okstra 0.186.4 → 0.186.6

package/runtime/python/okstra_token_usage/collect.py

@@ -2,6 +2,7 @@
 from __future__ import annotations
 
 import json
+import os
 from datetime import datetime, timezone
 from pathlib import Path
 from typing import Any
@@ -13,9 +14,18 @@ from .claude import (
     find_claude_agent_sessions,
     find_claude_team_sessions,
 )
-from .codex import codex_session_total, find_codex_sessions
+from .codex import (
+    codex_session_ids,
+    codex_session_is_worker,
+    codex_session_total,
+    find_codex_sessions,
+)
 from .antigravity import antigravity_session_total, find_antigravity_sessions
-from .grok import find_grok_sessions, grok_session_total
+from .grok import (
+    find_grok_sessions,
+    grok_session_is_non_interactive,
+    grok_session_total,
+)
 from .paths import claude_project_dir, utc_now
 from .pricing import antigravity_cost_usd, provider_cost_usd
 from okstra_ctl.dispatch_state import worker_session_ids
@@ -94,13 +104,9 @@ def _aggregate_totals(items: list[dict]) -> dict:
             aggregate["startedAt"] = s
         if e and (aggregate["endedAt"] is None or e > aggregate["endedAt"]):
             aggregate["endedAt"] = e
-    if aggregate["startedAt"] and aggregate["endedAt"]:
-        try:
-            a = datetime.fromisoformat(aggregate["startedAt"].replace("Z", "+00:00"))
-            b = datetime.fromisoformat(aggregate["endedAt"].replace("Z", "+00:00"))
-            aggregate["durationMs"] = max(0, int((b - a).total_seconds() * 1000))
-        except ValueError:
-            pass
+    wall = _wall_ms(aggregate["startedAt"], aggregate["endedAt"])
+    if wall is not None:
+        aggregate["durationMs"] = wall
     return aggregate
 
 
@@ -557,9 +563,18 @@ def wrapper_execution(status_path: Path | None) -> dict:
     }
     if status.exit_code is not None:
         execution["exitCode"] = status.exit_code
+    duration = _wrapper_duration_ms(status.raw.get("duration_ms"), execution)
+    if duration is not None:
+        execution["durationMs"] = duration
     return execution
 
 
+def _wrapper_duration_ms(raw_duration: object, execution: dict) -> int | None:
+    if isinstance(raw_duration, (int, float)) and not isinstance(raw_duration, bool):
+        return max(0, int(raw_duration))
+    return _wall_ms(execution.get("startedAt"), execution.get("endedAt"))
+
+
 def _execution_note(execution: dict) -> str:
     status = execution["status"]
     if status == "exited":
@@ -603,6 +618,8 @@ def collect_cli_usage(
         block = _cli_usage_block(provider, _aggregate_totals(totals), sessions)
 
     block["cliExecutionStatus"] = execution["status"]
+    if "durationMs" in execution:
+        block["durationMs"] = execution["durationMs"]
     if fallback_window_used:
         fallback_note = "wrapper status sidecar unavailable; used aggregate wrapper window fallback"
         prior_note = block.get("cliNote")
@@ -706,11 +723,10 @@ def _attach_cli_usage(
             block[key] = cli[key]
 
 
-def _collect_cli_runtime_usage(state: dict, project_root: Path) -> dict:
+def _collect_cli_runtime_usage(
+    state: dict, project_root: Path, team_state_path: Path | None = None,
+) -> dict:
     windows_by_worker = _codex_worker_windows(project_root, state)
-    state["leadUsage"] = na_block(
-        "Host lead token accounting is unavailable; CLI worker usage is collected only from attributable provider logs."
-    )
     for worker in state.get("workers", []):
         if not isinstance(worker, dict):
             continue
@@ -728,11 +744,128 @@ def _collect_cli_runtime_usage(state: dict, project_root: Path) -> dict:
             worker=worker,
             fallback_windows=windows_by_worker.get(worker_id, []),
         )
+    state["leadUsage"] = _cli_lead_usage(state, project_root, team_state_path)
     _populate_usage_summary(state, team_name=resolve_team_name(state),
                             sessions_found=0, needle_source="none")
     return state
 
