okstra 0.86.0 → 0.87.1

package/runtime/python/okstra_token_usage/paths.py

@@ -8,7 +8,6 @@ from pathlib import Path
 HOME = Path.home()
 CLAUDE_PROJECTS = HOME / ".claude" / "projects"
 CODEX_SESSIONS = HOME / ".codex" / "sessions"
-GEMINI_TMP = HOME / ".gemini" / "tmp"
 
 
 def utc_now() -> str:

package/runtime/python/okstra_token_usage/pricing.py

@@ -8,7 +8,8 @@ provider:
     `claude-haiku-4-5-*`, `claude-3-5-sonnet-*`, `claude-3-5-haiku-*`,
     `claude-3-opus-*`, `claude-3-haiku-*`.
   * OpenAI / Codex — `gpt-5*`, `gpt-4o*`, `gpt-4*`.
-  * Google / Gemini — `gemini-2.5-pro*`, `gemini-2.5-flash*`, `gemini-2.0-flash*`.
+  * Google / Gemini — `gemini-3.5-flash*`, `gemini-3.1-pro*`, `gemini-3-flash*`,
+    `gemini-2.5-pro*`, `gemini-2.5-flash*`, `gemini-2.0-flash*`.
 
 Insertion order is the match order, so list more specific keys first. Update
 when providers change list pricing.
@@ -112,7 +113,9 @@ GEMINI_PRICING = {
     # Both dotted (`gemini-3.1-pro`) and hyphenated (`gemini-3-1-pro`) id
     # forms appear in the wild, so include both for the new 3.x families.
 
-    # Gemini 3 series (preview).
+    # Gemini 3 series.
+    "3.5-flash":       (1.50, 9.00),
+    "3-5-flash":       (1.50, 9.00),
     "3.1-pro":         (2.00, 12.0),
     "3-1-pro":         (2.00, 12.0),
     "3-flash":         (0.50, 3.00),
@@ -203,7 +206,7 @@ def codex_cost_usd(model: str | None, input_t: int, cached_input_t: int, output_
     return round((fresh * pi + cached_input_t * pci + output_t * po) / 1_000_000, 4)
 
 
-def gemini_cost_usd(model: str | None, input_t: int, output_t: int) -> float | None:
+def antigravity_cost_usd(model: str | None, input_t: int, output_t: int) -> float | None:
     p = _match_pricing(model, GEMINI_PRICING)
     if p is None:
         return None

package/runtime/schemas/final-report-v1.0.schema.json

@@ -884,7 +884,7 @@
       "required": ["agent", "role", "model", "status", "summary"],
       "additionalProperties": false,
       "properties": {
-        "agent": { "enum": ["Claude Code", "Codex", "Gemini"] },
+        "agent": { "enum": ["Claude Code", "Codex", "Antigravity"] },
         "role": { "type": "string", "minLength": 1 },
         "model": { "type": "string", "minLength": 1 },
         "status": { "$ref": "#/$defs/WorkerStatus" },

package/runtime/skills/okstra-brief/SKILL.md

@@ -462,7 +462,7 @@ on a project-internal concrete object whenever possible.
   `<PROJECT_ROOT>`) and the canonical symbol name. Record the resolution
   under `Augmentation > Domain alignment` with the `evidence-link` label,
   and link the absolute path from `Related Artifacts`. This is what makes
-  the brief portable across downstream workers (Claude / Codex / Gemini)
+  the brief portable across downstream workers (Claude / Codex / Antigravity)
   that may operate in separate worktrees and need to point at the same
   object.
 - **Conversion-block signal** — when a reporter statement cannot be

package/runtime/skills/okstra-inspect/SKILL.md

@@ -1,12 +1,12 @@
 ---
 name: okstra-inspect
 description: |
-  Use for any read-side okstra inspection or status mutation. Single skill dispatches by sub-command to six facets — status, history, report, time, logs, cost. Trigger words include "okstra status", "task status", "current phase", "next phase", "okstra status set", "okstra mark", "<task-id> done|in-progress|진행중|완료", "okstra history", "past runs", "re-run", "resume", "list tasks", "find report", "show report for", "read the okstra report", "continue from report", "작업 시간", "소요 시간", "time summary", "duration", "elapsed", "얼마나 걸렸", "시간 분석", "okstra logs", "로그 현황", "로그 파일", "log size", "log status", "로그 정리", "log cleanup", "okstra context-cost", "context cost", "context-cost", "컨텍스트 비용", "읽기 비용", "산출물 비용".
+  Use for any read-side okstra inspection or status mutation. Single skill dispatches by sub-command to seven facets — status, history, report, time, logs, cost, errors. Trigger words include "okstra status", "task status", "current phase", "next phase", "okstra status set", "okstra mark", "<task-id> done|in-progress|진행중|완료", "okstra history", "past runs", "re-run", "resume", "list tasks", "find report", "show report for", "read the okstra report", "continue from report", "작업 시간", "소요 시간", "time summary", "duration", "elapsed", "얼마나 걸렸", "시간 분석", "okstra logs", "로그 현황", "로그 파일", "log size", "log status", "로그 정리", "log cleanup", "okstra context-cost", "context cost", "context-cost", "컨텍스트 비용", "읽기 비용", "산출물 비용", "okstra errors", "error report", "에러 리포트", "에러 보고", "에러 모아줘", "실패 로그 정리".
 ---
 
 # OKSTRA Inspect
 
-Single read-side entry point for okstra runtime inspection plus the one write-side mutation that belongs here (`workStatus`). Sub-commands:
+Single read-side entry point for okstra runtime inspection plus the one status mutation that belongs here (`workStatus`) and read-derived artifact rendering (`errors` report). Sub-commands:
 
 | Sub-command | What it does |
 |---|---|
@@ -14,8 +14,9 @@ Single read-side entry point for okstra runtime inspection plus the one write-si
 | `history` | List past okstra runs; assemble re-run or resume command. |
 | `report` | Resolve final-report path for a task-key. Optionally read it. |
 | `time` | Per-task-type and per-worker duration breakdown for a task. |
-| `logs` | Inventory codex/gemini wrapper `.log` sidecars; emit cleanup commands. |
+| `logs` | Inventory codex/antigravity wrapper `.log` sidecars; emit cleanup commands. |
 | `cost` | Estimate file/read context cost for a task bundle. |
+| `errors` | Aggregate okstra-run error logs for a task into a timestamped markdown report; print a summary. |
 
 ## Step 0: Verify okstra runtime + project setup (shared)
 
@@ -245,7 +246,7 @@ When a user selects a specific task:
 
 Builds a fresh run — new run-seq, new manifest, new report — using parameters from a previous `run-manifest-*.json`. Does NOT touch old artifacts; use `history.4` to continue an interrupted session.
 
