mirror of
https://github.com/adobe/skills.git
synced 2026-09-19 08:06:09 +08:00
0e7d9df830
Agent-Id: agent-abcf7901-1e84-43e3-a806-c3d7471124d5 Linked-Note-Id: ae22a77c-e673-4782-af85-f7daa999effa
180 lines
6.2 KiB
YAML
180 lines
6.2 KiB
YAML
name: Tessl Skill Eval
|
|
|
|
on:
|
|
pull_request:
|
|
branches: [main]
|
|
|
|
jobs:
|
|
tessl-eval:
|
|
runs-on: ubuntu-latest
|
|
# Uses the "eval" environment for protection rules (e.g. reviewer approval)
|
|
# to guard the TESSL_TOKEN secret against exfiltration via same-repo branches.
|
|
environment: eval
|
|
steps:
|
|
- uses: actions/checkout@v6
|
|
with:
|
|
fetch-depth: 0
|
|
ref: ${{ github.event.pull_request.head.sha }}
|
|
|
|
- name: Check eval trigger
|
|
id: trigger
|
|
run: |
|
|
LAST_MSG=$(git log -1 --format=%s)
|
|
echo "commit_message=$LAST_MSG"
|
|
if [[ ! "$LAST_MSG" =~ ^eval: ]]; then
|
|
echo "skip=true" >> "$GITHUB_OUTPUT"
|
|
echo "Last commit does not start with 'eval:' — skipping eval."
|
|
else
|
|
echo "skip=false" >> "$GITHUB_OUTPUT"
|
|
echo "Eval triggered by commit: $LAST_MSG"
|
|
fi
|
|
|
|
- name: Post skip summary (no eval trigger)
|
|
if: steps.trigger.outputs.skip == 'true'
|
|
run: |
|
|
{
|
|
echo "## Skill Eval"
|
|
echo ""
|
|
echo "Eval not requested — last commit message does not start with \`eval:\`."
|
|
echo "To trigger evals, create a commit with a message like:"
|
|
echo "\`\`\`"
|
|
echo 'git commit --allow-empty -m "eval: test building-blocks changes"'
|
|
echo "\`\`\`"
|
|
} >> "$GITHUB_STEP_SUMMARY"
|
|
|
|
- name: Detect changed skills with tiles
|
|
if: steps.trigger.outputs.skip != 'true'
|
|
id: detect
|
|
run: |
|
|
CHANGED_FILES=$(git diff --name-only origin/${{ github.base_ref }}...HEAD)
|
|
|
|
# Build list of all tile directories (parents of tile.json)
|
|
ALL_TILE_DIRS=$(find skills -name tile.json -exec dirname {} \; 2>/dev/null | sort)
|
|
|
|
if [ -z "$ALL_TILE_DIRS" ]; then
|
|
echo "skip=true" >> "$GITHUB_OUTPUT"
|
|
echo "No tile.json files found in repo."
|
|
else
|
|
# For each changed file, find which tile directory it belongs to
|
|
DIRS=""
|
|
for changed_file in $CHANGED_FILES; do
|
|
for tile_dir in $ALL_TILE_DIRS; do
|
|
case "$changed_file" in
|
|
"$tile_dir"/*)
|
|
# Check not already added
|
|
case " $DIRS " in
|
|
*" $tile_dir "*) ;;
|
|
*) DIRS="$DIRS $tile_dir" ;;
|
|
esac
|
|
break
|
|
;;
|
|
esac
|
|
done
|
|
done
|
|
|
|
DIRS=$(echo "$DIRS" | xargs) # trim whitespace
|
|
|
|
# Check for unmatched changes under skills/
|
|
UNMATCHED=false
|
|
for changed_file in $CHANGED_FILES; do
|
|
case "$changed_file" in
|
|
skills/*)
|
|
MATCHED=false
|
|
for tile_dir in $ALL_TILE_DIRS; do
|
|
case "$changed_file" in
|
|
"$tile_dir"/*) MATCHED=true; break ;;
|
|
esac
|
|
done
|
|
if [ "$MATCHED" = "false" ]; then
|
|
UNMATCHED=true
|
|
break
|
|
fi
|
|
;;
|
|
esac
|
|
done
|
|
|
|
if [ "$UNMATCHED" = "true" ]; then
|
|
echo "Unmatched changes under skills/ detected — evaluating all tiles"
|
|
DIRS=$(find skills -name tile.json -exec dirname {} \; 2>/dev/null | tr "\n" " " | xargs)
|
|
fi
|
|
|
|
if [ -z "$DIRS" ]; then
|
|
echo "skip=true" >> "$GITHUB_OUTPUT"
|
|
echo "Changed files don't belong to any tile — nothing to eval."
