You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: assets/for-ai/ar/execution-protocol.md
+1Lines changed: 1 addition & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -53,6 +53,7 @@ tags: [ai, protocol, ospec]
53
53
- بوابات القرار وخيارات brainstorm ملك للمستخدم: **لا تختر الخيار الموصى به تلقائياً ولا تحلّ بوابة بنفسك**——اعرض كل بوابة عبر سلّم القدرات (واجهة سؤال أصلية ← واجهة خطة/موافقة ← نص محادثة عادي) وانتظر اختيار المستخدم الفعلي؛ البوابات الإلزامية تحجب التنفيذ والإرسال حتى يجيب المستخدم، و`recommended` مجرد تلميح يُعرض على المستخدم
54
54
- اكتب كل وثيقة change وbrainstorm تنشئها بلغة وثائق المشروع (`documentLanguage` في `.skillrc` / إرشادات `for-ai/` المُدارة)؛ لا تخلط العربية والإنجليزية داخل change واحد
55
55
- استخدم `ospec execute dispatch [changes/active/<change>] [--task task-id] [--limit N]` لإنشاء batch آمن للتوازي من worker packets و`artifacts/agents/execution-session.json`؛ يتضمن كل packet project session brief snapshot وworker profile يوضح capability tier وrecommended target وtarget tool mapping وrationale وrequired behavior. استخدم `ospec execute complete <task-id> ...` لتسجيل نتائج worker. استخدم `--task` لمهمة واحدة صريحة و`--limit` لتحديد حجم dispatch batch. تزامن هذه الأوامر أيضا `artifacts/agents/worker-status.md`، وتحدّث OSpec artifacts فقط ولا تشغّل workers خارجيين؛ عندما تكون النتيجة `NEEDS_CONTEXT` أو `BLOCKED` يكتب `complete` ملفات blocker escalation تحت `artifacts/agents/blockers/`
56
+
- لا تضع القبول الخارجي أو اليدوي في المسار الحرج لمهام تنفيذ غير مرتبطة. إذا كان التنفيذ الدائم مسجلاً وكانت المهمة `BLOCKED` فقط لغياب القبول الخارجي، وبعد تفويض المستخدم الصريح لتأجيله إلى البوابة النهائية، شغّل `ospec execute defer-blocker <task-id> [change-path] --reason "..."`. تبقى المهمة محظورة وغير محددة، ويستمر فقط التنفيذ الآمن من ناحية التبعيات؛ وتظل final review وverify وfinalize وarchive محظورة حتى توفر الدليل الحقيقي
56
57
- توفير الـ tokens (لا يغيّر أي خطوة): مرّر `--brief` على `ospec execute …` لقراءة ملخّص مختصر بدل التقرير الكامل، وقُد كل خطوة من `ospec execute status --brief` بدل إعادة قراءة `task-graph.json` أو `worker-status.md` أو `launch-plan.md` الكاملة في كل دورة — تُكتب الـ artifacts كاملةً على القرص، فافتحها فقط عند الحاجة للتفاصيل
57
58
- بعد dispatch، استخدم `ospec execute launch [changes/active/<change>] [--task task-id] [--target codex|gpt|claude|gemini|opencode|cursor|copilot] [--dry-run] [--json]` لكتابة launch plan؛ يقبل `runtimeAdapter` فقط model-native subagent capability حالية ومرتبطة بالـ target. ينشئ single-task dispatch هذه الخطة تلقائيا
58
59
- نفّذ `runtimeAdapter.selected.nativeSubagent` وشغّل safe batch فقط بالتوازي. عند غياب capability أو انتهاء صلاحيتها يجب block من دون agent CLI أو fallback إلى current controller
Copy file name to clipboardExpand all lines: assets/for-ai/en-US/execution-protocol.md
+1Lines changed: 1 addition & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -55,6 +55,7 @@ tags: [ai, protocol, ospec]
55
55
- Decision gates and brainstorm options belong to the user: never auto-select the recommended option or resolve a gate yourself — present every gate with the capability ladder (a native question UI, else a plan/approval UI, else plain chat text) and wait for the user's actual choice; required gates block implementation and dispatch until the user answers, and `recommended` is only a hint to show the user
56
56
- Write every change document and brainstorm you author in the project's document language (`.skillrc``documentLanguage` / managed `for-ai/` guidance); do not mix Chinese and English within a change
57
57
- Use `ospec execute dispatch [changes/active/<change>] [--task task-id] [--limit N]` to create a parallel-safe batch of worker packets and `artifacts/agents/execution-session.json`; each packet includes the project session brief snapshot and a worker profile with capability tier, recommended target, target tool mapping, rationale, and required behavior. Use `ospec execute complete <task-id> ...` to record worker results. Use `--task` for one explicit task and `--limit` to cap dispatch batch size. These commands also sync `artifacts/agents/worker-status.md`, update OSpec artifacts only, and do not launch external workers; required pending user decisions block dispatch; `complete` writes blocker escalation artifacts under `artifacts/agents/blockers/` when the result is `NEEDS_CONTEXT` or `BLOCKED`
