docs(token): update cost evaluation for v0.23.13#3
Conversation
- Update skill count: 10→18 (include global/project/npm skills) - Update tool count: 13→24 (include built-in tools) - Add v0.18.0→v0.23.13 change summary section - Document SKILL.md ~60% token reduction - Add input hook stage-routing to advantages
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: ceb5b4c1e8
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| | super-pi 管线 skill | 8 | 01-brainstorm ~ 08-help | | ||
| | 全局通用 skill | 9 | agent-browser, caveman, cli-tool, context7, git, handoff, html-write, skill-write, … | | ||
| | 项目本地 skill | 1 | github-commit-super-pi | | ||
| | 外部 npm skill | 1 | super-cli | | ||
| | **合计** | **18** | | |
There was a problem hiding this comment.
The source breakdown currently sums to 19 skills (8 + 9 + 1 + 1), but the table reports a total of 18, which makes the fixed-cost accounting internally inconsistent and undermines the later cost totals derived from this section. Please reconcile either the per-source counts or the total so readers can trust the v0.23.13 baseline.
Useful? React with 👍 / 👎.
| | 内置 Tools | 未统计 | 10 | 首次纳入 | | ||
| | 总 Tools | 13 | 24 | +11 | | ||
| | Rules | 78 | 78 | 持平 | | ||
| | Hooks | 5（均 0 token） | 5（均 0 token） | 持平 | |
There was a problem hiding this comment.
This summary says there are 5 hooks, but the document’s own hook table in section one lists 4 entries and the extension currently registers 4 hook handlers (input, two tool_result, and session_before_tree), so the comparison row is stale. Keeping this count incorrect makes the version-to-version comparison unreliable for cost and architecture decisions.
Useful? React with 👍 / 👎.
1 participant
变更说明
更新 token 成本评估文档，从 v0.18.0 基准更新到 v0.23.13：
2,500→4,345 tokens（口径变化，非膨胀）变更范围