Skip to content

[aw] Make CI KBE issue search deterministic #133684

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
vitek-karas wants to merge 4 commits into
base: main
Choose a base branch
from
+922 −50
Open

[aw] Make CI KBE issue search deterministic #133684

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
3c7220b
Add trusted KBE search eval support
vitek-karas Sep 15, 2026
8557d11
Make CI KBE issue search deterministic
vitek-karas Sep 15, 2026
d16a45e
Fix KBE eval grader integration
vitek-karas Sep 17, 2026
745c27f
Test production KBE search validation
vitek-karas Sep 17, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
51 changes: 47 additions & 4 deletions .github/workflows/ci-eval.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -231,7 +231,7 @@ jobs:
with:
persist-credentials: false

- name: Preserve the trusted eval specs
- name: Preserve the trusted eval inputs
if: steps.guard.outputs.run == 'true'
env:
SPECS: ${{ steps.sel.outputs.specs }}
Expand All @@ -240,6 +240,18 @@ jobs:
for spec in $SPECS; do
cp "$spec" "$RUNNER_TEMP/trusted-evals/$(basename "$spec")"
done
if test -f .github/workflows/evals/search-kbe-issues.cjs; then
cp .github/workflows/evals/search-kbe-issues.cjs \
"$RUNNER_TEMP/trusted-evals/search-kbe-issues.cjs"
fi
if test -f .github/workflows/evals/kbe-candidate-reads-grader.mjs; then
cp .github/workflows/evals/kbe-candidate-reads-grader.mjs \
"$RUNNER_TEMP/trusted-evals/kbe-candidate-reads-grader.mjs"
fi
if test -f .github/workflows/evals/kbe-search.test.mjs; then
cp .github/workflows/evals/kbe-search.test.mjs \
"$RUNNER_TEMP/trusted-evals/kbe-search.test.mjs"
fi
if [ -f .github/workflows/shared/filter-scanner-kbes.sh ]; then
cp .github/workflows/shared/filter-scanner-kbes.sh \
"$RUNNER_TEMP/trusted-evals/filter-scanner-kbes.sh"
Expand All @@ -265,7 +277,7 @@ jobs:
GH_TOKEN: ${{ github.token }}
run: gh pr checkout ${{ github.event.issue.number }}

- name: Restore the trusted eval specs
- name: Restore the trusted eval inputs
if: steps.guard.outputs.run == 'true'
env:
SPECS: ${{ steps.sel.outputs.specs }}
Expand All @@ -274,19 +286,45 @@ jobs:
rm -f -- "$spec"
cp -- "$RUNNER_TEMP/trusted-evals/$(basename "$spec")" "$spec"
done
rm -f -- .github/workflows/evals/search-kbe-issues.cjs
if test -f "$RUNNER_TEMP/trusted-evals/search-kbe-issues.cjs"; then
cp -- "$RUNNER_TEMP/trusted-evals/search-kbe-issues.cjs" \
.github/workflows/evals/search-kbe-issues.cjs
fi
Comment thread
vitek-karas marked this conversation as resolved.
rm -f -- .github/workflows/evals/kbe-candidate-reads-grader.mjs
if test -f "$RUNNER_TEMP/trusted-evals/kbe-candidate-reads-grader.mjs"; then
cp -- "$RUNNER_TEMP/trusted-evals/kbe-candidate-reads-grader.mjs" \
.github/workflows/evals/kbe-candidate-reads-grader.mjs
fi
rm -f -- .github/workflows/evals/kbe-search.test.mjs
if test -f "$RUNNER_TEMP/trusted-evals/kbe-search.test.mjs"; then
cp -- "$RUNNER_TEMP/trusted-evals/kbe-search.test.mjs" \
.github/workflows/evals/kbe-search.test.mjs
fi
if [ -f "$RUNNER_TEMP/trusted-evals/filter-scanner-kbes.sh" ]; then
mkdir -p .github/workflows/shared
cp -- "$RUNNER_TEMP/trusted-evals/filter-scanner-kbes.sh" \
.github/workflows/shared/filter-scanner-kbes.sh
fi

- name: Test the KBE search support
if: steps.guard.outputs.run == 'true'
run: |
if test -f .github/workflows/evals/kbe-search.test.mjs; then
node --test .github/workflows/evals/kbe-search.test.mjs
fi

- name: Lint the selected spec(s)
if: steps.guard.outputs.run == 'true'
env:
SPECS: ${{ steps.sel.outputs.specs }}
run: |
grader_args=()
if test -f .github/workflows/evals/kbe-candidate-reads-grader.mjs; then
grader_args+=(--grader-plugin ./kbe-candidate-reads-grader.mjs)
fi
for spec in $SPECS; do
vally lint --eval-spec "$spec" --strict
vally lint --eval-spec "$spec" --strict "${grader_args[@]}"
done

- name: Run the eval(s)
Expand Down Expand Up @@ -317,11 +355,16 @@ jobs:
report="$workdir/report.md"
: > "$report"
rc=0
grader_args=()
if test -f .github/workflows/evals/kbe-candidate-reads-grader.mjs; then
grader_args+=(--grader-plugin ./kbe-candidate-reads-grader.mjs)
fi
for spec in $SPECS; do
name="$(basename "$spec" .eval.yaml)"
echo "## $name" >> "$report"
if vally eval --eval-spec "$spec" --skill-dir .github/workflows \
--workspace "$workdir/ws-$name" --output-dir "$workdir/out-$name" > "$workdir/log-$name.txt" 2>&1; then
--workspace "$workdir/ws-$name" --output-dir "$workdir/out-$name" \
"${grader_args[@]}" > "$workdir/log-$name.txt" 2>&1; then
echo "✅ passed" >> "$report"
else
rc=1
Expand Down
Loading