 
+def _cli_lead_provider(state: dict) -> str:
+    lead = state.get("lead") if isinstance(state.get("lead"), dict) else {}
+    return str(
+        lead.get("provider") or lead.get("agent") or state.get("leadRuntime") or ""
+    ).strip()
+
+
+_CLI_LEAD_PROVIDERS = frozenset({"grok", "codex"})
+
+
+def _cli_lead_usage(
+    state: dict, project_root: Path, team_state_path: Path | None,
+) -> dict:
+    """호스트 리드 세션은 워커가 가져간 경로를 뺀 뒤 같은 트랜스크립트에서 읽는다."""
+    missing = (
+        "Host lead token accounting is unavailable; CLI worker usage is "
+        "collected only from attributable provider logs."
+    )
+    provider = _cli_lead_provider(state)
+    if team_state_path is None:
+        return na_block(missing)
+    if provider == "antigravity":
+        return na_block(
+            "Antigravity host lead has no registered session transcript; "
+            "worker stream-json logs are collected only."
+        )
+    if provider not in _CLI_LEAD_PROVIDERS:
+        return na_block(missing)
+    run_since, run_until = resolve_run_window(team_state_path, state)
+    if not run_since or not run_until:
+        return na_block(
+            f"{provider} lead usage accounting is unavailable because the run window is missing."
+        )
+    worker_paths = {
+        Path(path)
+        for worker in state.get("workers") or []
+        if isinstance(worker, dict)
+        for path in ((worker.get("usage") or {}).get("cliSessionPaths") or [])
+    }
+    sessions = _select_cli_lead_sessions(
+        provider,
+        [
+            path
+            for path in _cli_sessions_for_windows(
+                provider, project_root, [(run_since, run_until)],
+            )
+            if path not in worker_paths
+        ],
+        state,
+    )
+    totals = _cli_session_totals(provider, sessions)
+    if not totals:
+        return na_block(
+            f"{provider} lead usage accounting is unavailable because no host session "
+            "started in the run window."
+        )
+    return _cli_usage_block(provider, _aggregate_totals(totals), sessions)
+
+
+def _cli_worker_session(provider: str, path: Path) -> bool:
+    if provider == "grok":
+        return grok_session_is_non_interactive(path)
+    if provider == "codex":
+        return codex_session_is_worker(path)
+    return False
+
+
+def _select_cli_lead_sessions(
+    provider: str, sessions: list[Path], state: dict,
+) -> list[Path]:
+    """워커 래퍼 세션을 빼고, 아이디가 있으면 그 세션, 없으면 벽시계가 가장 긴 대화."""
+    candidates = [
+        path for path in sessions if not _cli_worker_session(provider, path)
+    ]
+    if not candidates:
+        return []
+    pinned = _pinned_cli_lead_session(provider, candidates, state)
+    if pinned is not None:
+        return [pinned]
+    if len(candidates) == 1:
+        return candidates
+    best = candidates[0]
+    best_ms = -1
+    for path in candidates:
+        totals = _cli_session_totals(provider, [path])
+        total = totals[0] if totals else {}
+        wall = _wall_ms(total.get("startedAt"), total.get("endedAt")) if total else None
+        if wall is None:
+            wall = -1
+        if wall > best_ms:
+            best = path
+            best_ms = wall
+    return [best]
+
+
+def _pinned_cli_lead_session(
+    provider: str, candidates: list[Path], state: dict,
+) -> Path | None:
+    lead = state.get("lead") if isinstance(state.get("lead"), dict) else {}
+    wanted = {str(lead.get("sessionId") or "").strip()}
+    if provider == "grok":
+        wanted.add(str(os.environ.get("GROK_SESSION_ID") or "").strip())
+    wanted.discard("")
+    for path in candidates:
+        names = {path.parent.name, path.name, path.stem}
+        if provider == "codex":
+            names.update(codex_session_ids(path))
+        if wanted & names:
+            return path
+        if provider == "codex" and any(
+            session_id and session_id in path.name for session_id in wanted
+        ):
+            return path
+    return None
+
+
 def _populate_usage_summary(
     state: dict,
     *,
@@ -747,7 +880,7 @@ def _populate_usage_summary(
     lead_total = lead.get("totalTokens", 0) or 0
     lead_cache_read = lead.get("cacheReadTokens", 0) or 0
     lead_billable = lead.get("billableEquivalentTokens", 0) or 0
-    lead_cost = lead.get("estimatedCostUsd", 0) or 0
+    lead_cost = lead.get("estimatedCostUsd") or lead.get("cliEstimatedCostUsd") or 0
     worker_total = sum((w.get("usage") or {}).get("totalTokens", 0) or 0 for w in workers)
     worker_cache_read = sum((w.get("usage") or {}).get("cacheReadTokens", 0) or 0 for w in workers)
     worker_billable = sum((w.get("usage") or {}).get("billableEquivalentTokens", 0) or 0 for w in workers)
@@ -1063,7 +1196,7 @@ def collect_cli_runtime_usage(
 ) -> dict:
     state = json.loads(team_state_path.read_text())
     cwd = project_root or _infer_project_root(team_state_path, state)
-    return _collect_cli_runtime_usage(state, cwd)
+    return _collect_cli_runtime_usage(state, cwd, team_state_path)
 
