adlee-was-taken 8c76072eb2 fix(watchdog): compare the last tick in milliseconds, like opencode's call times ...

last_tick is epoch seconds while opencode's call start times are epoch
milliseconds, so every call ever made looked "since the last tick". Each
tick re-judged the whole session history: a dry run against live opencode
triggered three alerts (two critical) for sessions idle since the night
before, which could never resolve because their history never changes.

The tests hid it by stubbing call times in seconds. They now use
milliseconds, and the idle-session test carries a 60-call loop that would
flag if judged, plus the tables a full tick reads, so a swallowed crash can
no longer pass as "no alert". It fails without this fix.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N9biTbFC63yDfYfUsZmhgd

2026-09-26 15:15:18 -04:00

..

fixtures

feat: export calibration fixture for progress detection

2026-09-26 11:28:45 -04:00

conftest.py

feat(admin): re-poll the catalog after an allowlist edit

2026-09-09 19:16:08 -04:00

test_admin_allowlist_and_fallback_contracts.py

fix(admin): validate cloud base_urls, and let an allowlist note be corrected

2026-09-08 18:39:17 -04:00

test_admin_allowlist_repoll.py

feat(admin): re-poll the catalog after an allowlist edit

2026-09-09 19:16:08 -04:00

test_admin_allowlist.py

test(allowlist): full suite — config, poller, admin, integration

2026-09-06 18:33:14 -04:00

test_admin_bodies_are_strict.py

fix(admin): unknown request keys are errors, and base providers are read-only

2026-09-08 18:30:06 -04:00

test_admin_classifier_config.py

feat(admin): a Classifier card, reusing the write path and the live resolver

2026-09-05 23:54:43 -04:00

test_admin_cloud_fallback_config.py

feat(admin): add classifier.cloud_fallback control to Controls page

2026-09-07 18:47:40 -04:00

test_admin_config.py

refactor(session-cache): rename staleness_minutes to staleness_seconds

2026-09-18 00:45:08 -04:00

test_admin_frontend.py

fix(admin): decisions tile reports route decisions and client outcomes

2026-09-26 03:03:30 -04:00

test_admin_health.py

fix(admin): decisions tile reports route decisions and client outcomes

2026-09-26 03:03:30 -04:00

test_admin_history.py

refactor: move config data files into config/ and update all references

2026-08-30 00:03:05 -04:00

test_admin_js_units.py

feat(admin): one knob table merges runtime and persisted state

2026-09-26 03:03:30 -04:00

test_admin_knob_coverage.py

feat: no-progress-lift-a — full implementation (watchdog rework, admin, tests)

2026-09-26 13:09:27 -04:00

test_admin_models.py

fix(admin): show a profile's models, and pick them from a two-pane list

2026-09-08 10:59:04 -04:00

test_admin_nav_is_complete.py

refactor(admin): render nav links from navbar.js instead of eight copies

2026-09-26 03:03:06 -04:00

test_admin_override_binds_everywhere.py

fix: an admin availability override binds every path, and "stale" works

2026-09-08 18:33:50 -04:00

test_admin_proficiency.py

refactor(admin): render nav links from navbar.js instead of eight copies

2026-09-26 03:03:06 -04:00

test_admin_profile_is_default.py

fix(admin): the profiles page claimed a restart the API says is not needed

2026-09-12 11:54:25 -04:00

test_admin_profiles.py

feat(admin): set the routing default from the profiles page; fold the allowlist

2026-09-08 20:49:36 -04:00

test_admin_provider_readonly.py

fix(admin): unknown request keys are errors, and base providers are read-only

2026-09-08 18:30:06 -04:00

test_admin_providers.py

fix(admin): return provider provenance flags in list and detail endpoints

2026-09-06 02:02:57 -04:00

test_admin_routing_override.py

fix: an admin availability override binds every path, and "stale" works

2026-09-08 18:33:50 -04:00

test_admin_runtime.py

feat(admin): one knob table merges runtime and persisted state

2026-09-26 03:03:30 -04:00

test_admin_snapshot.py

fix(admin): decisions tile reports route decisions and client outcomes

2026-09-26 03:03:30 -04:00

test_admin_triggers.py

refactor: move config data files into config/ and update all references

2026-08-30 00:03:05 -04:00

test_admin_watchdog.py

feat: no-progress-lift-a — full implementation (watchdog rework, admin, tests)

