Skip to content

feat(results): rename result directory row field#1540

Merged
christso merged 1 commit into
mainfrom
result-dir-rename
Jun 27, 2026
Merged

feat(results): rename result directory row field#1540
christso merged 1 commit into
mainfrom
result-dir-rename

Conversation

@christso

Copy link
Copy Markdown
Collaborator

Summary

  • Renames the per-row run manifest directory field from artifact_dir to result_dir across core artifact writers, CLI result consumers, Dashboard API/query routing, and docs.
  • Updates detail disambiguation to use the result_dir query parameter and Dashboard internal props to use resultDir.
  • Keeps broader artifact-directory terminology only for the full run artifact workspace / generic sidecars; artifact_pointers remains unchanged.
  • Removes artifact_dir compatibility from new source, tests, HTTP/query surfaces, and public contract docs.

Beads / Handoff

  • Beads: av-504, av-504.2
  • Handoff issue: EntityProcess/agentv-beads#16

Verification

  • bun test apps/cli/test/commands/eval/artifact-writer.test.ts
  • bun test apps/cli/test/commands/results/export.test.ts
  • bun test apps/cli/test/commands/results/combine.test.ts
  • bun test apps/cli/test/commands/runs/rerun.test.ts
  • bun test apps/cli/test/commands/results/serve.test.ts
  • bun test apps/cli/test/commands/results/validate.test.ts
  • bun test apps/dashboard/src/lib/api.test.ts
  • bun test packages/core/test/evaluation/evaluate-programmatic-api.test.ts
  • bun --filter @agentv/core typecheck
  • bun --filter agentv typecheck
  • bun --cwd apps/dashboard tsc -b
  • git diff --name-only -z | xargs -0 bunx biome check

Dependency / Merge Notes

  • Started from origin/main at 3f2ddc56.
  • Local origin fetch URL is malformed in this worktree (https://github.com/..git), so I fetched latest main via explicit URL without rewriting remote config. Push URL is correct.
  • This changes artifact-layout/result contract surfaces. Per .agents/verification.md, do not merge until live-provider and real-LLM-grader dogfood evidence is completed and published privately as required.

@cloudflare-workers-and-pages

Copy link
Copy Markdown

Deploying agentv with  Cloudflare Pages  Cloudflare Pages

Latest commit: fa45431
Status: ✅  Deploy successful!
Preview URL: https://180992c4.agentv.pages.dev
Branch Preview URL: https://result-dir-rename.agentv.pages.dev

View logs

@christso

Copy link
Copy Markdown
Collaborator Author

Read-only review: no findings

I reviewed PR #1540 against main using the remote PR diff/head ref (fa45431485ba621cb63c20258f80f8ac2ccefe37). I did not find any blocking bugs or wire-contract regressions.

Coverage notes:

  • Checked the hard public rename from artifact_dir to result_dir across the changed core writer/schema, CLI result consumers, Dashboard API/query routing, and docs.
  • git grep -n "artifact_dir" refs/review/pr-1540-head -- . returned no hits on the PR head, so I did not find emitted artifacts, HTTP responses, Dashboard query params, or docs retaining the old public key.
  • Confirmed remaining broader artifact-directory terminology is generic/local naming, not a public artifact_dir contract leak.
  • Confirmed artifact_pointers remains named artifact_pointers and was not conflated with the result directory rename.
  • git diff --check refs/review/pr-1540-base refs/review/pr-1540-head passed.

Residual risk / missing verification:

  • I did not run the PR test suite locally because this worktree is still at base 3f2ddc56...; I reviewed the fetched PR refs read-only without checking out or editing files.
  • I found no direct serve/dashboard regression test that asserts duplicate eval disambiguation via ?result_dir= or rejection/non-use of the old ?artifact_dir= query key. The implementation looks consistent, but that specific public query-path contract is not directly locked by tests I inspected.
  • Per .agents/verification.md, live-provider + real-LLM-grader dogfood remains required before merge because this changes artifact/result contract surfaces.

@christso

Copy link
Copy Markdown
Collaborator Author

Dogfood closeout for av-504.2 / result_dir rename:

  • Live provider + real LLM-grader dogfood passed on branch result-dir-rename at fa454314.
  • Command used bun apps/cli/src/cli.ts eval run with dogfood-agent and dogfood-grader, both OpenAI-compatible live targets via .env references (OPENAI_ENDPOINT, OPENAI_API_KEY, OPENAI_MODEL); no mock, dry-run, transcript replay, or deterministic-only grading path.
  • Result: 1/1 pass, score 100%, run bundle .agentv/results/av-504-2-result-dir/20260627T085958Z.
  • Contract checks: generated index.jsonl row has top-level result_dir; structural check confirms no top-level artifact_dir; Dashboard detail API accepts/selects with ?result_dir=, rejects invalid result_dir, and legacy artifact_dir has no selector effect.
  • Private evidence: EntityProcess/agentv-private branch evidence/av-504-2-result-dir, root commit 866c7cb.

GitHub Actions are still green and the Workmux review had no blocking findings.

@christso christso marked this pull request as ready for review June 27, 2026 09:05
@christso christso merged commit c053a9e into main Jun 27, 2026
8 checks passed
@christso christso deleted the result-dir-rename branch June 27, 2026 09:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant