메뉴
Use when optional local workflow guidance or legacy v1 workflow compatibility needs diagnosis.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
| name | cybernetic-workflow-infrastructure |
| description | Use when optional local workflow guidance or legacy v1 workflow compatibility needs diagnosis. |
Workflow skills are optional guidance for v2 execution, not control authority or an installation prerequisite. Existing v1 runs may still require their recorded workflow keys.
This skill defines the repo-owned workflow infrastructure used by cybernetic skills.
Cybernetic skills compile approved work chains. The local cybernetic workflow registry and this policy provide the behavior discipline for planning, independent review, execution, debugging, delegation, and completion verification.
Use references/workflow-infrastructure-policy.md.
Workflow resolution is authoritative in
../references/workflow-registry.json. Every active
$cybernetic-*-workflow name must resolve to a real local skill or an explicitly
declared registry-only policy; a dangling name is an installation defect.
Cybernetic workflow disciplines are infrastructure dependencies, not optional style suggestions.
When a required workflow applies:
Approval requires a final observer pass after the last substantive mutation to the reviewed approved files, including required solution design. An older review result must not be transferred to a changed approved work chain unless the change is deterministic-only and guard-covered.
Pre-goal review subagents are authorized by an active requirements or pre-goal
transition gate whose next_action requires independent review and whose
may_ask_user: false marks the action as agent-owned, such as
RunInformationCounterexampleReview. Do not ask the user for separate subagent
authorization. Runtime target-work subagents are authorized only when the final
/goal explicitly contains the approved subagent-driven execution work
assignment and the user launches that /goal; parallel runtime subagents still
require explicit human approval recorded in the execution policy and review.
| Stage | Local cybernetic workflow | Required? |
|---|---|---|
| Exploratory requirements analysis | $cybernetic-brainstorming-workflow | Optional |
| Exploratory solution design | $cybernetic-brainstorming-workflow | Optional |
| Non-trivial execution policy generation | $cybernetic-planning-workflow | Required |
| Review approval | Independent subagent review discipline | Required for Approved unless explicit human approval exists |
| Runtime execution | $cybernetic-execution-workflow discipline | Required |
| Runtime target-work delegation | Approved bounded subagent delegation protocol | Required when execution policy selects serial or parallel subagent-driven work assignment |
| Runtime serial implementation-plan delegation | $cybernetic-serial-subagent-workflow discipline | Conditional: use only when the approved execution policy records Selected agent workflow: cybernetic-serial-subagent-workflow, Subagent execution mode: serial-single-active, and Max concurrent subagents: 1 |
| Runtime parallel independent-domain delegation | $cybernetic-parallel-subagent-workflow discipline | Conditional: use only when the approved execution policy records Selected agent workflow: cybernetic-parallel-subagent-workflow, Subagent execution mode: parallel-max-safe, and approved wave/lock/barrier/integration rules |
| Unclear or repeated runtime failure | $cybernetic-debugging-workflow | Required |
| Completion claim | $cybernetic-verification-workflow | Required |
This skill does not:
/goal commands;It only supplies the shared infrastructure policy those skills must obey.
Use when a formed task needs durable human-approved meaning, completion obligations, evidence claims, non-goals, or live/destructive authority boundaries.
Use when maintaining or resuming an existing v1 requirements.control.json run; new runs use forming-cybernetic-commitments.
Use when an existing v1 run must compile its reviewed control chain into runtime.control.json.
Use when an existing v1 control chain requires design.control.json; v2 strategies remain agent-owned.
Use when cybernetic routing is requested but user input is pre-task intent or role-ambiguous context: confusion, dissatisfaction, risk sense, observed symptoms, failed attempts, method preference, process distrust, source material, declared current state, or unclear requested transformation.
Use when resuming an existing v1 run that already requires pre-goal control-chain compilation.