documentation-and-adrs: worth installing? Measured results

Measured results for documentation-and-adrs (addyosmani/agent-skills) from Driftproof, one row per receipt, in plain words.

Skill skills/documentation-and-adrs in addyosmani/agent-skills.

One row per receipt. No figure on this page combines two rows.

Measured by: Measured by Driftproof, Report 014
Model: Claude Haiku 4.5 (claude-haiku-4-5-20251001)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-10-07
Tasks and runs: 1 task; 6 runs with the skill and 7 without
Result: Too few answers to tell: 0.739 with the skill and 0.605 without, but the scores varied too much to call it. Not enough draws to conclude at this effect floor. Draws needed: no draw count at these spreads (case comment-intent-not-implementation: the two arms' spreads sum to 0.177, at or above the 0.05 floor).
Tokens: 16600 input and 1203 output tokens per call with the skill; 14139 and 1075 without. 0.0226 US dollars per call with the skill, 0.0195 without, at list price from the receipt's price snapshot.
Freshness: Measured on Haiku 4.5. The registry lists no haiku model released after this run, as of the build date.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-haiku-4-5-20251001-2026-10-07.json

Measured by: Measured by Driftproof, Report 014
Model: Claude Haiku 4.5 (claude-haiku-4-5-20251001)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-10-07
Tasks and runs: 1 task; 5 runs with the skill and 4 without
Result: The skill clearly helped: 0.772 with it and 0.579 without, on 1 test task.
Tokens: 16600 input and 1344 output tokens per call with the skill; 14139 and 1112 without. 0.0233 US dollars per call with the skill, 0.0197 without, at list price from the receipt's price snapshot.
Freshness: Measured on Haiku 4.5. The registry lists no haiku model released after this run, as of the build date.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-haiku-4-5-20251001-2026-10-07.json

Measured by: Measured by Driftproof, Report 014
Model: Claude Haiku 5.5 (claude-haiku-5-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-10-07
Tasks and runs: 1 task; 5 runs with the skill and 3 without
Result: Too few answers to tell: 0.542 with the skill and 0.483 without, but the scores varied too much to call it. Not enough draws to conclude at this effect floor. Draws needed: no draw count at these spreads (case comment-intent-not-implementation: the two arms' spreads sum to 0.100, at or above the 0.05 floor).
Tokens: 23354 input and 937 output tokens per call with the skill; 19891 and 1185 without. 0.0028 US dollars per call with the skill, 0.0026 without, at list price from the receipt's price snapshot.
Freshness: Measured on claude-haiku-5-5. Freshness unknown: the model is not in the registry.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-haiku-5-5-2026-10-07.json

Measured by: Measured by Driftproof, Report 014
Model: Claude Haiku 4.5 (claude-haiku-4-5-20251001)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-10-07
Tasks and runs: 1 task; 5 runs with the skill and 3 without
Result: The skill clearly helped: 0.763 with it and 0.626 without, on 1 test task.
Tokens: 16600 input and 1246 output tokens per call with the skill; 14139 and 914 without. 0.0228 US dollars per call with the skill, 0.0187 without, at list price from the receipt's price snapshot.
Freshness: Measured on Haiku 4.5. The registry lists no haiku model released after this run, as of the build date.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-haiku-4-5-20251001-2026-10-07.json

Measured by: Measured by Driftproof, Report 014
Model: Claude Haiku 5.5 (claude-haiku-5-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-10-07
Tasks and runs: 1 task; 5 runs with the skill and 3 without
Result: Too few answers to tell: 0.500 with the skill and 0.589 without, but the scores varied too much to call it. Not enough draws to conclude at this effect floor. Draws needed: no draw count at these spreads (case comment-intent-not-implementation: the two arms' spreads sum to 0.092, at or above the 0.05 floor).
Tokens: 23354 input and 1111 output tokens per call with the skill; 19891 and 1071 without. 0.0029 US dollars per call with the skill, 0.0025 without, at list price from the receipt's price snapshot.
Freshness: Measured on claude-haiku-5-5. Freshness unknown: the model is not in the registry.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-haiku-5-5-2026-10-07.json

Measured by: Measured by Driftproof, Report 014
Model: Claude Haiku 5.5 (claude-haiku-5-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-10-07
Tasks and runs: 1 task; 3 runs with the skill and 3 without
Result: Too few answers to tell: 0.597 with the skill and 0.561 without, but the scores varied too much to call it. Not enough draws to conclude at this effect floor. Draws needed: 1376 draws per arm (case comment-intent-not-implementation), with the spreads held.
Tokens: 23354 input and 1137 output tokens per call with the skill; 19891 and 1223 without. 0.0029 US dollars per call with the skill, 0.0026 without, at list price from the receipt's price snapshot.
Freshness: Measured on claude-haiku-5-5. Freshness unknown: the model is not in the registry.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-haiku-5-5-2026-10-07.json

