Repository navigation
apr code eval E5: apr code driver — agentic loop + edit-format adapters #5286
Description
Drive apr code against a task: give it the problem statement and repo snapshot, a bounded turn and token budget, and tools inside the sandbox only. Parse its edits through the whole-file, search/replace and unified-diff adapters, and count malformed edits as format failures. Aider-style: two attempts, with the visible test output fed back after the first.
Acceptance: a fixed-seed run of the smoke suite produces a receipt byte-identical across two runs.
Part of #5281. Spec: docs/specifications/apr-code-eval-v1.md (APR-CODE-EVAL-001).
keep-open: child of #5281
Activity
feat(code-eval): agent driver for the apr code eval (E5, part 2)
noahgift commentedon Oct 14, 2026
merged into #5314 (master ticket for epic #4000; triage rule 7b, 2026-10-14). It is a checklist row there; reopen this issue when work on it starts or its PR opens.
noahgift commentedon Oct 14, 2026
Moved: this sub-ticket is now a row in #5342 (apr code eval: task suite, sandbox, metrics and nightly receipts (APR-CODE-EVAL-001)), split out of #5314 by theme.
feat(code-eval): spec §7 run receipt for the apr code eval (E5, part 3)
feat(code-eval): sandbox-backed test runner (E5, part 3b)
feat(code-eval): suite validation, evaluate and the prompt (E5, part 4)
feat(apr-cli): apr code eval, the receipted eval of apr code (E5, par…