-1. Pick the source run-manifest: `runManifestPath` from `history.2`, or task's `latestRunManifestPath` from `history.1`.
+1. Pick the source run-manifest: `runManifestPath` from `history.2`, or task's `latestRunManifestPath` from `history.1`. Note: `latestRunManifestPath` is projected from the timeline's latest run, so it is empty when the task has no recorded run yet (timeline absent) — in that case fall back to a `history.2` per-run `runManifestPath`, or ask the user.
 2. Read the manifest and extract required arguments:
    - `projectId` → `--project-id`
    - `taskGroup` → `--task-group`
@@ -255,8 +256,8 @@ Builds a fresh run — new run-seq, new manifest, new report — using parameter
 3. Optional arguments (include only when present in source):
    - `recommendedWorkers` → `--workers` (comma-separated)
    - `relatedTasks` → `--related-tasks`
-   - model overrides → `--claude-model`, `--codex-model`, `--gemini-model`
-   - for `taskType: implementation`: `teamContract.executor.provider` → `--executor <claude|codex|gemini>` when different from `claude`.
+   - model overrides → `--claude-model`, `--codex-model`, `--antigravity-model`
+   - for `taskType: implementation`: `teamContract.executor.provider` → `--executor <claude|codex|antigravity>` when different from `claude`.
 4. **`taskType: implementation` only — resolve `--base-ref`:** base ref is NOT in the run-manifest; it lives in the worktree registry at `~/.okstra/worktrees/registry.json`. If a worktree is already registered, existing branch & base are reused — omit `--base-ref` unless the user explicitly wants a different starting point. If no worktree is registered (cleaned up), `--base-ref` is mandatory — ask the user before running.
 5. Display the assembled command:
    ```bash
@@ -299,7 +300,7 @@ B. **`task-manifest.json` (direct):** if catalog missing, slugify task-group / t
 
 C. **`timeline.json` (specific run):** for a specific date or run, read `.okstra/tasks/<group-segment>/<id-segment>/history/timeline.json`, filter `runs[]` by `runTimestamp` / `status` / `taskType`, use `runs[].reportPath`.
 
-D. **Specific task-type (fallback):** `latestReportPath` is task-type-agnostic. For a specific task-type's latest report, look under `.okstra/tasks/<group-segment>/<id-segment>/runs/<task-type-segment>/reports/`, filename pattern `final-report-<task-type-segment>-<NNN>.md` per `scripts/okstra_ctl/sequence.py:31`. Highest seq is latest. Stage-isolated runs (`implementation`, single-stage `final-verification`) keep their reports one level deeper at `runs/<task-type-segment>/stage-<N>/reports/` — scan those subdirectories too; seq is independent per stage. Cross-verify with `timeline.json`'s `runs[].taskType` filter.
+D. **Specific task-type (fallback):** `latestReportPath` is task-type-agnostic. For a specific task-type's latest report, look under `.okstra/tasks/<group-segment>/<id-segment>/runs/<task-type-segment>/reports/`, filename pattern `final-report-<task-type-segment>-<NNN>.md` per `scripts/okstra_ctl/sequence.py:30`. Highest seq is latest. Stage-isolated runs (`implementation`, single-stage `final-verification`) keep their reports one level deeper at `runs/<task-type-segment>/stage-<N>/reports/` — scan those subdirectories too; seq is independent per stage. Cross-verify with `timeline.json`'s `runs[].taskType` filter.
 
 ### report.2 — Confirm existence
 
@@ -344,11 +345,11 @@ After reading the report, surface follow-up options:
 
 Trigger phrases: "작업 시간", "소요 시간", "time summary", "duration", "elapsed", "얼마나 걸렸", "시간 분석".
 
-Aggregate elapsed work time for a given task, grouped by **task type** and broken down by **worker** (lead, claude, codex, gemini, report-writer).
+Aggregate elapsed work time for a given task, grouped by **task type** and broken down by **worker** (lead, claude, codex, antigravity, report-writer).
 
 **Data sources** (both collected by okstra):
 
-1. `.okstra/tasks/<task-group>/<task-id>/history/timeline.json` — `runs` array with `runTimestamp`, `taskType`, `status`, `teamStatePath`, `taskRootPath`. Both path fields may be either project-root-relative or task-root-relative depending on which `render.py` version wrote the manifest.
+1. `.okstra/tasks/<task-group>/<task-id>/history/timeline.json` — `runs` array with `runTimestamp`, `taskType`, `status`, `teamStatePath`. `teamStatePath` is project-root-relative (written by `render.py` via `_rel(project_root, ...)`).
 2. Each run's `.../runs/<task-type>/state/team-state-<suffix>.json` — populated by `scripts/okstra-token-usage.py` at Phase 7. Contains `leadUsage.{startedAt, endedAt, durationMs}` and `workers[].{workerId, agent, usage.{startedAt, endedAt, durationMs}}`.
 