Measured by: Measured by Driftproof, Report 013
Model: Claude Opus 5.5 (claude-opus-5-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-09-29
Tasks and runs: 1 task; 3 runs with the skill and 6 without
Result: The skill clearly helped: 0.840 with it and 0.518 without, on 1 test task.
Tokens: 16253 input and 1104 output tokens per call with the skill; 12790 and 769 without. 0.0871 US dollars per call with the skill, 0.0665 without, at list price from the receipt's price snapshot.
Freshness: Measured on Opus 5.5. The registry lists no opus model released after this run, as of the build date.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-opus-5-5-2026-09-29.json

Measured by: Measured by Driftproof, Report 013
Model: Claude Sonnet 5 (claude-sonnet-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-09-29
Tasks and runs: 1 task; 5 runs with the skill and 5 without
Result: Too few answers to tell: 0.641 with the skill and 0.727 without, but the scores varied too much to call it. Not enough draws to conclude at this effect floor. Draws needed: no draw count at these spreads (case comment-intent-not-implementation: the two arms' spreads sum to 0.285, at or above the 0.05 floor).
Tokens: 22675 input and 686 output tokens per call with the skill; 19212 and 390 without. 0.0522 US dollars per call with the skill, 0.0423 without, at list price from the receipt's price snapshot.
Freshness: Measured on Sonnet 5. The registry lists no sonnet model released after this run, as of the build date.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-sonnet-5-2026-09-29.json

Measured by: Measured by Driftproof, Report 013
Model: Claude Sonnet 5.5 (claude-sonnet-5-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-09-29
Tasks and runs: 1 task; 3 runs with the skill and 3 without
Result: The skill clearly helped: 0.847 with it and 0.574 without, on 1 test task.
Tokens: 16257 input and 505 output tokens per call with the skill; 12794 and 506 without. 0.0376 US dollars per call with the skill, 0.0306 without, at list price from the receipt's price snapshot.
Freshness: Measured on claude-sonnet-5-5. Freshness unknown: the model is not in the registry.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-sonnet-5-5-2026-09-29.json

Measured by: Measured by Driftproof, Report 013
Model: Claude Opus 5.5 (claude-opus-5-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-09-29
Tasks and runs: 1 task; 3 runs with the skill and 3 without
Result: The skill clearly helped: 0.850 with it and 0.550 without, on 1 test task.
Tokens: 16253 input and 1036 output tokens per call with the skill; 12790 and 812 without. 0.0857 US dollars per call with the skill, 0.0674 without, at list price from the receipt's price snapshot.
Freshness: Measured on Opus 5.5. The registry lists no opus model released after this run, as of the build date.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-opus-5-5-2026-09-29.json

Measured by: Measured by Driftproof, Report 013
Model: Claude Sonnet 5 (claude-sonnet-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-09-29
Tasks and runs: 1 task; 5 runs with the skill and 5 without
Result: Too few answers to tell: 0.700 with the skill and 0.706 without, but the scores varied too much to call it. Not enough draws to conclude at this effect floor. Draws needed: no draw count at these spreads (case comment-intent-not-implementation: the two arms' spreads sum to 0.203, at or above the 0.05 floor).
Tokens: 22675 input and 545 output tokens per call with the skill; 19212 and 372 without. 0.0508 US dollars per call with the skill, 0.0421 without, at list price from the receipt's price snapshot.
Freshness: Measured on Sonnet 5. The registry lists no sonnet model released after this run, as of the build date.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-sonnet-5-2026-09-29.json

Measured by: Measured by Driftproof, Report 013
Model: Claude Sonnet 5.5 (claude-sonnet-5-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-09-29
Tasks and runs: 1 task; 7 runs with the skill and 3 without
Result: The skill clearly helped: 0.813 with it and 0.576 without, on 1 test task.
Tokens: 16257 input and 578 output tokens per call with the skill; 12794 and 449 without. 0.0383 US dollars per call with the skill, 0.0301 without, at list price from the receipt's price snapshot.
Freshness: Measured on claude-sonnet-5-5. Freshness unknown: the model is not in the registry.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-sonnet-5-5-2026-09-29.json

