feat(reports): --report-md and an automatic GitHub step summary by Chemaclass · Pull Request #1043 · TypedDevs/bashunit

Chemaclass · 2026-08-10T18:36:20Z

🤔 Background

Related #1015

Every existing report format targets a machine, and the HTML one has to be uploaded as an artifact and downloaded to be read. Nothing targeted the page a developer looks at first: the pull request.

💡 Changes

--report-md <file> writes a Markdown summary: verdict, counts table, and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists.
Inside GitHub Actions there is nothing to configure: with GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended to it and renders on the job page. Appended, never written, since that file belongs to the whole job.
Failure messages are fenced and ANSI-stripped; test names are escaped so |, *, _ or a backtick cannot break the table.
Only the outermost run writes the step summary, matching the annotations: the variable is inherited by every child, so a nested run would otherwise append its fixtures' results to the parent's job page.

🩹 Follow-up fix in this branch

The coverage percentage in the summary was read before the hit records were finalized, and before a --parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a lower-than-console figure sequentially. The coverage numbers are now computed before the summary is written; the console and file reports still render where they did. Covered by test_the_coverage_percentage_matches_the_console_report.

✅ Build & Verify

The earlier block on this PR is gone: the standalone size budget was raised to 544 KiB on main in #1046. The artifact built from this branch is 523306 bytes, inside the budget.

Every existing report format targets a machine, and the HTML one has to be uploaded and downloaded to be read. Nothing targeted the page a developer actually looks at first. The summary carries the verdict, a counts table and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists. Failure messages go inside a fence with ANSI stripped so they render verbatim, and test names are escaped so a name containing a pipe, asterisk, underscore or backtick cannot break the table. With GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended there. Appended, never written: that file belongs to the whole job, so truncating it would discard the other steps' output. Only the outermost run writes it, for the same reason the annotations do -- the variable is inherited by every child, so a nested run would append its own fixtures' results to the parent's job page. Generation moved ahead of print_profile_and_reset, which deletes the profile records the summary reads, and load_spooled moved with it so a --parallel run still reports the rows its workers spooled (#1004). Closes#1015

GITHUB_STEP_SUMMARY is set on every GitHub Actions runner, which makes the Markdown summary a third reason to collect rows. These tests ask whether a file report was configured, so the ambient CI environment has to be pinned off for the answer to mean anything.

…nsole nor parallel The summary read the coverage percentage before the hit records were finalized, and before a parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a low figure sequentially. Compute the coverage numbers before the summary is written; the reports still render where they did.

Chemaclass · 2026-08-10T18:36:20Z

🤔 Background

Related #1015

Every existing report format targets a machine, and the HTML one has to be uploaded as an artifact and downloaded to be read. Nothing targeted the page a developer looks at first: the pull request.

💡 Changes

--report-md <file> writes a Markdown summary: verdict, counts table, and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists.
Inside GitHub Actions there is nothing to configure: with GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended to it and renders on the job page. Appended, never written, since that file belongs to the whole job.
Failure messages are fenced and ANSI-stripped; test names are escaped so |, *, _ or a backtick cannot break the table.
Only the outermost run writes the step summary, matching the annotations: the variable is inherited by every child, so a nested run would otherwise append its fixtures' results to the parent's job page.

🩹 Follow-up fix in this branch

The coverage percentage in the summary was read before the hit records were finalized, and before a --parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a lower-than-console figure sequentially. The coverage numbers are now computed before the summary is written; the console and file reports still render where they did. Covered by test_the_coverage_percentage_matches_the_console_report.

✅ Build & Verify

The earlier block on this PR is gone: the standalone size budget was raised to 544 KiB on main in #1046. The artifact built from this branch is 523306 bytes, inside the budget.

Every existing report format targets a machine, and the HTML one has to be uploaded and downloaded to be read. Nothing targeted the page a developer actually looks at first. The summary carries the verdict, a counts table and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists. Failure messages go inside a fence with ANSI stripped so they render verbatim, and test names are escaped so a name containing a pipe, asterisk, underscore or backtick cannot break the table. With GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended there. Appended, never written: that file belongs to the whole job, so truncating it would discard the other steps' output. Only the outermost run writes it, for the same reason the annotations do -- the variable is inherited by every child, so a nested run would append its own fixtures' results to the parent's job page. Generation moved ahead of print_profile_and_reset, which deletes the profile records the summary reads, and load_spooled moved with it so a --parallel run still reports the rows its workers spooled (#1004). Closes#1015

GITHUB_STEP_SUMMARY is set on every GitHub Actions runner, which makes the Markdown summary a third reason to collect rows. These tests ask whether a file report was configured, so the ambient CI environment has to be pinned off for the answer to mean anything.

…nsole nor parallel The summary read the coverage percentage before the hit records were finalized, and before a parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a low figure sequentially. Compute the coverage numbers before the summary is written; the reports still render where they did.

Chemaclass · 2026-08-10T18:36:20Z

🤔 Background

Related #1015

Every existing report format targets a machine, and the HTML one has to be uploaded as an artifact and downloaded to be read. Nothing targeted the page a developer looks at first: the pull request.