 If a run never reached Phase 7, its `team-state` lacks `durationMs`. Mark such runs as `unavailable` rather than guessing.
@@ -366,10 +367,7 @@ If `task-catalog.json` is missing: `No okstra history found. Run /okstra-run fir
 
 For each entry in `timeline.json`'s `runs` array:
 
-1. Resolve the `team-state` file via two-step lookup:
-   a. First try `<projectRoot>/<teamStatePath>`.
-   b. If that file does not exist, fall back to `<projectRoot>/<taskRootPath>/<teamStatePath>` (the manifest's `taskRootPath` is task-root relative to project root; `teamStatePath` written by `render.py` is task-root-relative in many runs).
-   If neither resolves, treat the run as `unavailable`.
+1. Resolve the `team-state` file at `<projectRoot>/<teamStatePath>` — `teamStatePath` is project-root-relative. If `teamStatePath` is empty or the file does not exist on disk, treat the run as `unavailable`.
 2. Extract `taskType` from the timeline entry (authoritative), `leadUsage.durationMs` and `leadUsage.{startedAt,endedAt}`, and for each `worker` in `workers[]`: `workerId`, `agent`, `usage.durationMs`. Read defensively — `usage` (and `leadUsage`) may be:
    - a normal `usage_block` with `durationMs >= 0`,
    - a `na_block` with `{"source": "unavailable", "durationMs": 0, "note": ...}` when Phase 7 collection failed,
@@ -398,7 +396,7 @@ Worked example for one run with three concurrent workers:
 lead       [================================]  durationMs = 1800000 (30:00)
 claude       [============]                     durationMs =  720000 (12:00)
 codex        [==============]                   durationMs =  840000 (14:00)
-gemini       [========]                         durationMs =  480000 (08:00)
+antigravity       [========]                         durationMs =  480000 (08:00)
 ```
 
 - `CPU sum` for the run = `1800000 + 720000 + 840000 + 480000` = `3840000` (`01:04:00`)
@@ -417,7 +415,7 @@ For each task type, list one row per `workerId` actually present, plus `lead`. A
 - If `Runs == 0` for a worker, **omit the row entirely** rather than dividing by zero.
 - `Avg/run = Total / Runs` (integer ms, then format to `HH:MM:SS`).
 
-Use the `workerId` from team-state. Valid enum: `lead, claude, codex, gemini, report-writer`.
+Use the `workerId` from team-state. Valid enum: `lead, claude, codex, antigravity, report-writer`.
 
 Display rule for `workerId` vs `agent`:
 - If every run of this task type used `agent == workerId` for this row, display the bare `workerId`.
@@ -463,7 +461,7 @@ team-state written by collector ≥ v0.73 carries a `phaseTimeline` block — th
 | lead           | 2    | 00:12:08 | 00:06:04 |
 | claude         | 2    | 00:09:12 | 00:04:36 |
 | codex          | 2    | 00:07:40 | 00:03:50 |
-| gemini         | 2    | 00:03:12 | 00:01:36 |
+| antigravity         | 2    | 00:03:12 | 00:01:36 |
 | report-writer  | 2    | 00:01:00 | 00:00:30 |
 
 > Unavailable: 1 run (implementation / 2026-04-30_03-03-48) — team-state has no durationMs (Phase 7 not reached)
@@ -595,14 +593,14 @@ Interpretation rules:
 
 Trigger phrases: "okstra logs", "로그 현황", "로그 파일", "log files", "log size", "log status", "로그 정리", "log cleanup".
 
-Read-only inventory of codex/gemini wrapper log files written next to each prompt history file (`<prompt>.log`). Reports sizes, ages, totals, and suggests cleanup commands. **Does not delete** — the user runs whichever `find … -delete` line they like.
+Read-only inventory of codex/antigravity wrapper log files written next to each prompt history file (`<prompt>.log`). Reports sizes, ages, totals, and suggests cleanup commands. **Does not delete** — the user runs whichever `find … -delete` line they like.
 
-**Background:** codex/gemini wrappers (`okstra-codex-exec.sh`, `okstra-gemini-exec.sh`) write a sidecar log next to each prompt history file:
+**Background:** codex/antigravity wrappers (`okstra-codex-exec.sh`, `okstra-antigravity-exec.sh`) write a sidecar log next to each prompt history file:
 
