BaaRai 9a927846ef Intercept judge: closed-question mode, fail-closed to ask ...

The LLM fallback judge gets a closed mode (default when the judge is on):
the model answers eight code-owned questions with yes|no|unknown and one
evidence line each, in one call; the decision is taken in code. Every
failure (invalid output, missing key, value outside the vocabulary, a
failed call, arguments above the 256 KiB cap) is ask. The custom judge
prompt becomes bounded operator-policy data (8000 runes). The answers are
stored in the intercept audit and shown in the approval detail.

freeform keeps the previous behaviour; defaultJudgeFailAction changes
from allow to ask.

2026-11-12 22:35:11 +01:00

..

adversarial_gate_seed_test.go

feat: adversarial verification gate before confirmed findings

2026-10-12 06:26:51 +02:00

agentic_guards_test.go

agent: frame agent-derived fields in a per-message nonce envelope (#124, F3)

2026-11-12 01:30:17 +01:00

assembly.go

Tool RAG: lexical projection of the bound tool catalog (off by default)

2026-11-12 15:42:31 +01:00

auto_scope_hygiene_test.go

fix(coverage): count auto-scope only under a declared root (F8)

2026-11-12 07:37:58 +01:00

bbot_ingest_test.go

agent: typed parser for bbot JSONL output (F10b)

2026-11-12 12:15:37 +01:00

bbot_ingest_tool_pg_test.go

Add ingest_bbot tool (bbot JSONL, typed scope split)

2026-11-12 12:35:06 +01:00

bbot_ingest_tool_test.go

Add ingest_bbot tool (bbot JSONL, typed scope split)

2026-11-12 12:35:06 +01:00

bbot_ingest_tool.go

Add ingest_bbot tool (bbot JSONL, typed scope split)

2026-11-12 12:35:06 +01:00

bbot_ingest.go

agent: typed parser for bbot JSONL output (F10b)

2026-11-12 12:15:37 +01:00

blackboard_inheritance_test.go

Skip the four known-broken tests in code instead of CI -skip flags

2026-10-10 18:12:42 +02:00

budget_test.go

Hierarchical run budgets: min(child, parent) with forced wrap-up (off by default)

2026-11-12 17:27:28 +01:00

budget.go

Coverage ledger: record examined, ruled-out and untested risk classes

2026-11-12 20:16:23 +01:00

cancelcause.go

feat(engine): intent claim heartbeat, TTL and fencing token (S3c)

2026-11-10 17:45:11 +01:00

capture_approval_test.go

Intercept judge: closed-question mode, fail-closed to ask

2026-11-12 22:35:11 +01:00

capture_usage_test.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

capture.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

chat.go

Untrusted boundary ON by default (opt-out via ARTEX_UNTRUSTED_BOUNDARY=0)

2026-11-12 14:46:44 +01:00

coldgraph_test.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

coldgraph.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

compaction.go

agent: fact provenance (declared source plus server-set origin)

2026-11-12 00:37:44 +01:00

constraints.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

coverage_ledger_test.go

Coverage ledger: review fixes (one-line capped fields, framed gap lines, per-run end_reason)

2026-11-12 20:48:34 +01:00

coverage_ledger.go

Coverage ledger: review fixes (one-line capped fields, framed gap lines, per-run end_reason)

2026-11-12 20:48:34 +01:00

data_envelope_test.go

agent: frame agent-derived fields in a per-message nonce envelope (#124, F3)

2026-11-12 01:30:17 +01:00

data_envelope.go

agent: frame agent-derived fields in a per-message nonce envelope (#124, F3)

2026-11-12 01:30:17 +01:00

deadline_contract_seed_test.go

feat: deadline contract in the executor seeds

2026-10-12 07:30:27 +02:00

deferred_test.go

chore: ARTEX 初始化

2026-07-26 04:27:45 -04:00

deferred.go

feat: 增加 Agent 流量绑定开关并补齐证据交接

2026-09-10 14:08:40 +08:00

fact_provenance_test.go

agent: fact provenance (declared source plus server-set origin)

2026-11-12 00:37:44 +01:00

finding_dedup_seed_test.go

feat: duplicate discipline before report_finding and in the reporter

2026-10-12 07:29:59 +02:00

finding_quality_seed_test.go

feat: measured impact and attacker-goal proof standard for findings

2026-10-12 06:25:35 +02:00

finding_reasoning_test.go

feat(findings): optional counterevidence and confidence_rationale

2026-11-11 23:43:12 +01:00

finding_recorder_test.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

finding_recorder.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

finding_workflow_test.go

feat: 增加 Agent 流量绑定开关并补齐证据交接

2026-09-10 14:08:40 +08:00

finding_workflow.go

feat: verified findings backend (S1)

2026-11-09 20:21:27 +01:00

goals.go

fix(coverage): count auto-scope only under a declared root (F8)

2026-11-12 07:37:58 +01:00

hooks_reach_test.go

feat(guard): planner and main-agent hooks, target-gate scope in every session (S3b 1.1)

2026-11-10 22:14:24 +01:00

insert_assets_test.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

mainagent.go

Untrusted boundary ON by default (opt-out via ARTEX_UNTRUSTED_BOUNDARY=0)