Measured by: Measured by Driftproof, Report 013
Model: Claude Opus 5.5 (claude-opus-5-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-09-29
Tasks and runs: 1 task; 6 runs with the skill and 3 without
Result: Too few answers to tell: 0.650 with the skill and 0.534 without, but the scores varied too much to call it. Not enough draws to conclude at this effect floor. Draws needed: no draw count at these spreads (case comment-intent-not-implementation: the two arms' spreads sum to 0.141, at or above the 0.05 floor).
Tokens: 16253 input and 1030 output tokens per call with the skill; 12790 and 778 without. 0.0856 US dollars per call with the skill, 0.0667 without, at list price from the receipt's price snapshot.
Freshness: Measured on Opus 5.5. The registry lists no opus model released after this run, as of the build date.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-opus-5-5-2026-09-29.json

Measured by: Measured by Driftproof, Report 013
Model: Claude Sonnet 5 (claude-sonnet-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-09-29
Tasks and runs: 1 task; 3 runs with the skill and 4 without
Result: Too few answers to tell: 0.602 with the skill and 0.661 without, but the scores varied too much to call it. Not enough draws to conclude at this effect floor. Draws needed: no draw count at these spreads (case comment-intent-not-implementation: the two arms' spreads sum to 0.191, at or above the 0.05 floor).
Tokens: 22675 input and 537 output tokens per call with the skill; 19212 and 342 without. 0.0507 US dollars per call with the skill, 0.0418 without, at list price from the receipt's price snapshot.
Freshness: Measured on Sonnet 5. The registry lists no sonnet model released after this run, as of the build date.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-sonnet-5-2026-09-29.json

Measured by: Measured by Driftproof, Report 013
Model: Claude Sonnet 5.5 (claude-sonnet-5-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-09-29
Tasks and runs: 1 task; 6 runs with the skill and 3 without
Result: The skill clearly helped: 0.781 with it and 0.572 without, on 1 test task.
Tokens: 16257 input and 592 output tokens per call with the skill; 12794 and 422 without. 0.0384 US dollars per call with the skill, 0.0298 without, at list price from the receipt's price snapshot.
Freshness: Measured on claude-sonnet-5-5. Freshness unknown: the model is not in the registry.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-sonnet-5-5-2026-09-29.json

Measured by: Measured by Driftproof, Report 011
Model: Claude Opus 5 (claude-opus-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-09-23
Tasks and runs: 1 task; 7 runs with the skill and 3 without
Result: The skill clearly helped: 0.804 with it and 0.567 without, on 1 test task.
Tokens: 8187 input and 1735 output tokens per call with the skill; 4724 and 1147 without. 0.0843 US dollars per call with the skill, 0.0523 without, at list price from the receipt's price snapshot.
Freshness: Measured on Opus 5. The registry lists no opus model released after this run, as of the build date.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-opus-5-2026-09-23.json

Measured by: Measured by Driftproof, Report 011
Model: Claude Opus 5.5 (claude-opus-5-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-09-23
Tasks and runs: 1 task; 6 runs with the skill and 3 without
Result: Too few answers to tell: 0.661 with the skill and 0.567 without, but the scores varied too much to call it. Not enough draws to conclude at this effect floor. Draws needed: no draw count at these spreads (case comment-intent-not-implementation: the two arms' spreads sum to 0.148, at or above the 0.05 floor).
Tokens: 7187 input and 857 output tokens per call with the skill; 3724 and 612 without. 0.0459 US dollars per call with the skill, 0.0271 without, at list price from the receipt's price snapshot.
Freshness: Measured on Opus 5.5. The registry lists no opus model released after this run, as of the build date.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-opus-5-5-2026-09-23.json

Measured by: Measured by Driftproof, Report 009
Model: Claude Opus 5 (claude-opus-5)
Harness: Driftproof runner 0.10.1 on claude-cli
Date: 2026-09-15
Tasks and runs: 1 task; 3 runs with the skill and 4 without
Result: The skill clearly helped: 0.822 with it and 0.532 without, on 1 test task.
Tokens: 6077 input and 1165 output tokens per call with the skill; 2614 and 922 without. 0.0595 US dollars per call with the skill, 0.0361 without, at list price from the receipt's price snapshot.
Freshness: Measured on Opus 5. Registered opus models released after this run, as of the build date: Opus 5.5.
Skill commit: be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39
Receipt: documentation-and-adrs-claude-opus-5-2026-09-15.json

Receipts by Driftproof (driftproofhq/driftproof), licensed Apache-2.0.

Disagree, or want a re-run?

Built on 2026-10-15 from committed receipts.