feat(eval): add agent-based iterative format optimization framework and history by gspencergoog · Pull Request #7 · gspencergoog/A2UI

feat(eval): add agent-based iterative format optimization framework and history

gspencergoog

committed

refactor(eval): apply code review recommendations for parallel log dump loading and sys.path resilience

gspencergoog

committed

fix(eval): address subagent code review findings for dict normalization, path resolution, and recursion safety

gspencergoog

committed

fix(eval): address second-pass review comments for metric keys, slug deduplication, and safe dict access

gspencergoog

committed

fix(eval): address third-pass review findings for dict metrics fallback, archiver log fallback, and sync race conditions

gspencergoog

committed

fix(eval): address fourth-pass review findings for sync scanning, baseline dict normalization, and sys.path

gspencergoog

committed

fix(eval): address fifth-pass review findings for run re-indexing, archiver metrics fallback, and transport format support

gspencergoog

committed

fix(eval): address sixth-pass review findings for None metric conversions, scores/usage guards, and mtime sorting

gspencergoog

committed

fix(eval): address seventh-pass review findings for null fallback guards, None coalescing, and glob de-duplication

gspencergoog

committed

fix(eval): address eighth-pass review findings for active_samples filtering, S_opt zero fallbacks, and PID log dir

gspencergoog

committed

fix(eval): address ninth-pass review findings for PID temp log dir resolution, results.json archiving, and run_meta metric extraction

gspencergoog

committed

fix(eval): address tenth-pass review findings for dict/list s_scores, samples dict normalization, and mtime eval file sorting

gspencergoog

committed

fix(eval): address eleventh-pass review findings for meta_data uninitialized guard, module level sync import, and failure explanation list scores

gspencergoog

committed

fix(eval): address twelfth-pass review findings for list score extraction, non-none metric resolution, utf-8 subprocess, and baseexception cleanup

gspencergoog

committed

fix(eval): address thirteenth-pass review findings for null dict key guards, reporter metric accuracy fallback, and worktree history path filtering

gspencergoog

committed

fix(eval): address fourteenth-pass review findings for scalar accuracy dict guard, judging_val N/A check, and archiver results.json handling

gspencergoog

committed

fix(ci): format reporter.py and declare extra-build-dependencies for a2ui-agent-sdk in workspace

gspencergoog

committed

docs(eval): add comprehensive code documentation and README guides for iterative optimization scripts

gspencergoog

committed

feat(eval): convert format iteration workflow into portable inference-format-optimizer skill

gspencergoog

committed

docs(eval): update subagent protocol lifecycle management and sync_history root scope

gspencergoog

committed

Remove redundant tests

gspencergoog

committed

Remove express history

gspencergoog

committed

fix(eval): update script path resolution in skill tests and ignore evaluation history in prettier

gspencergoog

committed

style(eval): apply pyink formatting to skill scripts

gspencergoog

committed

feat(eval): update all-format baselines and framework evaluation scripts

gspencergoog

committed

style(eval): apply pyink formatting to python scripts

gspencergoog

committed

feat(eval): update format baselines with 5-epoch rerun metrics and process timeout fixes

gspencergoog

committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(eval): add agent-based iterative format optimization framework and history - #7

feat(eval): add agent-based iterative format optimization framework and history#7
gspencergoog wants to merge 27 commits into
mainfrom
iterative_optimization

Commits on Jul 24, 2026

feat(eval): add agent-based iterative format optimization framework and history

refactor(eval): apply code review recommendations for parallel log dump loading and sys.path resilience

fix(eval): address subagent code review findings for dict normalization, path resolution, and recursion safety

fix(eval): address second-pass review comments for metric keys, slug deduplication, and safe dict access

fix(eval): address third-pass review findings for dict metrics fallback, archiver log fallback, and sync race conditions

fix(eval): address fourth-pass review findings for sync scanning, baseline dict normalization, and sys.path

fix(eval): address fifth-pass review findings for run re-indexing, archiver metrics fallback, and transport format support

fix(eval): address sixth-pass review findings for None metric conversions, scores/usage guards, and mtime sorting

fix(eval): address seventh-pass review findings for null fallback guards, None coalescing, and glob de-duplication

fix(eval): address eighth-pass review findings for active_samples filtering, S_opt zero fallbacks, and PID log dir

fix(eval): address ninth-pass review findings for PID temp log dir resolution, results.json archiving, and run_meta metric extraction

fix(eval): address tenth-pass review findings for dict/list s_scores, samples dict normalization, and mtime eval file sorting

fix(eval): address eleventh-pass review findings for meta_data uninitialized guard, module level sync import, and failure explanation list scores

fix(eval): address twelfth-pass review findings for list score extraction, non-none metric resolution, utf-8 subprocess, and baseexception cleanup

fix(eval): address thirteenth-pass review findings for null dict key guards, reporter metric accuracy fallback, and worktree history path filtering

fix(eval): address fourteenth-pass review findings for scalar accuracy dict guard, judging_val N/A check, and archiver results.json handling

fix(ci): format reporter.py and declare extra-build-dependencies for a2ui-agent-sdk in workspace

docs(eval): add comprehensive code documentation and README guides for iterative optimization scripts

feat(eval): convert format iteration workflow into portable inference-format-optimizer skill

docs(eval): update subagent protocol lifecycle management and sync_history root scope

Remove redundant tests

Remove express history

fix(eval): update script path resolution in skill tests and ignore evaluation history in prettier

style(eval): apply pyink formatting to skill scripts

feat(eval): update all-format baselines and framework evaluation scripts

style(eval): apply pyink formatting to python scripts

feat(eval): update format baselines with 5-epoch rerun metrics and process timeout fixes