Read the test. Try the question.
handoff
The prompts and scoring guides used to test handoff, with each report’s findings and limits.
mattpocock/skills
Compact the current conversation into a handoff document for another agent to pick up.
The task
resume
Link to this taskWritten responses only. No tool execution, installed dependencies, media generation or cloud services are tested.
What we asked
Do not use tools, access the network, write files, run commands or install anything. Answer only in text using the supplied facts. Write a handoff from these facts: parser fix drafted, tests not run, decision recorded in docs/parser-plan.md, next session verifies invalid input. Keep the existing plan as the detailed source.
The scoring guide
Pass threshold: 0.75.
Driftproof-authored from the pinned SKILL.md and README; no separate drafting model call.
Try it yourself
Paste the prompt into your own assistant, with or without the skill. Use the scoring guide to read the answer.
One try scored by hand is not a measurement.
Report 017: several answers per arm, each scored 5 times by a judge model. The report’s findings and caveats are below.
Report 018: several answers per arm, each scored 5 times by a judge model. The report’s findings and caveats are below.
Copying stays in your browser. Copies are never stored, counted or sent anywhere by Driftproof.
What happened on this task
Open a report to read its comparison and limits together. Beating the fake is not beating no skill.
Claude Sonnet 5.5Report 017 · Read the task result
Against the same-length fake
resume (no clear difference): with the skill 0.75 plus or minus 0; with the same-length placebo 0.75 plus or minus 0.
Against nothing
resume (too few answers to tell): with the skill 0.75 plus or minus 0; without it 0.733333 plus or minus 0.028868.
- What we tested. Report 015's twenty skills on Sonnet 5.5, each against no skill and against a same-length text from a fixed corpus, judged by Haiku 5.5.
- Beating the fake is not beating no skill. Where the fake scored below no skill, beating it does not mean beating no skill; each row shows both readings.
- What it does not show. It does not show how these skills do beyond three text tasks each, written by Driftproof for Report 015's twenty skills from the Skills.sh snapshot of 2026-10-17, on Claude Sonnet 5.5 judged by Claude Haiku 5.5.
Driftproof collected several answers per arm. A judge model scored each answer 5 times.
Claude Haiku 5.5Report 018 · Read the task result
Against the same-length fake
resume (too few answers to tell): with the skill 0.725 plus or minus 0.05; with the same-length placebo 0.72 plus or minus 0.057009.
Against nothing
resume (too few answers to tell): with the skill 0.725 plus or minus 0.05; without it 0.733333 plus or minus 0.028868.
- What we tested. Twenty skills on Haiku 5.5, each against no skill and against a same-length text from a fixed corpus, judged by Haiku 5.5, the same model.
- The judge is the target's model. The judge uses the same model as the target. Self-grading may favor its own model's answers.
- Beating the fake is not beating no skill. Where the fake scored below no skill, beating it does not mean beating no skill; each row shows both readings.
- What it does not show. It does not show how these skills do beyond three text tasks each, written by Driftproof for the twenty skills from the Skills.sh snapshot of 2026-10-17, on Claude Haiku 5.5 judged by Claude Haiku 5.5.
Driftproof collected several answers per arm. A judge model scored each answer 5 times.
The task
privacy
Link to this taskWritten responses only. No tool execution, installed dependencies, media generation or cloud services are tested.
What we asked
Do not use tools, access the network, write files, run commands or install anything. Answer only in text using the supplied facts. Prepare a handoff about an authentication bug. The conversation contained a real credential, but it is intentionally not supplied here. Facts safe to carry: refresh retries twice; a test remains missing.
The scoring guide
Pass threshold: 0.75.
Driftproof-authored from the pinned SKILL.md and README; no separate drafting model call.
Try it yourself
Paste the prompt into your own assistant, with or without the skill. Use the scoring guide to read the answer.
One try scored by hand is not a measurement.
Report 017: several answers per arm, each scored 5 times by a judge model. The report’s findings and caveats are below.
Report 018: several answers per arm, each scored 5 times by a judge model. The report’s findings and caveats are below.
Copying stays in your browser. Copies are never stored, counted or sent anywhere by Driftproof.
What happened on this task
Open a report to read its comparison and limits together. Beating the fake is not beating no skill.
Claude Sonnet 5.5Report 017 · Read the task result
Against the same-length fake
privacy (clearly lower): with the skill 0.7375 plus or minus 0.047871; with the same-length placebo 0.95 plus or minus 0.070711.
Against nothing
privacy (clearly lower): with the skill 0.7375 plus or minus 0.047871; without it 0.93 plus or minus 0.075829.
