fix: make Session.history immutable + fix {}.history crash
- Session.history now exposes a COPY of the internal _history list
- add_message() replaces history with a fresh copy each time
- get_context_messages() derives from _history directly
- replace_messages() updates both _history and history
- truncate_messages() updates both _history and history
- _persist_message() line 207: fixed {}.history fallback crash
- Added 11 tests for session isolation and edge cases
Addresses #135 root cause #1: shared mutable references
* fix: task scheduler uses SessionManager methods instead of overwriting sessions
- Added ensure_task_session() to SessionManager (checks cache first)
- Task scheduler now uses ensure_task_session() instead of direct dict assignment
- Task scheduler now uses SessionManager.add_message() for message persistence
- Removed direct sess_obj.history.append() that was silently losing data
Addresses #135 root causes #2 and #3
* fix: add age guard to cleanup_empty_sessions — don't delete sessions <1h old
Prevents the cleanup task from deleting sessions that were just created
and haven't received any messages yet (message_count == 0).
Addresses #135 root cause #5
* test: comprehensive session isolation tests (10/10 passing)
* refactor: consolidate _session_manager into singleton pattern
- Added set_session_manager_instance / get_session_manager_instance to core/models
- kept backward-compat aliases (set_session_manager, get_session_manager)
- session_manager.py re-exports the singleton functions
- ai_interaction.set_session_manager now syncs with the core singleton
- context_compactor uses get_session_manager_instance() instead of getattr hack
- app.py initializes the singleton once
Addresses #135 root cause #4: fragile global wiring
* test: add concurrent session isolation integration tests
Verifies:
- Concurrent add_message to different sessions doesn't cross-contaminate
- Rapid parallel writes maintain isolation
- Read-write concurrent access is safe
All 3 async tests pass, proving the immutable history fix works under concurrency
* fix: pre-import core.models in conftest to prevent test pollution
test_agent_loop.py stubs sys.modules['core.models'] = MagicMock() at
module level during collection. Any test collected after it imports
Session as a MagicMock. Pre-importing core.models in conftest.py
before test_agent_loop.py's module-level code runs prevents this.
* fix: make .history authoritative mutable list, address PR review
Per review feedback: keep .history as the authoritative mutable list so
existing code doing .history.pop(), .history = [...], etc. still works.
Fix the cross-contamination bug by ensuring __post_init__() gives each
Session its OWN unique history list (never shared).
Changes:
- core/models.py: .history IS the authoritative list. _history aliases it.
Each Session gets its own list in __post_init__.
- core/session_manager.py: add_message() delegates to Session.add_message()
instead of appending directly — no double-append, single source of truth.
- tests/test_session_manager.py: updated test to reflect that .history
references see new messages (same list, not a snapshot).
- docs/plans/2026-06-01-fix-chat-context-drifting.md: removed (not for
shipping — useful design context but too much process/doc to ship).
All 272 tests pass (3 pre-existing failures unrelated).
* Fix session manager message persistence
* Fix session history alias regressions
* Fix session history aliasing and task delivery
On summary LLM call failure, maybe_compact was returning system_msgs+recent
(dropping the older half) with was_compacted=False, misleading the caller into
thinking the list was unchanged. Return the original messages list unchanged so
no history is lost; the next trim_for_context call handles length if needed.
Fixes#2160
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
* fix: context_compactor token helpers crash on non-string message text
* fix: _truncate_text_to_token_budget returns an empty string for non-string text, not the raw value
context_compactor.maybe_compact built its summary text with
msg.get('content', '')[:2000], which raised
TypeError: 'NoneType' object is not subscriptable on assistant turns
whose content is None (turns that carried only native tool_calls).
Once a conversation crossed the 85% compaction threshold — reached
after only a few turns on small-context local models plus the large
agent prompt — every subsequent message failed ("send more than three
messages and it stops working").
Flatten message content to text first via a _content_as_text helper
(str passthrough, multimodal list blocks joined, None -> "") and
tolerate a missing role. Adds tests/test_context_compactor.py covering
the helper and a >=4-message conversation that forces compaction with
a None-content tool-call turn (fails before this change, passes after).
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The context compactor computed split_point against convo_msgs (system
messages filtered out) but applied it directly to session.history which
includes the system messages. After compaction, the original system
prompt was dropped and replaced by an off-by-N slice of the full history.
This silently dropped the system prompt (preset, persona, RAG context)
from every compacted session — the model would lose persona, RAG, and
preset guidance on the next turn after a long conversation.
The split in maybe_compact does:
convo_msgs = [m for m in messages if m['role'] != 'system']
split_point = len(convo_msgs) // 2
so split_point is indexed against the system-stripped list. But the
helper _update_session_history took (session, split_point, summary) and
did session.history[split_point:]. session.history is the full list
including the leading system messages, so this dropped the first
system_msg_count messages.
Fix: pass system_msg_count=len(system_msgs) into _update_session_history
and use session.history[system_msg_count + split_point:] as the recent
slice, with session.history[:system_msg_count] prepended to preserve
persona/preset/RAG system messages.
Validated: tests/test_compactor_data_loss.py both tests now pass (were
failing). tests/test_context_compactor.py 12 pre-existing tests still
pass.
Symptom was: post-compaction history = [summary] + assistant_1 + user_2
+ assistant_2 (system_A was lost).
Co-authored-by: Ernest Hysa <ernest@example.com>