58
+
- Keep external/manual acceptance out of the critical path for unrelated implementation tasks. If a durable implementation is already recorded as `BLOCKED` only because external acceptance is unavailable, and the user explicitly authorizes deferring that acceptance, run `ospec execute defer-blocker <task-id> [change-path] --reason "..."`. This leaves the task blocked and unchecked while dependency-safe implementation continues; final review, verification, finalization, and archive remain blocked until the real evidence exists
58
59
- Use `ospec execute launch [changes/active/<change>] [--task task-id] [--target codex|gpt|claude|gemini|opencode|cursor|copilot] [--dry-run] [--json]` after dispatch to write the launch plan. Its `runtimeAdapter` accepts only a current, target-bound model-native subagent capability. A single-task dispatch generates this plan automatically
59
60
- Execute `runtimeAdapter.selected.nativeSubagent`; parallelize only the safe batch. Missing or expired native capability blocks dispatch with no agent CLI or controller-context fallback
60
61
-`IDE-CONTROLLER-AUTO-DISPATCH`: L1 is report-only. For L2/L3, the IDE AI owns tick -> execute all `actions[]` through each action's `runtimeAdapter.selected.nativeSubagent` -> persist heartbeat/result evidence -> immediate tick. An empty `actions[]` with `pending` is observation only and must never be relaunched
Copy file name to clipboardExpand all lines: assets/global-skills/codex/ospec-goal/SKILL.md
+3-1Lines changed: 3 additions & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -24,6 +24,7 @@ Use this skill for complex work that needs the full OSpec workflow. A goal is in
24
24
-**Guards are enforced before new work.** Pause/STOP, iteration/deadline/token/time budgets, no-progress limits, comprehension-review checkpoints, required decisions, approved document reviews, ready workspace evidence, and L3 allowlists can stop or pause the loop.
25
25
-**Document review is convergent.** Specialist design/plan review uses two rounds and 30 minutes as default convergence thresholds. In continuous mode, a new structured finding-ID set may continue automatically; a repeated or cycling set stops. Cache/pending reuse, heartbeat, lease recovery, and deterministic preflight do not consume rounds. Never use `--force` to bypass a guard. Strict mode retains the exact user-authorized extra-round window. Read the prior findings sidecar and resolution evidence before the full document when revising. Every structured finding must have a non-empty unique ID.
26
26
-**Review repair is convergent and regression-aware.** A downstream task that shares target files inherits transitive upstream regression obligations. The default two-round values are convergence thresholds. Changed structured finding IDs continue automatically. A stable ID may also continue only when both its structured finding fingerprint and its authorized repair-scope code snapshot changed; wording-only or code-only churn, exact repeats, and cycles stop. A blocked final review requires blocker resolution and must not enter grouped repair. Never raise a limit to bypass unchanged evidence.
27
+
-**External acceptance may be deferred, never waived.** Keep device, credential, manual, and third-party acceptance out of unrelated implementation critical paths. When durable implementation exists and the user explicitly authorizes moving only external acceptance to the final gate, use `ospec execute defer-blocker`; the task stays `BLOCKED` and unchecked, while dependency-safe implementation may continue. Final review, verification, finalization, and archive remain hard-blocked until the real evidence exists.
