Evidence - xsyetopz/claudetto GitHub Wiki
This page is the index of the evidence pages. Read it to find the page and the source file of a measurement. Each row gives what the page measures, its date, its Claude Code version, and its source file. If a page does not state a date or a version, the row says "not stated".
| Page | What it measures | Date | Claude Code version | Source file |
|---|---|---|---|---|
| Handoff Cost | The cost of one /dotclaude:handoff of dotclaude 1.0.0 and one /compact, from the same session |
2026-10-12 | 2.1.296 |
wiki/evidence/Handoff-Cost.md |
| Usage Evidence | The plans, the prices, the models, and the usage of claudetto 0.1.0, with one measured week of dotclaude 0.x | Measured week 2026-09-21 to 09-28, measures of 2026-10-03, prices of 2026-10-13 |
2.1.287 for one fact, otherwise not stated |
wiki/evidence/Usage-Evidence.md |
| Prompt Variants | The system prompt variant of each model family, the verification line of each subagent block, and the size bounds of the first request | Not stated | Not stated |
wiki/evidence/Prompt-Variants.md, with the variants in src/plugins/claudetto/slices/prompt/domain/variants.ts
|
| Evals | The suite of 16 cases, how to run it, and where its results are | Newest full run 2026-10-10, of the 12 cases of dotclaude 1.0.0 |
2.1.296 for the runs |
wiki/evidence/Evals.md, with the cases in src/plugins/claudetto/evals/, src/plugins/claudetto/slices/*/evals/, and src/plugins/claudetto-you-should-know/evals/
|
| Model Guidance | What the prompting guides, the claude-api skill, and the system cards say about false completion and the end of a turn |
2026-10-16 | 2.1.301 |
wiki/evidence/Model-Guidance.md |
| Failure Gaps of 0.1.0 | The agent failure dossier compared with the mechanisms of claudetto 0.1.0 | 2026-10-16 | 2.1.301 |
wiki/evidence/Failure-Gaps-0.1.0.md |
The design records of claudetto 0.1.0 are in openspec/changes/claudetto-0-1-0/.
The measurements of earlier releases are in Evals History and Open Items.
- Handoff Cost: the first measurement page to read after this index.
- Usage Evidence
- Prompt Variants
- Evals
- Evals of 0.1.0
- Model Guidance
- Failure Gaps of 0.1.0