Repository navigation
Commit 672b214
authored
Standardize the Self-selected Top50 benchmark (#72)
* feat: enforce per-rule evidence budgets
* feat: run self-selected Top50 episodes
* feat: validate and score Top50 contracts
* feat: freeze Top50 evidence contract
* feat: package high-information observations (#73)
* test: retain successful command summaries
* refactor: simplify benchmark protocol and triage prompt
* refactor: converge on single benchmark protocol
* refactor: fold API workflow into run-benchmark1 parent 6f1cdef commit 672b214
80 files changed
Lines changed: 5897 additions & 4991 deletions
File tree
- .agents/skills
- add-agent-harness
- agents
- references
- run-api-benchmark
- agents
- run-benchmark
- agents
- references
- scripts
- run-cli-benchmark
- agents
- references
- submit-benchmark-result
- .claude/skills
- .github
- scripts
- workflows
- benchmark
- docs
- tests
- docker
- intake/cloudflare-worker
- src
- test
- site
Some content is hidden
Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.
This file was deleted.
This file was deleted.
This file was deleted.
0 commit comments