 
 def collect(

package/runtime/python/okstra_token_usage/grok.py

@@ -6,6 +6,7 @@
 """
 from __future__ import annotations
 
+import json
 import os
 from datetime import datetime, timezone
 from pathlib import Path
@@ -87,11 +88,17 @@ def grok_session_total(updates_path: Path) -> dict:
     }
 
 
-def _session_in_window(updates_path: Path, started_at: str, ended_at: str) -> bool:
+def _session_started_in_window(
+    updates_path: Path, started_at: str, ended_at: str,
+) -> bool:
+    """첫 usage 시각이 창 안에 있을 때만 이 창의 세션이다.
+
+    창 중간의 갱신만 보면 같은 cwd 의 리드 대화가 워커 창에 섞인다.
+    """
     for record in iter_jsonl(updates_path):
         iso = _iso_from_unix(record.get("timestamp"))
-        if iso and ts_in_window(iso, started_at, ended_at):
-            return True
+        if iso:
+            return ts_in_window(iso, started_at, ended_at)
     return False
 
 
@@ -102,7 +109,7 @@ def find_grok_sessions(
     *,
     session_root: Path | None = None,
 ) -> list[Path]:
-    """cwd 로 인코딩된 세션 중 창 안에 갱신이 있는 updates.jsonl."""
+    """cwd 로 인코딩된 세션 중 창 안에서 시작된 updates.jsonl."""
     if not started_at or not ended_at:
         return []
     root = session_root or grok_sessions_root()
@@ -118,10 +125,22 @@ def find_grok_sessions(
             continue
         for session in child.iterdir():
             updates = session / "updates.jsonl"
-            if updates.is_file() and _session_in_window(updates, started_at, ended_at):
+            if updates.is_file() and _session_started_in_window(
+                updates, started_at, ended_at
+            ):
                 matches.append(updates)
     return sorted(matches)
 
 
 def grok_session_dir_name(cwd: Path) -> str:
     return quote(str(cwd), safe="")
+
+
+def grok_session_is_non_interactive(updates_path: Path) -> bool:
+    """워커 래퍼 세션은 prompt_context.is_non_interactive 가 true 이다."""
+    context_path = updates_path.parent / "prompt_context.json"
+    try:
+        payload = json.loads(context_path.read_text(encoding="utf-8"))
+    except (OSError, json.JSONDecodeError):
+        return False
+    return payload.get("is_non_interactive") is True

package/runtime/python/okstra_token_usage/report.py

@@ -18,6 +18,7 @@ from okstra_ctl.design_prep import (  # noqa: E402
     DesignPrepError,
     materialize_design_prep_requests,
 )
+from okstra_ctl.usage_cells import duration_ms_from_bounds  # noqa: E402
 