2026-09-26 13:09:27 -04:00

test_allowlist_filter_structure.py

feat(admin): set the routing default from the profiles page; fold the allowlist

2026-09-08 20:49:36 -04:00

test_allowlist.py

test(allowlist): full suite — config, poller, admin, integration

2026-09-06 18:33:14 -04:00

test_apply_tiering.py

refactor: move config data files into config/ and update all references

2026-08-30 00:03:05 -04:00

test_balance_config.py

feat(config): DispatchProvider.balance_url with fail-closed parser/telemetry validators

2026-09-06 12:47:49 -04:00

test_baseline_report.py

feat(baseline_report): incumbent-free counterfactual for the challenger dial

2026-09-13 20:11:56 -04:00

test_blocked_routing.py

fix: ruff style compliance on new files

2026-09-26 13:15:25 -04:00

test_capabilities.py

test: pin iter_image_url_values contract on well-typed-but-unparseable input

2026-08-23 16:02:25 -04:00

test_chat_completions.py

refactor(session-cache): rename staleness_minutes to staleness_seconds

2026-09-18 00:45:08 -04:00

test_cheapest_classifier_candidate.py

feat(classifier): configurable backend -- local_llm / cloud_llm / local_encoder

2026-09-05 23:53:44 -04:00

test_circuit_breaker.py

fix: gate streaming breaker on stream_completed, register content-fault warning tripwire

2026-09-09 22:55:26 -04:00

test_classifier_backoff.py

fix(tests): migrate backoff/modes/gaming fixtures to per-provider refusal state

2026-09-06 02:43:19 -04:00

test_classifier_candidate_categories.py

fix(local-encoder): update test references from confidence_threshold to confidence_min

2026-09-23 01:22:59 -04:00

test_classifier_cascade.py

test(dispatch): multi-provider pin resolution, telemetry, and refusal tests

2026-09-06 01:43:52 -04:00

test_classifier_input.py

fix: resolve context-pruning-and-framing review findings

2026-08-23 22:23:08 -04:00

test_classifier_modes_config.py

fix(local-encoder): update test references from confidence_threshold to confidence_min

2026-09-23 01:22:59 -04:00

test_classifier_modes_dispatch.py

fix(local-encoder): update test references from confidence_threshold to confidence_min

2026-09-23 01:22:59 -04:00

test_config_endpoints.py

fix(config): adoption_window_seconds comments match the validator

2026-09-20 21:43:21 -04:00

test_config_watchdog.py

test: config validation tests for new watchdog config sections

2026-09-26 11:42:05 -04:00

test_config.py

feat(config): incumbent knobs — challenger dial, refresh TTL, pricing floor; gated off

2026-09-13 19:38:12 -04:00

test_context_prune.py

fix(pinch): cap outsized tool results even inside the protected window

2026-09-06 19:10:24 -04:00

test_conversation_identity.py

fix(conversation): first occurrence of a duplicate header wins

2026-09-20 21:27:52 -04:00

test_conversation_wiring.py

feat(dispatcher): session_key from X-Router-Conversation, agent and parent recorded

2026-09-20 20:21:09 -04:00

test_credit_attenuation_config.py

feat(config): objective.credit_attenuation block, off by default

2026-09-06 12:50:27 -04:00

test_credit_attenuation_routing.py

fix(test): reprice attenuation fixture so OFF→ON flip is observable

2026-09-06 14:12:54 -04:00

test_decisions_page_structure.py

fix(admin): drop the Kind column, keep the 0.2% it was hiding

2026-09-10 22:21:09 -04:00

test_dispatcher_helpers.py

test: make the HTTP fakes exercise the boundary instead of modelling it

2026-09-08 14:05:16 -04:00

test_dispatcher_local_energy.py

fix(local-encoder): update test references from confidence_threshold to confidence_min