 ```
 .okstra/tasks/<task-group>/<task-id>/runs/<phase>/prompts/
-  <worker>-prompt-<phase>-<seq>.md    <-- prompt (git-tracked)
-  <worker>-prompt-<phase>-<seq>.log   <-- live stdout+stderr mirror
+  <worker>-worker-prompt-<phase>-<seq>.md    <-- prompt (git-tracked)
+  <worker>-worker-prompt-<phase>-<seq>.log   <-- live stdout+stderr mirror
 ```
 
 The log is truncated at each dispatch (`: > "$log_path"`) — only the latest run for a given seq is preserved. Different seqs keep separate files. Long-running implementation dispatches can produce multi-MB logs; analysis-phase dispatches are typically smaller.
@@ -634,7 +632,7 @@ If empty, report `No wrapper log files found under <projectRoot>` and exit.
 | # | Task | Phase | Worker | Seq | Size | Age | Path |
 |---|------|-------|--------|-----|------|-----|------|
 
-Parse fields from the path: task-group / task-id from the `tasks/<task-group>/<task-id>/` segment; phase from `runs/<phase>/`; worker from filename prefix before `-prompt-`; seq from filename suffix (last 3-digit segment). Format sizes as KB/MB. Format age as `Nd` (days) or `Nh` (hours) from `mtime` relative to "now".
+Parse fields from the path: task-group / task-id from the `tasks/<task-group>/<task-id>/` segment; phase from `runs/<phase>/`; worker from filename prefix before `-worker-prompt-` (the wrapper names files `<worker>-worker-prompt-<phase>-<seq>.log`); seq from filename suffix (last 3-digit segment). Format sizes as KB/MB. Format age as `Nd` (days) or `Nh` (hours) from `mtime` relative to "now".
 
 **Table B — Per-task totals:**
 
@@ -694,6 +692,72 @@ Substitute the literal `<PROJECT_ROOT>` with the resolved absolute path so the c
 
 ---
 
+## errors
+
+Trigger phrases: "okstra errors", "error report", "에러 리포트", "에러 보고", "에러 모아줘", "실패 로그 정리".
+
+Aggregate a task's okstra-run error logs (`runs/*/logs/errors-*.jsonl`, lead-observed + worker-reported) into a timestamped markdown report and summarize it. This sub-command renders a **read-derived artifact** (a `.md` file) but never mutates task state (`task-manifest.json`, catalog, timeline).
+
+### errors.1 — Resolve target
+
+Accepted target forms (same as `cost`):
+
+1. Full task-key: `<project-id>:<task-group>:<task-id>`.
+2. Task id only — read `.okstra/discovery/task-catalog.json`, match `taskId` case-insensitively (multiple → list and ask; none → report not found).
+3. Task root path.
+
+### errors.2 — Run the renderer
+
+Use the CLI output as the source of truth:
+
+```bash
+okstra error-report <resolved-target> --project-root <projectRoot>
+```
+
+For a task-root path, run `okstra error-report <path>` directly. Do not parse the jsonl files by hand unless the CLI fails and the user explicitly asks for a manual fallback.
+
+### errors.3 — Summarize output
+
+Parse the stdout JSON and report:
+
+| Field | Source |
+|---|---|
+| Report file | `reportPath` (project-root-relative `.md`) |
+| Total errors | `totals.errorCount` |
+| Logs (runs) | `totals.runCount` |
+| By errorType | `totals.byErrorType` (tool-failure / cli-failure / contract-violation) |
+| By source | `totals.bySource` (lead-observed / worker-reported) |
+| By phase | `byPhase[]` |
+| By agent | `byAgent[]` |
+| Parse-skipped lines | `parseSkipped` |
+
+- If `reportPath` is empty AND `totals.errorCount == 0`: report `이 task 에는 기록된 에러 로그가 없습니다.` and do not claim a file was written.
+- Otherwise show the `.md` path and offer to read it.
+- If `parseSkipped > 0`, surface it (do not silently hide malformed lines).
+
+### errors — Output template
+
+```markdown
+## okstra Error Report — <task-key>
+
+- Report: `<reportPath-or-->`
+- Total errors: <N> across <runCount> log(s)
+- By type: <tool-failure: a, cli-failure: b, ...>
+- By source: <lead-observed: x, worker-reported: y>
+
+| Phase | Count |
+|---|---:|
+| implementation | 3 |
+
+| Agent | Count |
+|---|---:|
+| codex-worker | 2 |
+
+<If parseSkipped > 0: "⚠ 파싱 건너뛴 줄: <N>">
+```
+
+---
+
 ## Output Rules (shared)
 
 - Responses should be concise and written in Korean unless the user requests otherwise.

package/runtime/skills/okstra-run/SKILL.md

@@ -49,7 +49,7 @@ The wizard tells you *which UI to use* via `kind` (and the optional `multi` flag
 
 The `branch_confirm` step (shown just before `confirm`) is a normal `pick` step and is rendered the same way — no special handling needed. Its options always include `중단` (abort); `base-ref 다시 고르기` (edit) appears only when a new branch would be created from the user-chosen base-ref. For `implementation` runs the wizard previews the **stage worktree** this run will actually use (not the task-key directory); render its label verbatim.
 
-Never invent additional questions. Never reorder. **Never drop, hide, or merge a `pick` / `pick_group` option** — render every `options[]` entry as its own selectable `AskUserQuestion` choice, including entries that carry a `(default)` / `(recommended)` suffix. Do NOT collapse a multi-option pick into a "recommended + 직접 입력 / Other" shortlist: the wizard's `options[]` array IS the complete, authoritative choice set. Example: the `executor` step always emits `claude` / `codex` / `gemini` — show all three, never just `claude`. The run-prompt recommendation rule (1–2 추천 + 직접 입력) applies ONLY to prompts this skill authors itself (e.g. the conformance-waiver picker), never to wizard-provided `options[]`. Never use `AskUserQuestion` for `text` prompts — the wizard explicitly chose `text` to avoid the picker-Other re-render lag.
+Never invent additional questions. Never reorder. **Never drop, hide, or merge a `pick` / `pick_group` option** — render every `options[]` entry as its own selectable `AskUserQuestion` choice, including entries that carry a `(default)` / `(recommended)` suffix. Do NOT collapse a multi-option pick into a "recommended + 직접 입력 / Other" shortlist: the wizard's `options[]` array IS the complete, authoritative choice set. Example: the `executor` step always emits `claude` / `codex` / `antigravity` — show all three, never just `claude`. The run-prompt recommendation rule (1–2 추천 + 직접 입력) applies ONLY to prompts this skill authors itself (e.g. the conformance-waiver picker), never to wizard-provided `options[]`. Never use `AskUserQuestion` for `text` prompts — the wizard explicitly chose `text` to avoid the picker-Other re-render lag.
 
 ## Step 1: Verify okstra runtime + project setup
 
@@ -95,7 +95,7 @@ Repeat until `next.kind == "done"` (or `"aborted"` — terminal cancel, see "How
 
 1. **Render** the prompt according to `kind` (and `multi` for pick):
    - `pick` + `multi: false` → `AskUserQuestion` with `multiSelect: false`, `label`, and `options`. The user's chosen option's `value` is the answer string.
-   - `pick` + `multi: true` → `AskUserQuestion` with `multiSelect: true`, `label`, and `options`. Join the selected `value`s with `,` into a single literal CSV string (e.g. `"claude,codex,gemini"`) and submit it as a single `--answer "claude,codex,gemini"`. Empty selection submits `--answer ""` and the wizard re-prompts.
+   - `pick` + `multi: true` → `AskUserQuestion` with `multiSelect: true`, `label`, and `options`. Join the selected `value`s with `,` into a single literal CSV string (e.g. `"claude,codex,antigravity"`) and submit it as a single `--answer "claude,codex,antigravity"`. Empty selection submits `--answer ""` and the wizard re-prompts.