 from .task_totals import task_cumulative_usage  # noqa: E402
 
@@ -81,8 +82,11 @@ def _populate_execution_row(row: dict, source: dict) -> None:
     """
     usage = source.get("usage") or {}
     if usage.get("source") == "unavailable":
-        # Leave cells null; renderer emits `--`. A note in the row's
-        # summary is the worker's responsibility, not ours.
+        # Token cells stay null (`--`). Dispatch wall-clock still belongs
+        # on the row: a wrapper that failed still ran for some time.
+        duration = duration_ms_from_bounds(source.get("startedAt"), source.get("endedAt"))
+        if duration is not None:
+            row["durationMs"] = duration
         return
     if "totalTokens" in usage:
         row["totalTokens"] = usage["totalTokens"]
@@ -98,6 +102,12 @@ def _populate_execution_row(row: dict, source: dict) -> None:
         row["cliTotalTokens"] = usage["cliTotalTokens"]
     if "cliEstimatedCostUsd" in usage:
         row["cliCostUsd"] = usage["cliEstimatedCostUsd"]
+    if not row.get("durationMs"):
+        duration = duration_ms_from_bounds(source.get("startedAt"), source.get("endedAt"))
+        if duration is None:
+            duration = duration_ms_from_bounds(usage.get("startedAt"), usage.get("endedAt"))
+        if duration is not None:
+            row["durationMs"] = duration
 
 
 def _worker_detail_label(worker: dict) -> str:

package/runtime/skills/okstra-user-response/SKILL.md

@@ -59,7 +59,9 @@ Read the absolute path in the fixed `Relay contract` line. In that file, take th
 - When `native-single` is available and the option count fits `nativeLimits` (unique labels, within min/max): call `interactions.native-single.function` once with one question and every option as `{label, description}` in original order. Do not print a numbered list in chat while the native tool is available. Claude Code's function is `AskUserQuestion`, Grok's is `ask_user_question`, Codex's is `request_user_input` — copy the relay field; do not substitute one name for another.
 - Otherwise render a 1-based numbered Markdown list and wait for the next message. Do not drop options to force the native tool.
 
-Pass only the choices this step already owns — the `list-view` rows, the report `options[]`, or the two confirmation labels. Do not append `Enter directly`. Claude Other, Grok `z`, and Codex's free-form row already collect a custom answer; that row is `Enters an answer`. When native-single is unavailable, a next message that is not a listed label or its 1-based number is the same `Enters an answer`. Do not ask a second question for the custom value.
+Copy the view's `Picker:` `- Label:` / `Description:` pairs into that function in that order. Do not rebuild labels from the `Options:` dump. `--option-number` is the 1-based `Option N:` index, which is the same order as `Picker:`. The HTML report's `<select>` uses the same `option.answer` values.
+
+Pass only the choices this step already owns — the `list-view` `Picker:` rows, the report `Picker:` rows, or the two confirmation labels. Do not append `Enter directly`. Claude Other, Grok `z`, and Codex's free-form row already collect a custom answer; that row is `Enters an answer`. When native-single is unavailable, a next message that is not a listed label or its 1-based number is the same `Enters an answer`. Do not ask a second question for the custom value.
 
 Never invent a picker function. Never ask the user to type a number when the native tool is available.
 
