菜单
Use before any approved Mythos-style defensive cybersecurity run, or when reviewing a Fable/Mythos security workflow. Requires authorization, bounded targets, private evidence handling, independent validation, and a patch/report owner before the model explores vulnerabilities.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Stop Claude Fable 5 from over-planning, re-deriving settled facts, or listing options it won't pursue. Use in interactive sessions where turns feel slow on simple asks, or agentic sessions at high/xhigh effort where the model gathers more context than the task needs.
Keep Claude Fable 5 running to completion in unattended pipelines -- no stalling on a stated intent without the matching tool call, no mid-run permission questions, no self-imposed session splits from context-budget worry. Use for overnight runs, scheduled jobs, and CI agents where nobody is watching to type "continue".
Pick and re-evaluate the output_config.effort level for a Claude Fable 5 workload. Use when designing a pipeline that mixes routine and hard tasks, when a call feels slow or expensive for what it's doing, or when porting an Opus 4.7/4.8 harness that defaulted to xhigh.
Bind every progress claim Claude Fable 5 makes during a long or autonomous run to an actual tool result from that session. Use for multi-hour agents, overnight pipelines, CI runs, or any workflow where a status update has previously claimed work that wasn't actually done.
Maintain a file-based lesson memory that Claude Fable 5 reads and updates across sessions. Use for recurring agents (daily jobs, long projects, team assistants) where the same correction keeps being re-made, or when bootstrapping a new agent from past session history.
Keep Claude Fable 5's diffs mapped one-to-one to the request -- no unrequested refactors, speculative abstractions, or defensive error handling for states that can't occur. Use in code review and any coding session where the diff comes back larger than the ask, especially at high or xhigh effort.
| name | mythos-scope-gate |
| description | Use before any approved Mythos-style defensive cybersecurity run, or when reviewing a Fable/Mythos security workflow. Requires authorization, bounded targets, private evidence handling, independent validation, and a patch/report owner before the model explores vulnerabilities. |
Mythos 5 is not "Fable 5 but with fewer annoying refusals." It is the gated Project Glasswing deployment of the same underlying capability class without Fable's relevant dual-use classifier layer. Treat that as a stronger duty to scope the work, not as permission to run broader tasks.
Do not start a Mythos-style cyber run until the request has a scope packet with:
If any item is missing, stop and ask for the packet. Do not compensate with a more careful prompt.
For claude-fable-5, the control surface includes refusal logging and fallback
handling. For claude-mythos-5, do not treat the absence of a refusal as a
safety signal. The safety signal is scope provenance: what was authorized,
what target was touched, what artifact was produced, who validated it, and who
owns the patch or disclosure.
Reports should be patch-first and scope-bound. Describe exploitability only to the level required for validation and remediation, keep artifacts private, and never publish operational proof-of-concept details from the agent run.
The companion repo includes a runnable validator:
python mythos_scope/scope_gate.py mythos_scope/sample_mythos_scope.json