27
28
-**Allowlist updates are replacement or explicit CAS.** Repeated `loop configure --allow-*` calls replace the complete selected list and print a diff; they never append. For L3 derive/check/apply the exact task-graph permissions, review the diff, and pass `--approve-expansion` only for an intended expansion.
28
29
-**Stop condition is three-stage:** run the project's real tests, record evidence with `ospec execute verify --status`, then confirm with `ospec verify`.
29
30
@@ -78,7 +79,7 @@ For legacy root-layout projects, use the same paths without the `.ospec/` prefix
78
79
4. If the matching active goal already exists, continue it instead of duplicating it.
79
80
5. Draft or update `design.md` from the requirement, `proposal.md`, and project context before editing `implementation-plan.md`, deriving `artifacts/agents/task-graph.json`, editing `tasks.md`, or editing code.
80
81
6. Draft or update `implementation-plan.md` from `design.md`; identify target files, expected results, verification commands, dependencies, parallelizable work, and conflicts.
81
-
7. Derive `artifacts/agents/task-graph.json` from `implementation-plan.md`; give every task a `documentation_updates` array (`[]` when none), include every declared docs path in the same task's `target_files`, and require meaningful-change evidence from dispatch to completion; derive `tasks.md` from the task graph. Mark every dependency/file-safe task `parallelizable: true`; every generated `parallelizable: false` task must include `serial_reason`, and `maxParallel=1` must include `maxParallelReason`. Split tasks with more than six `target_files`; when one atomic verification boundary genuinely requires that breadth, record a concrete `scope_reason`. For L3 use `ospec loop allowlist derive/check/apply --from-task-graph` so exact paths and commands fail before expensive document reviewers run; never assume repeated configure calls append. Finalize also generates one indexed `docs/project/changes/<archive-path>.md` for this goal; its preflight refuses to overwrite a human-owned file at that path, and the generated summary does not replace required architecture, API, module, or operational documentation.
82
+
7. Derive `artifacts/agents/task-graph.json` from `implementation-plan.md`; give every task a `documentation_updates` array (`[]` when none), include every declared docs path in the same task's `target_files`, and require meaningful-change evidence from dispatch to completion; derive `tasks.md` from the task graph. Mark every dependency/file-safe task `parallelizable: true`; every generated `parallelizable: false` task must include `serial_reason`, and `maxParallel=1` must include `maxParallelReason`. Split tasks with more than six `target_files`; when one atomic verification boundary genuinely requires that breadth, record a concrete `scope_reason`. Split implementation/automatic checks from external device, credential, third-party, or manual acceptance so an unavailable external gate cannot block unrelated implementation; downstream work may depend on the implementation slice, while the external acceptance remains a final hard gate. For L3 use `ospec loop allowlist derive/check/apply --from-task-graph` so exact paths and commands fail before expensive document reviewers run; never assume repeated configure calls append. Finalize also generates one indexed `docs/project/changes/<archive-path>.md` for this goal; its preflight refuses to overwrite a human-owned file at that path, and the generated summary does not replace required architecture, API, module, or operational documentation.
82
83
8. Run `ospec execute doc-review [changes/active/<goal>] --stage design`. When it reports `Reused approval: yes`, do not launch another reviewer. Otherwise execute one fresh independent design reviewer through `dispatch.runtimeAdapter`, immediately claim its real executor id, wait for Markdown plus structured findings, then complete that executor. Approval must validate before plan review.
83
84
9. Run `ospec execute doc-review [changes/active/<goal>] --stage plan`. Reuse a valid approval; otherwise execute one fresh independent plan reviewer through `dispatch.runtimeAdapter`, claim its real executor id, wait for findings, then complete it before worker dispatch. The controlling AI must not self-approve either review.
84
85
10. Use `ospec execute decision` for direction, architecture, API, UI, risk, or scope choices that need explicit user selection, and always include `--answered-by user` when persisting the user's answer. Persist every named browser/E2E/manual verification requirement before implementation.
0 commit comments