2026-09-23 01:22:59 -04:00

test_docs_content.py

docs: client conversation headers and conversation_id

2026-09-20 20:28:22 -04:00

test_eval_proficiency_logging.py

fix(eval): call_model's call site had a positional/keyword argument collision

2026-09-06 22:36:10 -04:00

test_eval_scoring.py

test(eval_scoring): bound infinite-loop test timeout to 1s

2026-09-01 23:05:09 -04:00

test_events.py

fix(events): guard publish_decision against concurrent subscriber churn

2026-08-28 19:02:47 -04:00

test_exploration.py

test(proficiency): step-2 gate (persisted-cost inertness) + property tests for feedback + exploration

2026-09-01 22:10:22 -04:00

test_feedback_preview.py

fix(feedback): align dry-run preview count with projected snapshot; emit CSV header on empty backlog

2026-09-18 00:00:19 -04:00

test_feedback.py

fix(feedback): align dry-run preview count with projected snapshot; emit CSV header on empty backlog

2026-09-18 00:00:19 -04:00

test_gaming_mode.py

fix(tests): migrate backoff/modes/gaming fixtures to per-provider refusal state

2026-09-06 02:43:19 -04:00

test_incumbent_routing.py

test(dispatcher): flag gate proves DB-query absence off, dial kwarg presence on

2026-09-13 20:53:34 -04:00

test_iteration.py

feat(iteration): budget retries in prompt re-bills; prefer cache-preserving same-model retry on malformed

2026-09-18 00:32:55 -04:00

test_leaderboard.py

refactor(leaderboard): thread provider into apply_priors from config key

2026-09-04 20:55:41 -04:00

test_load_candidates.py

refactor: move config data files into config/ and update all references

2026-08-30 00:03:05 -04:00

test_local_dispatch_fallback.py

test: make the HTTP fakes exercise the boundary instead of modelling it

2026-09-08 14:05:16 -04:00

test_local_dispatch.py

test(local_dispatch, proficiency): honor new default_provider validation

2026-09-04 21:01:04 -04:00

test_local_encoder.py

fix(local-encoder): dispatch pooling strategy by model_id, not load order

2026-09-23 16:20:15 -04:00

test_local_energy.py

feat: session attribution + local energy ledger (Phase 1/2)

2026-09-01 00:20:10 -04:00

test_logs.py

feat: one line per request saying what was decided, and why

2026-08-22 21:52:58 -04:00

test_measured_cache_rates.py

feat(dispatcher): measured per-model cache rates with TTL + independent pricing floor

2026-09-13 19:47:24 -04:00

test_metrics_calibration_and_latency.py

feat(metrics): report the cost-estimator's error and the latency nobody could see

2026-09-16 14:34:27 -04:00

test_metrics_endpoint.py

feat(metrics): report the cost-estimator's error and the latency nobody could see