|
|
else
|
|
echo "skip=false" >> "$GITHUB_OUTPUT"
|
|
echo "dirs=$DIRS" >> "$GITHUB_OUTPUT"
|
|
echo "Changed tiles: $DIRS"
|
|
fi
|
|
fi
|
|
|
|
- name: Post skip summary (no qualifying tiles)
|
|
if: steps.trigger.outputs.skip != 'true' && steps.detect.outputs.skip == 'true'
|
|
run: |
|
|
{
|
|
echo "## Skill Eval"
|
|
echo ""
|
|
echo "No qualifying skills to evaluate — either no skill files were changed or changed skills have no \`tile.json\`."
|
|
} >> "$GITHUB_STEP_SUMMARY"
|
|
|
|
- uses: tesslio/setup-tessl@v2
|
|
if: steps.trigger.outputs.skip != 'true' && steps.detect.outputs.skip != 'true'
|
|
with:
|
|
token: ${{ secrets.TESSL_TOKEN }}
|
|
|
|
- name: Run evals
|
|
if: steps.trigger.outputs.skip != 'true' && steps.detect.outputs.skip != 'true'
|
|
run: |
|
|
PASS=0
|
|
FAIL=0
|
|
SUMMARY_ROWS=""
|
|
|
|
for tile_dir in ${{ steps.detect.outputs.dirs }}; do
|
|
TILE_NAME=$(basename "$tile_dir")
|
|
echo "::group::Evaluating $TILE_NAME ($tile_dir)"
|
|
|
|
EXIT_CODE=0
|
|
OUTPUT=$(tessl eval run "$tile_dir" 2>&1) || EXIT_CODE=$?
|
|
echo "$OUTPUT"
|
|
echo "::endgroup::"
|
|
|
|
if [ "$EXIT_CODE" -ne 0 ]; then
|
|
echo "::warning::tessl eval run failed for $TILE_NAME (exit code $EXIT_CODE)"
|
|
FAIL=$((FAIL + 1))
|
|
SUMMARY_ROWS="$SUMMARY_ROWS| $TILE_NAME | error | ❌ |\n"
|
|
else
|
|
PASS=$((PASS + 1))
|
|
SUMMARY_ROWS="$SUMMARY_ROWS| $TILE_NAME | passed | ✅ |\n"
|
|
# Show detailed results only for successful eval
|
|
echo "::group::Eval results for $TILE_NAME"
|
|
tessl eval view --last 2>&1 || true
|
|
echo "::endgroup::"
|
|
fi
|
|
done
|
|
|
|
TOTAL=$((PASS + FAIL))
|
|
|
|
{
|
|
echo "## Skill Eval"
|
|
echo ""
|
|
echo "| Tile | Result | Status |"
|
|
echo "|------|--------|--------|"
|
|
echo -e "$SUMMARY_ROWS"
|
|
echo "| **Total** | **$PASS/$TOTAL passed** | $([ "$FAIL" -eq 0 ] && echo '✅' || echo '❌') |"
|
|
} >> "$GITHUB_STEP_SUMMARY"
|
|
|
|
echo ""
|
|
echo "============================="
|
|
echo " Skill Eval Summary"
|
|
echo "============================="
|
|
echo " Total: $TOTAL"
|
|
echo " Passed: $PASS"
|
|
echo " Failed: $FAIL"
|
|
echo "============================="
|
|
|
|
if [ "$FAIL" -gt 0 ]; then
|
|
echo "::error::$FAIL tile(s) failed evaluation"
|
|
exit 1
|
|
fi
|
|
|