2026-11-12 14:46:44 +01:00

noa.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

operator_handoff_seed_test.go

feat: ask_operator environmental handoff discipline

2026-10-12 06:45:21 +02:00

planner.go

Coverage ledger: review fixes (one-line capped fields, framed gap lines, per-run end_reason)

2026-11-12 20:48:34 +01:00

prompt_injection_seed_test.go

agent: untrusted-content tail for every chat agent; evidence wording fits the reporter

2026-11-11 21:59:09 +01:00

prompt_now_test.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

prompt_test.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

prompt.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

promptcatalog.go

agent: fact provenance (declared source plus server-set origin)

2026-11-12 00:37:44 +01:00

provider_capture_e2e_test.go

feat(llmrec): 录制 LLM 请求/响应 HTTP 原文

2026-08-25 09:40:05 -04:00

provider_capture_test.go

feat(llm): optional custom User-Agent for LLM requests

2026-11-08 17:11:43 +01:00

provider_env_test.go

fix(agent): match ARTEX_LLM_PROVIDER case-insensitively

2026-11-03 23:20:21 +01:00

provider_quota_test.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

provider_responses_test.go

feat(llm): 支持 OpenAI Responses API 格式的 LLM 配置

2026-08-25 11:39:24 -04:00

provider.go

feat(llm): optional custom User-Agent for LLM requests

2026-11-08 17:11:43 +01:00

proxyenv_test.go

feat(proxy): 新增全局出口代理配置

2026-09-07 10:43:38 -04:00

retester.go

feat: adversarial verification gate before confirmed findings

2026-10-12 06:26:51 +02:00

review_context_test.go

Intercept judge: closed-question mode, fail-closed to ask

2026-11-12 22:35:11 +01:00

runinfo.go

feat(tasks): add runtime controls and inherited context

2026-08-24 19:29:00 +08:00

session_header_test.go

feat(llm): 每个 LLM 配置支持自定义会话头

2026-09-07 11:21:50 -04:00

side_questions_test.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

side_questions.go

feat(agent): 支持持久化 /btw 旁路提问

2026-09-10 21:42:25 +08:00

stagnation_test.go

feat(engine): ouroboros takes - goal ambiguity gate, failover escalation, stagnation surfacing

2026-10-21 14:46:56 +02:00

target_gate_tools_test.go

feat(intercept): target extraction, resolver, exemptions and the gate in the hook (S3b 1.5-1.7)

2026-11-11 00:34:55 +01:00

taskclock.go

chore: ARTEX 初始化

2026-07-26 04:27:45 -04:00

terminalreason_test.go

feat(engine): intent claim heartbeat, TTL and fencing token (S3c)

2026-11-10 17:45:11 +01:00

terminalreason.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

testmain_test.go

feat: each test package runs on its own database

2026-10-12 18:43:29 +02:00

toolcatalog_test.go

fix(tools): harden obj against blank required names and heal seeded list_task_credentials rows

2026-11-10 09:46:49 +01:00

toolcatalog.go

feat(tasks): per-task test-account credentials

2026-10-09 14:28:08 +02:00

toolrag_test.go

Tool RAG: pin the keep set (harness, Skill, deferred) in a test

2026-11-12 15:58:34 +01:00

toolrag.go

Tool RAG: lexical projection of the bound tool catalog (off by default)

2026-11-12 15:42:31 +01:00

tools_creds.go

fix(tools): drop empty required entry from list_task_credentials schema

2026-11-02 09:05:03 +03:00

tools_digest.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

tools_insert.go

Coverage ledger: record examined, ruled-out and untested risk classes

2026-11-12 20:16:23 +01:00

tools_nil_store_test.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

tools_overview_test.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00

tools.go

Coverage ledger: record examined, ruled-out and untested risk classes

2026-11-12 20:16:23 +01:00

untrusted_boundary_test.go

Untrusted boundary ON by default (opt-out via ARTEX_UNTRUSTED_BOUNDARY=0)

2026-11-12 14:46:44 +01:00

untrusted_boundary.go

Untrusted boundary ON by default (opt-out via ARTEX_UNTRUSTED_BOUNDARY=0)

2026-11-12 14:46:44 +01:00

untrusted_prompt.go

agent: untrusted-content tail for every chat agent; evidence wording fits the reporter

2026-11-11 21:59:09 +01:00

user_agent_test.go

fix(llm): validate custom User-Agent in the connection-test path

2026-11-08 18:15:33 +01:00

webfetch_dialer_test.go

feat(intercept): pinned dialer for WebFetch and the custom http tool (S3b 2.1)

2026-11-11 19:19:55 +01:00

webfetch_egress_test.go

WebFetch: coalesce only, never cache; pin the gate-off redirect test to the off profile

2026-11-12 16:33:14 +01:00

webfetch_redirect_test.go

WebFetch: coalesce only, never cache; pin the gate-off redirect test to the off profile

2026-11-12 16:33:14 +01:00

webfetch.go

Egress: return large unshared responses without an error, namespace the transport cache per client, pace WebFetch only

2026-11-12 16:44:08 +01:00

worker_intervention_test.go

新增worker直接对话

2026-08-29 21:55:27 +08:00

worker.go

Coverage ledger: record examined, ruled-out and untested risk classes

2026-11-12 20:16:23 +01:00

wrapup.go

Translate all Chinese text to English

2026-10-08 23:14:34 +02:00