    - `pick_group` → one `AskUserQuestion` with one question per `questions[]` entry (tab). Map each tab's selected `value` back by `questions[].step`, assemble a JSON object, and submit it as a single literal `--answer '<json>'`.
    - `text` → plain text message containing `label`. Consume the user's next reply verbatim as the answer string (empty reply = empty string).
 2. **Submit** the answer — call `okstra wizard step` with the literal state-file path from Step 2 and the literal user answer (no shell variables, no `$(...)`):
@@ -175,7 +175,7 @@ okstra render-bundle \
   --lead-model   "<args.lead-model>" \
   --claude-model "<args.claude-model>" \
   --codex-model  "<args.codex-model>" \
-  --gemini-model "<args.gemini-model>" \
+  --antigravity-model "<args.antigravity-model>" \
   --report-writer-model "<args.report-writer-model>" \
   --related-tasks "<args.related-tasks>" \
   --clarification-response "<args.clarification-response>" \

package/runtime/skills/okstra-setup/SKILL.md

@@ -222,10 +222,10 @@ field is optional; declare only what your conformance scripts use.
 provisions `<PROJECT_ROOT>/.claude/settings.local.json` as a symlink to
 `~/.okstra/templates/settings.local.json`. The template is installed
 by `okstra install` 0.14.0+ and contains the Bash permission rules
-required for the codex/gemini worker wrappers:
+required for the codex/antigravity worker wrappers:
 
 - `Bash($HOME/.okstra/bin/okstra-codex-exec.sh:*)`
-- `Bash($HOME/.okstra/bin/okstra-gemini-exec.sh:*)`
+- `Bash($HOME/.okstra/bin/okstra-antigravity-exec.sh:*)`
 
 Claude Code automatically loads `.claude/settings.local.json` whenever
 it operates inside that project, so okstra workers dispatched from

package/runtime/templates/prd/brief.template.md

@@ -14,7 +14,7 @@ taskType: "{{FM_TASK_TYPE}}"
 # OKSTRA Task Brief
 
 <!--
-이 brief은 okstra 워커(Claude/Codex/Gemini)가 코드베이스/도메인 사전지식 없이 분석할 수 있도록 준비하는 단일 source-of-truth 문서입니다.
+이 brief은 okstra 워커(Claude/Codex/Antigravity)가 코드베이스/도메인 사전지식 없이 분석할 수 있도록 준비하는 단일 source-of-truth 문서입니다.
 
 원칙:
 1. 워커는 외부 링크에 접근할 수 없음 → 모든 1차 증거는 inline으로 박을 것 (fenced code block).
@@ -86,7 +86,7 @@ taskType: "{{FM_TASK_TYPE}}"
 ## Domain Glossary
 
 <!--
-Codex/Gemini는 이 코드베이스를 모릅니다. 핵심 용어 5-15개를 한 문장씩 정의.
+Codex/Antigravity는 이 코드베이스를 모릅니다. 핵심 용어 5-15개를 한 문장씩 정의.
 없는 용어를 워커가 추측하면 분석 품질이 즉시 떨어집니다.
 -->
 

package/runtime/templates/reports/i18n/en.json

@@ -53,7 +53,7 @@
     "rowLead": "Lead",
     "rowWorkerTotal": "Worker subtotal",
     "rowGrandTotal": "**Grand total**",
-    "rowCliExtra": "Codex/Gemini CLI add-on"
+    "rowCliExtra": "Codex/Antigravity CLI add-on"
   },
   "verdictCard": {
     "tableHeaderLabel": "Item",

package/runtime/templates/reports/i18n/ko.json

@@ -53,7 +53,7 @@
     "rowLead": "Lead",
     "rowWorkerTotal": "Worker 합계",
     "rowGrandTotal": "**전체 합계**",
-    "rowCliExtra": "Codex/Gemini CLI 추가 비용"
+    "rowCliExtra": "Codex/Antigravity CLI 추가 비용"
   },
   "verdictCard": {
     "tableHeaderLabel": "항목",

package/runtime/templates/reports/report.js

@@ -3,10 +3,9 @@
  * Responsibilities:
  *   1. Collect entry values for every <tr data-response-id> whose
  *      Status is open/answered (disabled rows skipped automatically).
- *      Widgets: <select> (enum decision), <input data-other-for> (기타
- *      input revealed when select == "__other__"), <input
- *      data-response-id> (material/data-point single-line), <textarea>
- *      (everything else / fallback).
+ *      Widgets: <select> (enum decision), <textarea data-other-for>
+ *      (기타 input revealed when select == "__other__"), <textarea
+ *      data-response-id> (everything else / fallback).
  *   2. Serialise the entries into markdown whose bytes are IDENTICAL
  *      to scripts/okstra_ctl/report_views.py serialize_user_response.
  *   3. Write the result to <pre id="user-response-output">, offer a
@@ -56,7 +55,7 @@
       if (picked === "") return "";
       if (picked === "__other__") {
         var rid = sel.getAttribute("data-response-id") || "";
-        var other = row.querySelector('input[data-other-for="' + rid + '"]');
+        var other = row.querySelector('textarea[data-other-for="' + rid + '"]');
         return other ? trimMultiline(other.value) : "";
       }
       var opt = sel.options[sel.selectedIndex];
@@ -69,11 +68,6 @@
       if (ta.disabled) return "";
       return trimMultiline(ta.value);
     }
-    var inp = row.querySelector("input[data-response-id]");
-    if (inp) {
-      if (inp.disabled) return "";
-      return trimMultiline(inp.value);
-    }
     return "";
   }
 
@@ -119,7 +113,7 @@
     for (var i = 0; i < selects.length; i++) {
       (function (sel) {
         var rid = sel.getAttribute("data-response-id") || "";
-        var other = document.querySelector('input[data-other-for="' + rid + '"]');
+        var other = document.querySelector('textarea[data-other-for="' + rid + '"]');
         if (!other) return;
         var update = function () {
           other.hidden = sel.value !== "__other__";

package/runtime/templates/reports/settings.template.json

@@ -21,8 +21,8 @@
       "Bash($HOME/.okstra/bin/:*)",
       "Bash(codex)",
       "Bash(codex:*)",
-      "Bash(gemini)",
-      "Bash(gemini:*)",
+      "Bash(agy)",
+      "Bash(agy:*)",
       "Bash(claude)",
       "Bash(claude:*)"
     ]

package/runtime/templates/reports/task-brief.template.md

@@ -152,14 +152,14 @@ To declare servers, add entries shaped `{ "name": "<server>", "description": "..
 How to invoke (worker-by-worker):
 
 - **Claude lead / Claude worker / Report writer worker**: invoke the MCP tool **directly by its tool name** (e.g. `mcp__<server>__<tool>`) through the host's tool interface. **Do NOT call it via `Bash`** — these names are MCP tools, not shell commands; running them in a shell will always fail with `command not found` regardless of permission settings.
-- **Codex worker / Gemini worker**: invoke through the external CLI's own MCP transport (e.g. `codex mcp call <server> <tool> <args>` for Codex CLI; the equivalent Gemini CLI MCP invocation for Gemini). If the worker's CLI has no matching MCP config, treat the server as unavailable for this run and record `MCP not available in this CLI` in `Missing Information or Assumptions` — do **not** attempt a shell fallback such as `mysql -h ...` or piping a tool name into `bash`.
+- **Codex worker / Antigravity worker**: invoke through the external CLI's own MCP transport (e.g. `codex mcp call <server> <tool> <args>` for Codex CLI; the equivalent Antigravity CLI MCP invocation for Antigravity). If the worker's CLI has no matching MCP config, treat the server as unavailable for this run and record `MCP not available in this CLI` in `Missing Information or Assumptions` — do **not** attempt a shell fallback such as `mysql -h ...` or piping a tool name into `bash`.
 - All workers: cite the exact server, tool, and SELECT (or `WHERE` filters) used in the result file. Tool-call failures must be logged in the worker's `*-errors.json` (commandKind `mcp_call`) so the lead can decide whether to retry under a different worker.
 
 Usage policy:
 
 - **Allowed phases**: `requirements-discovery`, `error-analysis`, `implementation-planning`, `final-verification`. Use only when local schema/data evidence improves the answer. Always cite the server, table, and the SELECT used as evidence in worker output.