2026-09-16 14:34:27 -04:00

test_metrics.py

fix(admin): decisions tile reports route decisions and client outcomes

2026-09-26 03:03:30 -04:00

test_models_page_structure.py

fix(admin): tighten the models table and show what routing turns on

2026-09-08 22:26:39 -04:00

test_multi_provider_dispatch.py

feat(telemetry): a NULL cached_tokens now says which kind of nothing it was

2026-09-13 11:04:02 -04:00

test_multi_provider_poller.py

feat(poller): add allowlist filter+prune for openrouter provider

2026-09-06 18:11:01 -04:00

test_no_header_snapshot.py

test(dispatcher): baseline no-header route_decisions snapshot fixture

2026-09-20 20:03:07 -04:00

test_notifier.py

fix: ruff style compliance on new files

2026-09-26 13:15:25 -04:00

test_openrouter_poller.py

feat(poller): fetch and normalize OpenRouter catalog

2026-09-06 01:16:27 -04:00

test_outcome_attribution.py

feat(outcome): attribute a report to an exact conversation

2026-09-20 20:26:16 -04:00

test_plans_declare_status.py

docs(plans): give every plan a Status line so the queue is greppable

2026-09-08 18:55:16 -04:00

test_poller_freshness.py

feat(poller): multi-provider main loop with failure isolation and tiering

2026-09-06 01:29:15 -04:00

test_poller_parsing.py

fix(poller): cap the output reserve so a catalog field cannot zero a model

2026-09-09 18:34:35 -04:00

test_prefix_probe.py

feat(telemetry): a prefix-stability probe that stores no prefix

2026-09-13 11:37:59 -04:00

test_proficiency.py

fix(proficiency): peer_rate is the sample-weighted pooled rate per plan §2

2026-09-08 21:24:58 -04:00

test_progress_backtest.py

fix: use sys.executable for backtest subprocess instead of worktree-path-based VENV_PYTHON

2026-09-26 14:58:01 -04:00

test_progress_detect.py

fix: canonicalize args for dup/slow, pass RO keywords to backtest

2026-09-26 14:46:29 -04:00

test_provider_balance.py

feat: store total_credits_usd and total_usage_usd from balance poller

2026-09-10 22:05:12 -04:00

test_route_decisions.py

feat(route-decisions): agent and parent_key columns, surfaced on purpose

2026-09-20 19:54:14 -04:00

test_router_cli.py

feat(cli): router_cli one-shot routing-decision printer

2026-08-23 20:19:46 -04:00

test_router_wall_clock.py

feat(telemetry): router-observed wall-clock and TTFT columns, with plan docs

2026-09-13 12:25:39 -04:00

test_routing_profiles_integration.py

fix(api): /route and /dispatch honour routing.default_profile

2026-09-09 21:19:09 -04:00

test_routing.py

test(routing): pin the load-bearing challenger clamp across the dial class

2026-09-13 20:53:34 -04:00

test_scoring.py

feat: retarget router to NeuralWatt-only, add serving-class routing and an OpenAI-compatible dispatcher

2026-08-12 00:03:43 -04:00

test_seed_local_dispatch.py

fix: local dispatch metering + seed module LOC ceiling

2026-09-02 02:09:59 -04:00

test_seed_sweep.py

refactor(seed_energy): cloud provider from config key

2026-09-04 20:56:15 -04:00

test_session_cache.py

feat(session_cache): pure in-memory classification cache module

2026-08-26 20:58:12 -04:00

test_session_identity.py

feat: session attribution + local energy ledger (Phase 1/2)

2026-09-01 00:20:10 -04:00

test_suite_ignores_machine_overlay.py

fix(config): stop the test suite reading the machine's config.local.yaml

2026-09-08 18:51:22 -04:00

test_task_set.py

fix(evals,seed): unanswerable diff_checking question + two seed-script defects

2026-09-03 19:33:14 -04:00

test_textcut.py

fix: cut text on character boundaries, and write up the mutation-point audit

2026-09-08 14:15:35 -04:00

test_tiering.py

fix: price is a market signal, not a capability measurement

2026-08-21 19:03:17 -04:00

test_truncation_is_cluster_safe.py

fix: cut text on character boundaries, and write up the mutation-point audit

2026-09-08 14:15:35 -04:00

test_tui_schema_drift.py

docs(tests): drift-test comments match the registries

2026-09-20 21:43:21 -04:00

test_tui_warnings.py

feat(metrics): a cache-rate series, and an expiry check on assumed_cache_rate

2026-09-13 11:19:13 -04:00

test_tui.py

fix(tui): render blank ctx cell instead of literal "None"

2026-09-18 23:31:07 -04:00

test_verification.py

fix: an agent turn that calls a tool is not a failed answer

2026-08-21 19:02:07 -04:00

test_watchdog.py

fix(watchdog): compare the last tick in milliseconds, like opencode's call times

2026-09-26 15:15:18 -04:00