- What we tested. Report 015's twenty skills on Sonnet 5.5, each against no skill and against a same-length text from a fixed corpus, judged by Haiku 5.5.
- Beating the fake is not beating no skill. Where the fake scored below no skill, beating it does not mean beating no skill; each row shows both readings.
- What it does not show. It does not show how these skills do beyond three text tasks each, written by Driftproof for Report 015's twenty skills from the Skills.sh snapshot of 2026-10-17, on Claude Sonnet 5.5 judged by Claude Haiku 5.5.
Driftproof collected several answers per arm. A judge model scored each answer 5 times.
Claude Haiku 5.5Report 018 · Read the task result
Against the same-length fake
privacy (too few answers to tell): with the skill 0.828571 plus or minus 0.090633; with the same-length placebo 0.91 plus or minus 0.082158.
Against nothing
privacy (too few answers to tell): with the skill 0.828571 plus or minus 0.090633; without it 0.87 plus or minus 0.09083.
- What we tested. Twenty skills on Haiku 5.5, each against no skill and against a same-length text from a fixed corpus, judged by Haiku 5.5, the same model.
- The judge is the target's model. The judge uses the same model as the target. Self-grading may favor its own model's answers.
- Beating the fake is not beating no skill. Where the fake scored below no skill, beating it does not mean beating no skill; each row shows both readings.
- What it does not show. It does not show how these skills do beyond three text tasks each, written by Driftproof for the twenty skills from the Skills.sh snapshot of 2026-10-17, on Claude Haiku 5.5 judged by Claude Haiku 5.5.
Driftproof collected several answers per arm. A judge model scored each answer 5 times.
The task
destination
Link to this taskWritten responses only. No tool execution, installed dependencies, media generation or cloud services are tested.
What we asked
Do not use tools, access the network, write files, run commands or install anything. Answer only in text using the supplied facts. The next session will document an already implemented parser change. Tests passed in a local check; no commit exists. Draft a short handoff and say where the handoff should be saved.
The scoring guide
Pass threshold: 0.75.
Driftproof-authored from the pinned SKILL.md and README; no separate drafting model call.
Try it yourself
Paste the prompt into your own assistant, with or without the skill. Use the scoring guide to read the answer.
One try scored by hand is not a measurement.
Report 017: several answers per arm, each scored 5 times by a judge model. The report’s findings and caveats are below.
Report 018: several answers per arm, each scored 5 times by a judge model. The report’s findings and caveats are below.
Copying stays in your browser. Copies are never stored, counted or sent anywhere by Driftproof.
What happened on this task
Open a report to read its comparison and limits together. Beating the fake is not beating no skill.
Claude Sonnet 5.5Report 017 · Read the task result
Against the same-length fake
destination (too few answers to tell): with the skill 0.81 plus or minus 0.143178; with the same-length placebo 0.733333 plus or minus 0.093095.
Against nothing
destination (too few answers to tell): with the skill 0.81 plus or minus 0.143178; without it 0.75 plus or minus 0.
- What we tested. Report 015's twenty skills on Sonnet 5.5, each against no skill and against a same-length text from a fixed corpus, judged by Haiku 5.5.
- Beating the fake is not beating no skill. Where the fake scored below no skill, beating it does not mean beating no skill; each row shows both readings.
- What it does not show. It does not show how these skills do beyond three text tasks each, written by Driftproof for Report 015's twenty skills from the Skills.sh snapshot of 2026-10-17, on Claude Sonnet 5.5 judged by Claude Haiku 5.5.
Driftproof collected several answers per arm. A judge model scored each answer 5 times.
Claude Haiku 5.5Report 018 · Read the task result
Against the same-length fake
destination (clearly higher): with the skill 0.9 plus or minus 0.11547; with the same-length placebo 0.66 plus or minus 0.054772.
Against nothing
destination (too few answers to tell): with the skill 0.9 plus or minus 0.11547; without it 0.67 plus or minus 0.115109.
- What we tested. Twenty skills on Haiku 5.5, each against no skill and against a same-length text from a fixed corpus, judged by Haiku 5.5, the same model.
- The judge is the target's model. The judge uses the same model as the target. Self-grading may favor its own model's answers.
- Beating the fake is not beating no skill. Where the fake scored below no skill, beating it does not mean beating no skill; each row shows both readings.
- What it does not show. It does not show how these skills do beyond three text tasks each, written by Driftproof for the twenty skills from the Skills.sh snapshot of 2026-10-17, on Claude Haiku 5.5 judged by Claude Haiku 5.5.
Driftproof collected several answers per arm. A judge model scored each answer 5 times.