 - **`implementation` phase**: read-only MCP queries are permitted as cross-checks; MCP MUST NOT be used as a write path even if a write tool becomes available — schema/data mutations belong in the codebase migration files reviewed by humans.
-- **External CLI workers (Codex, Gemini)**: can use these MCP servers ONLY if their CLI's own MCP configuration is set up to mirror the same servers. If not configured, the worker should record `MCP not available in this CLI` in `Missing Information or Assumptions` rather than guessing.
+- **External CLI workers (Codex, Antigravity)**: can use these MCP servers ONLY if their CLI's own MCP configuration is set up to mirror the same servers. If not configured, the worker should record `MCP not available in this CLI` in `Missing Information or Assumptions` rather than guessing.
 - **Forbidden**: connecting to non-listed databases, running anything that mutates state (server is read-only — flagged write attempts are a contract violation), persisting query results outside the run's own artifact directories.
 - **Limits**: `maxRows: 1000` per query (server-enforced). For larger scans, page via `WHERE` predicates and document the strategy.
 

package/runtime/templates/worker-prompt-preamble.md

@@ -16,7 +16,7 @@ Different recipients need different files. Do NOT include `final-report-template
 
 | Recipient | Files included in `[Required reading]` |
 |---|---|
-| Claude / Codex / Gemini analysis workers | analysis-packet.md as the primary compact input; task-brief, analysis-profile, analysis-material, reference-expectations, and clarification-response remain source/fallback paths, not automatic first-read files |
+| Claude / Codex / Antigravity analysis workers | analysis-packet.md as the primary compact input; task-brief, analysis-profile, analysis-material, reference-expectations, and clarification-response remain source/fallback paths, not automatic first-read files |
 | Report writer worker (Phase 6) | all of the above **plus** the instruction-set-local `final-report-template.md` (phase-stripped) and `final-report-schema.json` (per-task-type excerpt) — NOT the full `templates/reports/...` / `schemas/...` sources |
 | Reverify dispatches (Phase 5.5, lightweight mode) | **do NOT inject `[Required reading]` at all** — see [convergence](../prompts/lead/convergence.md) "Reverify prompt: required-reading suppression". |
 
@@ -55,7 +55,7 @@ If the sidecar file does not exist, create it with `{"schemaVersion": 1, "errors
 ### Rules
 
 - Do NOT include `source` / `recordedAt` / `agent` / `agentRole` / `model` / `taskKey` — the lead fills those in when dumping the sidecar to the run-level errors log via `okstra-error-log.py append-from-worker`.
-- Do NOT use `errorType` values other than `"tool-failure"` in the sidecar. `cli-failure` events are recorded by CLI wrapper agents (codex / gemini) directly to the run-level errors log, not via the sidecar. `contract-violation` events are recorded by the lead after inspecting worker outputs.
+- Do NOT use `errorType` values other than `"tool-failure"` in the sidecar. `cli-failure` events are recorded by CLI wrapper agents (codex / antigravity) directly to the run-level errors log, not via the sidecar. `contract-violation` events are recorded by the lead after inspecting worker outputs.
 - Continue your task after recording; do not abort unless the failure makes the task impossible.
 
 ### Path extraction (BLOCKING)
@@ -65,7 +65,7 @@ Before recording anything, extract the absolute paths verbatim from the lead's d
 - `**Errors log path:** <abs-path>` — the run-level errors JSONL. Used by CLI wrapper agents via `okstra-error-log.py append-observed` for `cli-failure` events.
 - `**Errors sidecar path:** <abs-path>` — this worker's per-run sidecar JSON. Used by all workers for `tool-failure` events.
 
-If a required header line is absent from the dispatch prompt, return `<SENTINEL_PREFIX>_ERRORS_PATH_MISSING: lead prompt did not include **Errors log path:** / **Errors sidecar path:** headers` (substitute the worker's sentinel prefix — `CLAUDE_WORKER`, `CODEX`, `GEMINI`, or `REPORT_WRITER`) without proceeding. Do NOT synthesize the path from `<runDir>/logs/...` — that template syntax is documentation only; historical failure mode produced silently empty run-level error logs.
+If a required header line is absent from the dispatch prompt, return `<SENTINEL_PREFIX>_ERRORS_PATH_MISSING: lead prompt did not include **Errors log path:** / **Errors sidecar path:** headers` (substitute the worker's sentinel prefix — `CLAUDE_WORKER`, `CODEX`, `ANTIGRAVITY`, or `REPORT_WRITER`) without proceeding. Do NOT synthesize the path from `<runDir>/logs/...` — that template syntax is documentation only; historical failure mode produced silently empty run-level error logs.
 
 ---
 
@@ -76,7 +76,7 @@ Every worker prompt MUST begin with these anchor headers, in this exact order, b
 1. `**Project Root:** <absolute-path>` — required so the worker can self-anchor without relying on inherited cwd.
 2. `**Prompt History Path:** <project-relative-path>`
 3. `**Result Path:** <project-relative-path>` — canonical destination for the worker's result file.
-4. `Assigned worker prompt history path: <absolute-path>` — same as the prompt-history path but resolved against `Project Root`. Codex / Gemini wrapper subagents extract this exact line.
+4. `Assigned worker prompt history path: <absolute-path>` — same as the prompt-history path but resolved against `Project Root`. Codex / Antigravity wrapper subagents extract this exact line.
 5. `**Worker Preamble Path:** <absolute-path>` — points to THIS file. Workers Read it end-to-end before doing anything else.
 6. `**Errors log path:** <absolute-path>` — run-level JSONL (see Error reporting above).
 7. `**Errors sidecar path:** <absolute-path>` — per-worker JSON (see Error reporting above).
@@ -84,7 +84,7 @@ Every worker prompt MUST begin with these anchor headers, in this exact order, b
 For the **implementation phase** specifically, the dispatched prompt MUST also include:
 
 - `**Worktree:** <absolute-path>` — the task worktree path.
-- `cwd for every mutating command: <absolute-path>` — same as Worktree path; used by codex / gemini wrappers as `--add-dir` / `--include-directories`.
+- `cwd for every mutating command: <absolute-path>` — same as Worktree path; used by codex / antigravity wrappers as `--add-dir` / `--include-directories`.
 
 For the **final-verification phase** specifically, the dispatched prompt MUST also include the verification target snapshot so every analyser verifies the SAME target the lead captured at the entry gate:
 
@@ -99,7 +99,7 @@ When a worker reads any project-relative path from the prompt, it MUST resolve i
 
 ## Worker output sections (analysis workers)
 
-Every analysis worker (Claude / Codex / Gemini) produces sections 1–5 with the same dimensions. The lead does NOT bias worker prompts with per-worker emphasis — sections 1–5 are the common core, and specialization lives only in optional Section 6.
+Every analysis worker (Claude / Codex / Antigravity) produces sections 1–5 with the same dimensions. The lead does NOT bias worker prompts with per-worker emphasis — sections 1–5 are the common core, and specialization lives only in optional Section 6.
 
 1. **Findings** — what you identified
 2. **Missing Information or Assumptions** — gaps in the analysis
@@ -122,7 +122,7 @@ The short message you return inline to the lead — the text shown to the user a
 **Model:** <Role>, <modelExecutionValue>
 ```
 
