菜单
Analyze how Codex context grows, what resident instructions cost, and where fidelity stops. Use for context-window pressure, growth curves, loaded-surface audits, or compaction questions.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
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。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/agentmc15/polytropos --skill context-weight命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Run Copilot CLI work through the Goliath five-role pipeline, with the architect as planner and explicit model fallbacks for implementation, test authoring, verification, review, and red-team checks.
Have Fable 5 do the expensive meta-work once — deep plan a complex task (codebase review, greenfield project, large migration) AND build the execution kit (task briefs, model-pinned subagents, skills, verification loops) so Opus/Sonnet can execute at near-Fable quality. Use when the user says "plan this with Fable", "architect this", "review this codebase", or escalates a complex task.
Run an execution kit produced by /polytropos:architect — loop through TASKS.md, dispatching each task to the kit's model-pinned subagents, verifying, and updating state. Use when the user says to execute/continue/resume a kit or plan. Args: the kit slug (optional if only one kit is active).
基于 SOC 职业分类
正在显示 SKILL.md
| name | context-weight |
| description | Analyze how Codex context grows, what resident instructions cost, and where fidelity stops. Use for context-window pressure, growth curves, loaded-surface audits, or compaction questions. |
Set POLYTROPOS_ROOT from this file's real location: in plugin mode, this file is
<root>/codex/skills/context-weight/SKILL.md, so ascend to <root>; in a managed copied install,
use the installer-resolved POLYTROPOS_ROOT="{{POLYTROPOS_ROOT}}". Reject a literal placeholder.
Before shelling out, verify $POLYTROPOS_ROOT/data/pricing.codex.json and
$POLYTROPOS_ROOT/bin/context_weight.py exist. If proof fails, stop and direct the user to
python3 bin/harness_select.py doctor --harness codex; never run a guessed or stale path.
Use only the engine's real Codex surfaces:
python3 "$POLYTROPOS_ROOT/bin/context_weight.py" session --harness codex --codex-home <codex-home>
python3 "$POLYTROPOS_ROOT/bin/context_weight.py" overview --harness codex --codex-home <codex-home>
python3 "$POLYTROPOS_ROOT/bin/context_weight.py" audit --project <repo>
python3 "$POLYTROPOS_ROOT/bin/context_weight.py" demo
Codex rollout logs provide token-count growth curves and record-type byte shares, not content
provenance. Never turn those byte shares into claims about which prompt, instruction, or tool
filled the window. watch is Claude-only, and constraints --harness codex can only return the
engine's explicit fidelity-limit line; Codex cannot answer live pruning or GUARDRAILS.md residency.
Relay the engine's estimated-token and unpriced labels. The audit is about resident tokens, never dollars, and no context estimate is a billing statement.