@@ -79,7 +81,7 @@ Present up to three task choices through the host picker. A host free-text row o
 okstra user-response show-view --report <reportPath> --project-root <projectRoot>
 ```
 
-The view contains the report identity, contract version, every open clarification question, its expected form, its current response and disposition, its options, approval context, plan option candidates, current plan decision, resolved context, why the row is asked, linked plan items, and cited artifacts. Question text and `options[]` come only from this view. Do not open a report record to select fields.
+The view contains the report identity, contract version, every open clarification question, its expected form, its current response and disposition, its options, a `Picker:` block, approval context, plan option candidates, current plan decision, resolved context, why the row is asked, linked plan items, and cited artifacts. Question text and `options[]` come only from this view. Do not open a report record to select fields.
 
 Each entry in `options[]` corresponds to `{role, answer, rationale, scopeImpact, addedWork, directionChange, disposition}`. Put the `recommended` option first and suffix its label with `(Recommended)`. Then put the alternatives in view order.
 

package/runtime/templates/reports/html/assets/base.css

@@ -135,9 +135,7 @@ button[data-action="export-user-response"]:hover { background: color-mix(in srgb
   align-items: flex-end;
   gap: .4rem;
 }
-.back-to-top-actions { display: flex; gap: .4rem; }
-.back-to-top,
-.back-to-top-toggle {
+.back-to-top {
   padding: .55rem .9rem;
   border-radius: 8px;
   border: 1px solid color-mix(in srgb, CanvasText 40%, transparent);
@@ -148,12 +146,9 @@ button[data-action="export-user-response"]:hover { background: color-mix(in srgb
   font-size: .9rem;
   font-weight: 600;
   cursor: pointer;
-  box-shadow: 0 2px 10px color-mix(in srgb, CanvasText 28%, transparent);
 }
-.back-to-top:hover,
-.back-to-top-toggle:hover { background: color-mix(in srgb, CanvasText 12%, Canvas); }
-.back-to-top:focus-visible,
-.back-to-top-toggle:focus-visible { outline: 2px solid Highlight; outline-offset: 2px; }
+.back-to-top:hover { background: color-mix(in srgb, CanvasText 12%, Canvas); }
+.back-to-top:focus-visible { outline: 2px solid Highlight; outline-offset: 2px; }
 .back-to-top-index {
   display: none;
   box-sizing: border-box;
@@ -170,7 +165,6 @@ button[data-action="export-user-response"]:hover { background: color-mix(in srgb
 .back-to-top-index ol { margin: 0; padding-left: 1.2rem; }
 .back-to-top-index li { margin: .25em 0; }
 .back-to-top-index a { color: inherit; }
-.back-to-top-wrap.is-open .back-to-top-index { display: block; }
 @media (hover: hover) {
   .back-to-top-wrap:hover .back-to-top-index { display: block; }
 }

package/runtime/templates/reports/html/assets/base.js

@@ -2,25 +2,4 @@
   "use strict";
 
   document.documentElement.classList.add("js-enabled");
-
-  // 목차는 hover 만으로는 안 열린다. 클릭으로 연다.
-  var wrap = document.querySelector(".back-to-top-wrap");
-  var toggle = wrap && wrap.querySelector(".back-to-top-toggle");
-  if (!wrap || !toggle) return;
-
-  function setOpen(open) {
-    wrap.classList.toggle("is-open", open);
-    toggle.setAttribute("aria-expanded", open ? "true" : "false");
-  }
-
-  toggle.addEventListener("click", function (event) {
-    event.stopPropagation();
-    setOpen(!wrap.classList.contains("is-open"));
-  });
-  document.addEventListener("click", function (event) {
-    if (!wrap.contains(event.target)) setOpen(false);
-  });
-  document.addEventListener("keydown", function (event) {
-    if (event.key === "Escape") setOpen(false);
-  });
 })();

package/runtime/templates/reports/html/base.template.html

@@ -116,10 +116,7 @@
   </footer>{% endif %}
   <div class="back-to-top-wrap">
     <nav class="back-to-top-index" id="back-to-top-index" aria-label="{{ t('base.contents') }}"><!--report-index-items--></nav>
-    <div class="back-to-top-actions">
-      <button type="button" class="back-to-top-toggle" aria-expanded="false" aria-controls="back-to-top-index">{{ t('base.contents') }}</button>
-      <a class="back-to-top" href="#top">{{ t('base.back-to-top') }}</a>
-    </div>
+    <a class="back-to-top" href="#top">{{ t('base.back-to-top') }}</a>
   </div>
   <script id="run-meta" type="application/json">{{ {
     "task-key": runMeta.task_key,