-Emit that line first, then your normal status / summary text on the following lines. This keeps the acting model visible in every worker box — across every provider (Claude / Codex / Gemini / report-writer) and every dispatch type (initial analysis, convergence reverify, report authoring). Copy the value exactly; never guess, abbreviate, or substitute a provider default. If your prompt carries no `**Model:**` line, say so in the return message rather than inventing one. Each worker spec applies this rule at its own concrete return point — see `agents/workers/claude-worker.md` "Stop Condition", `agents/workers/_cli-wrapper-template.md` step 8d, and `agents/workers/report-writer-worker.md` "Authoring Contract".
+Emit that line first, then your normal status / summary text on the following lines. This keeps the acting model visible in every worker box — across every provider (Claude / Codex / Antigravity / report-writer) and every dispatch type (initial analysis, convergence reverify, report authoring). Copy the value exactly; never guess, abbreviate, or substitute a provider default. If your prompt carries no `**Model:**` line, say so in the return message rather than inventing one. Each worker spec applies this rule at its own concrete return point — see `agents/workers/claude-worker.md` "Stop Condition", `agents/workers/_cli-wrapper-template.md` step 8d, and `agents/workers/report-writer-worker.md` "Authoring Contract".
 
 ## Writing style (all prose output — analysis workers + report-writer)
 

package/runtime/validators/forbidden_actions.py

@@ -12,7 +12,7 @@ Bash tool_use 명령을 phase deny-list 와 대조해 위반을 기계적으로
 - Bash tool_use 의 `input.command` 만 본다 — text 블록(설명/금지어 언급)이나 다른
   tool 의 인자는 보지 않으므로 "npm publish 하지 말 것" 같은 산문은 잡히지 않는다.
 
