Issues

• HTTP agent hooks: decide whether setup state may choose the turn URL host

  langwatch/tasks#1058 · Filter by author drewdrewthis opened on Oct 31, 2026

• Presidio truncates particle surnames in ordinary casing, dropping the final token

  bugF04:evalsP3 - lowsize:S
  langwatch/tasks#1057 · Filter by author sergioestebance opened on Oct 31, 2026

• default-test-lane guard: dedupe lanes by resolved config, name the CI lane, hint and loop-guard test

  langwatch/tasks#1056 · Filter by author drewdrewthis opened on Oct 30, 2026

• Person names written in lowercase are detected at 75%, against a 97.9% control

  bugF01:tracesP3 - lowsize:S
  langwatch/tasks#1055 · Filter by author sergioestebance opened on Oct 30, 2026

• Normalise ALL-CAPS tokens before person-name analysis so capitalised surnames are detected

  bugF01:tracesP1 - highsize:M
  langwatch/tasks#1054 · Filter by author sergioestebance opened on Oct 30, 2026

• Build the pi OTLP exporter extension as a standalone package and publish it

  architectureenhancementF02:sdkP2 - mediumsize:M
  langwatch/tasks#1053 · Filter by author sergioestebance opened on Oct 30, 2026

• Redaction consumes the opening quote when a finding starts a JSON string value

  bugF01:tracesP2 - mediumsize:S
  langwatch/tasks#1052 · Filter by author sergioestebance opened on Oct 30, 2026

• The Google DLP fallback stores a generic [REDACTED] instead of the entity marker

  bugF01:tracesP2 - mediumsize:S
  langwatch/tasks#1051 · Filter by author sergioestebance opened on Oct 30, 2026

• Redaction records an incomplete scan on spans but not on logs, metrics or the analysis-only route

  bugF01:tracesP2 - mediumsize:M
  langwatch/tasks#1050 · Filter by author sergioestebance opened on Oct 30, 2026

• Strict PII redaction misses person names whose surname is written in capitals

  bugF01:tracesP1 - highsize:M
  langwatch/tasks#1049 · Filter by author sergioestebance opened on Oct 30, 2026

• Strict PII redaction stores model, agent and tool names as [PERSON] under unlisted attribute keys

  bugF01:tracesP2 - mediumsize:XS
  langwatch/tasks#1048 · Filter by author sergioestebance opened on Oct 30, 2026

• Share the API_KEY_PEPPER → CREDENTIALS_SECRET → NEXTAUTH_SECRET chain in packages/secrets

  langwatch/tasks#1047 · Filter by author drewdrewthis opened on Oct 30, 2026

• Make the per-span PII redaction character budget configurable

  enhancementF01:tracesP2 - mediumsize:S
  langwatch/tasks#1045 · Filter by author sergioestebance opened on Oct 30, 2026

• EPIC: finish the Instant Evals decision model, from the catalogue to the models it cannot reach

  epicF04:evals
  langwatch/tasks#1043 · Filter by author sergioestebance opened on Oct 30, 2026

• HTTP agent hooks: low follow-ups from the 9193 security review

  langwatch/tasks#1042 · Filter by author drewdrewthis opened on Oct 30, 2026
  1

• Flaky tests: lantern-bleed-stacking browser test and use-instant-eval-runs integration test

  langwatch/tasks#1041 · Filter by author drewdrewthis opened on Oct 30, 2026

• LangWatchQL telemetry: production alert rule, log group and post-deploy check

  langwatch/tasks#1040 · Filter by author drewdrewthis opened on Oct 30, 2026

• proposal: Scenario Connector agent type with setup and teardown hooks and dataset fixtures

  langwatch/tasks#1039 · Filter by author drewdrewthis opened on Oct 30, 2026
  3

• inline run parameters: follow-ups from review (cap, shared save rules, one secret message, shared highlighter)

  langwatch/tasks#1038 · Filter by author drewdrewthis opened on Oct 30, 2026

• Scenario save should refuse a situation that declares a stored secret inline

  langwatch/tasks#1037 · Filter by author drewdrewthis opened on Oct 30, 2026

• HTTP agent update that omits the auth object drops the stored auth secret

  langwatch/tasks#1036 · Filter by author drewdrewthis opened on Oct 30, 2026

• Run plan results table shows 'Failed' for a run whose setup hook failed (should be 'Not scored')

  langwatch/tasks#1035 · Filter by author drewdrewthis opened on Oct 30, 2026

• Run dialog Parameters field: Tab picks a suggestion over the typed value

  langwatch/tasks#1034 · Filter by author drewdrewthis opened on Oct 30, 2026

• scenario-child: export the worker egress port so the hook SSRF test uses the real wiring

  langwatch/tasks#1033 · Filter by author drewdrewthis opened on Oct 30, 2026

• CLI governance unit tests overwrite the real Claude settings when CLAUDE_CONFIG_DIR is set

  bugP1 - highplanned
  langwatch/tasks#1032 · Filter by author drewdrewthis opened on Oct 30, 2026