package/runtime/templates/reports/html/tasks/implementation-planning.template.html

@@ -85,12 +85,14 @@
 {% endif %}
 
 {# The verification checklist, cross-project dependencies, migration risk,
-   rollback strategy and plan-body verification rounds are the implementer's
-   and the auditor's working material, not the approver's: this reader is
-   deciding whether to greenlight the plan, and none of those tables move that
-   decision. They stay in the markdown report, which is what the next phase and
-   the audit trail read. Requirement coverage is the exception — "does this plan
-   actually do what the brief asked" is the approval question itself. #}
+   rollback strategy, plan-body verification rounds, and the agent-activity
+   ledger are the implementer's and the auditor's working material, not the
+   approver's: this reader is deciding whether to greenlight the plan, and none
+   of those tables move that decision. They stay in the markdown report, which
+   is what the next phase and the audit trail read. Requirement coverage is the
+   exception — "does this plan actually do what the brief asked" is the
+   approval question itself. A-NNN citations still need an id in this document
+   or the checkRef links land nowhere. #}
 <section data-report-section="requirement-coverage" data-report-field="implementationPlanning.requirementCoverage">
   <h2>{{ t('tasks.implementation-planning.how-each-requirement-gets-met') }}</h2>
   {% if planning.get("planningContract") == "selected-direction" %}
@@ -134,28 +136,7 @@
 {% endif %}
 
 {% if agentActivities %}
-<section data-report-section="agent-activity">
-  <h2>{{ t('tasks.implementation-planning.what-each-agent-did') }}</h2>
-  <div class="activity-list">
-    {% for row in agentActivities %}
-    <article class="activity-card" id="id-{{ row.activityId }}">
-      <p class="eyebrow">{{ row.activityId | inline_code }} · {{ row.agent }}</p>
-      <h3>{{ row.summary | inline_code }}</h3>
-      <p>{{ t('tasks.implementation-planning.result') }}: {{ row.outcome | inline_code }}</p>
-      <details>
-        <summary>{{ t('tasks.implementation-planning.commands-and-evidence') }}</summary>
-        <p><code>{{ row.kind }}</code></p>
-        {% if row.planItemIds %}<p>{{ row.planItemIds | join(', ') | inline_code }}</p>{% endif %}
-        {% for command in row.commands %}
-        <p><code>{{ command.command }}</code> · exit {{ command.exitCode }} · {{ command.outputSummary | inline_code }}</p>
-        {% endfor %}
-        {% if row.evidenceRefs %}<p>{{ row.evidenceRefs | join(', ') | inline_code }}</p>{% endif %}
-        {% if row.resultPath %}<p><code>{{ row.resultPath }}</code></p>{% endif %}
-      </details>
-    </article>
-    {% endfor %}
-  </div>
-</section>
+<div hidden>{% for row in agentActivities %}<span id="id-{{ row.activityId }}"></span>{% endfor %}</div>
 {% endif %}
 
 <section data-report-section="open-decisions">

package/runtime/validators/lib/runners.sh

@@ -102,8 +102,12 @@ if not process.stdout.strip():
 payload = json.loads(process.stdout)
 actual_status = payload.get("validationStatus")
 if actual_status != expected_status:
+    extra = ""
+    if expected_status == "passed":
+        listed = payload.get("failures") or []
+        extra = "\n" + "\n".join(str(item) for item in listed)
     raise SystemExit(
-        f"validator status mismatch: expected {expected_status}, got {actual_status}"
+        f"validator status mismatch: expected {expected_status}, got {actual_status}{extra}"
     )
 
 if expected_status == "passed" and process.returncode != 0:

package/runtime/validators/validate-report-views.py

@@ -40,6 +40,7 @@ for _ssot_dir in (_VALIDATORS_DIR.parent / "scripts", _VALIDATORS_DIR.parent / "
         sys.path.insert(0, str(_ssot_dir))
 
 from okstra_ctl.clarification_items import (  # noqa: E402
+    STRUCTURED_REPORT_VERSIONS,
     parse_clarification_items,
     section_1_present_but_unparsed,
 )
@@ -87,7 +88,7 @@ def _load_v2_data(report_path: Path) -> tuple[Path, dict] | None:
         data = json.loads(data_path.read_text(encoding="utf-8"))
     except (OSError, json.JSONDecodeError):
         return None
-    if isinstance(data, dict) and data.get("schemaVersion") == "2.0":
+    if isinstance(data, dict) and data.get("schemaVersion") in STRUCTURED_REPORT_VERSIONS:
         return data_path, data
     return None