-한계 1: claude-code 세션 jsonl 만 스캔한다. codex/gemini lead·worker 는 동일 형식의
+한계 1: claude-code 세션 jsonl 만 스캔한다. codex/antigravity lead·worker 는 동일 형식의
 tool_use transcript 를 ~/.claude/projects 에 남기지 않으므로 이 스캐너의 사정권 밖이며,
 해당 런타임의 금지 행위는 여전히 수동 감사에 의존한다.
 한계 2: deny-list 는 Bash 명령 문자열 전체를 검색하므로, 금지 토큰이 인자/메시지에

package/runtime/validators/lib/fixtures.sh

@@ -190,13 +190,13 @@ final_status_path = project_root / run_manifest["expectedStatusPath"]
 status_by_worker_id = {
     "claude": "completed",
     "codex": "timeout",
-    "gemini": "error",
+    "antigravity": "error",
     "report-writer": "completed",
 }
 reason_by_worker_id = {
     "claude": "",
     "codex": "Validation fixture timeout",
-    "gemini": "Validation fixture execution error",
+    "antigravity": "Validation fixture execution error",
     "report-writer": "",
 }
 
@@ -371,7 +371,7 @@ report_lines.extend(
         "| Lead | `1` | `1` | `$0.01` |",
         "| Worker 합계 | `1` | `1` | `$0.01` |",
         "| **전체 합계** | **`2`** | **`2`** | **`$0.02`** |",
-        "| Codex/Gemini CLI 추가 비용 |  |  | `$0.00` |",
+        "| Codex/Antigravity CLI 추가 비용 |  |  | `$0.00` |",
         "",
         "## 7. Final Verdict",
         "",

package/runtime/validators/lib/validate-assets.sh

@@ -43,7 +43,7 @@ def check(source_path: Path, target_path: Path) -> None:
 #
 # `_cli-wrapper-template.md` is a build-time render INPUT, not a shipped
 # artifact — tools/build.mjs renders it (with each *.params.json) into
-# codex-worker.md / gemini-worker.md and excludes the template itself from the
+# codex-worker.md / antigravity-worker.md and excludes the template itself from the
 # runtime payload. Skip it here so parity is checked only on shipped files.
 # Keep this set in sync with TEMPLATE_INPUT_BASENAMES in tools/build.mjs.
 template_input_basenames = {"_cli-wrapper-template.md"}

package/runtime/validators/validate-run.py

@@ -610,7 +610,7 @@ _TOKEN_USAGE_SENTINEL_VALUES = frozenset(
 )
 
 # Numeric "valid zero" patterns. These ARE allowed in the CLI row when no
-# Codex/Gemini CLI work was billed; rejected everywhere else.
+# Codex/Antigravity CLI work was billed; rejected everywhere else.
 _TOKEN_USAGE_ZERO_VALUES = frozenset({"0", "$0.00", "$0", "0.00"})
 
 
@@ -630,7 +630,7 @@ def _scan_token_usage_summary(
     real source of `0` / `$0.00` / `pending` shipping in real reports.
 
     Rules:
-    - The Codex/Gemini CLI 추가 비용 row may carry an empty cell or
+    - The Codex/Antigravity CLI 추가 비용 row may carry an empty cell or
       `$0.00` (no CLI work was billed). Sentinel words are still
       rejected.
     - Every other row's backtick-wrapped cells must be either a
@@ -665,7 +665,7 @@ def _scan_token_usage_summary(
             continue
         label_cell = cells[0].strip("* `")
         # The CLI row's label always contains the word "CLI" — matching
-        # `Codex/Gemini CLI 추가 비용` regardless of formatting variations.
+        # `Codex/Antigravity CLI 추가 비용` regardless of formatting variations.
         is_cli_row = "CLI" in label_cell
         for raw_cell in cells[1:]:
             for value in _TOKEN_USAGE_BACKTICK_CELL_RE.findall(raw_cell):
@@ -689,7 +689,7 @@ def _scan_token_usage_summary(
                         "tokens. Re-run `okstra token-usage "
                         "<team-state> --write --summary --substitute-data "
                         "<report-path>` to repopulate from session jsonls. The "
-                        "Codex/Gemini CLI row is the only place `$0.00` is "
+                        "Codex/Antigravity CLI row is the only place `$0.00` is "
                         "allowed (when no CLI work was billed)."
                     )
 
@@ -1017,9 +1017,9 @@ def validate_report(
 
 # Worker-results filename pattern: `<worker-role>-<task-type>-<seq>.md`.
 # Every analysis-worker role name ends in `-worker` (`claude-worker`,
-# `codex-worker`, `gemini-worker`, `report-writer-worker`), so anchor the
-# split on that suffix — otherwise `gemini-worker-error-analysis-001.md`
-# ambiguously parses as `worker=gemini, task=worker-error-analysis`.
+# `codex-worker`, `antigravity-worker`, `report-writer-worker`), so anchor the
+# split on that suffix — otherwise `antigravity-worker-error-analysis-001.md`
+# ambiguously parses as `worker=antigravity, task=worker-error-analysis`.
 # Audit sidecars (`*-audit-*`) and errors sidecars (`.json`) are not
 # matched here.
 _WORKER_RESULT_BASENAME_RE = re.compile(
@@ -2037,7 +2037,7 @@ def _accuracy_failures(updated: dict) -> list[str]:
                 f"worker `{role}` (status=completed) has no usage data — "
                 f"{usage.get('note', 'reason unknown')}."
             )
-        if worker.get("agent") in ("codex", "gemini") and usage.get("source") != "unavailable":
+        if worker.get("agent") in ("codex", "antigravity") and usage.get("source") != "unavailable":
             if "cliTotalTokens" not in usage:
                 reasons.append(
                     f"worker `{role}` ({worker.get('agent')}) wrapper jsonl was located "

package/src/cli-registry.mjs

@@ -103,6 +103,13 @@ export const COMMAND_REGISTRY = [
     category: "introspection",
     summary: ["Estimate file/read context cost for a task bundle"],
   },
+  {
+    name: "error-report",
+    module: "./error-report.mjs",
+    export: "run",
+    category: "introspection",
+    summary: ["Aggregate okstra-run error logs into a task-level report"],
+  },
   {
     name: "worktree-lookup",
     module: "./worktree-lookup.mjs",