💡 Changes

--report-md <file> writes a Markdown summary: verdict, counts table, and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists.
Inside GitHub Actions there is nothing to configure: with GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended to it and renders on the job page. Appended, never written, since that file belongs to the whole job.
Failure messages are fenced and ANSI-stripped; test names are escaped so |, *, _ or a backtick cannot break the table.
Only the outermost run writes the step summary, matching the annotations: the variable is inherited by every child, so a nested run would otherwise append its fixtures' results to the parent's job page.

🩹 Follow-up fix in this branch

The coverage percentage in the summary was read before the hit records were finalized, and before a --parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a lower-than-console figure sequentially. The coverage numbers are now computed before the summary is written; the console and file reports still render where they did. Covered by test_the_coverage_percentage_matches_the_console_report.

✅ Build & Verify

The earlier block on this PR is gone: the standalone size budget was raised to 544 KiB on main in #1046. The artifact built from this branch is 523306 bytes, inside the budget.

Every existing report format targets a machine, and the HTML one has to be uploaded and downloaded to be read. Nothing targeted the page a developer actually looks at first. The summary carries the verdict, a counts table and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists. Failure messages go inside a fence with ANSI stripped so they render verbatim, and test names are escaped so a name containing a pipe, asterisk, underscore or backtick cannot break the table. With GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended there. Appended, never written: that file belongs to the whole job, so truncating it would discard the other steps' output. Only the outermost run writes it, for the same reason the annotations do -- the variable is inherited by every child, so a nested run would append its own fixtures' results to the parent's job page. Generation moved ahead of print_profile_and_reset, which deletes the profile records the summary reads, and load_spooled moved with it so a --parallel run still reports the rows its workers spooled (#1004). Closes#1015

GITHUB_STEP_SUMMARY is set on every GitHub Actions runner, which makes the Markdown summary a third reason to collect rows. These tests ask whether a file report was configured, so the ambient CI environment has to be pinned off for the answer to mean anything.

…nsole nor parallel The summary read the coverage percentage before the hit records were finalized, and before a parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a low figure sequentially. Compute the coverage numbers before the summary is written; the reports still render where they did.

Chemaclass · 2026-08-10T18:36:20Z

🤔 Background

Related #1015

Every existing report format targets a machine, and the HTML one has to be uploaded as an artifact and downloaded to be read. Nothing targeted the page a developer looks at first: the pull request.

💡 Changes

--report-md <file> writes a Markdown summary: verdict, counts table, and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists.
Inside GitHub Actions there is nothing to configure: with GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended to it and renders on the job page. Appended, never written, since that file belongs to the whole job.
Failure messages are fenced and ANSI-stripped; test names are escaped so |, *, _ or a backtick cannot break the table.
Only the outermost run writes the step summary, matching the annotations: the variable is inherited by every child, so a nested run would otherwise append its fixtures' results to the parent's job page.

🩹 Follow-up fix in this branch

The coverage percentage in the summary was read before the hit records were finalized, and before a --parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a lower-than-console figure sequentially. The coverage numbers are now computed before the summary is written; the console and file reports still render where they did. Covered by test_the_coverage_percentage_matches_the_console_report.

✅ Build & Verify

The earlier block on this PR is gone: the standalone size budget was raised to 544 KiB on main in #1046. The artifact built from this branch is 523306 bytes, inside the budget.

Every existing report format targets a machine, and the HTML one has to be uploaded and downloaded to be read. Nothing targeted the page a developer actually looks at first. The summary carries the verdict, a counts table and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists. Failure messages go inside a fence with ANSI stripped so they render verbatim, and test names are escaped so a name containing a pipe, asterisk, underscore or backtick cannot break the table. With GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended there. Appended, never written: that file belongs to the whole job, so truncating it would discard the other steps' output. Only the outermost run writes it, for the same reason the annotations do -- the variable is inherited by every child, so a nested run would append its own fixtures' results to the parent's job page. Generation moved ahead of print_profile_and_reset, which deletes the profile records the summary reads, and load_spooled moved with it so a --parallel run still reports the rows its workers spooled (#1004). Closes#1015

GITHUB_STEP_SUMMARY is set on every GitHub Actions runner, which makes the Markdown summary a third reason to collect rows. These tests ask whether a file report was configured, so the ambient CI environment has to be pinned off for the answer to mean anything.

…nsole nor parallel The summary read the coverage percentage before the hit records were finalized, and before a parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a low figure sequentially. Compute the coverage numbers before the summary is written; the reports still render where they did.

Chemaclass · 2026-08-10T18:36:20Z

🤔 Background

Related #1015

Every existing report format targets a machine, and the HTML one has to be uploaded as an artifact and downloaded to be read. Nothing targeted the page a developer looks at first: the pull request.

💡 Changes

--report-md <file> writes a Markdown summary: verdict, counts table, and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists.
Inside GitHub Actions there is nothing to configure: with GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended to it and renders on the job page. Appended, never written, since that file belongs to the whole job.
Failure messages are fenced and ANSI-stripped; test names are escaped so |, *, _ or a backtick cannot break the table.
Only the outermost run writes the step summary, matching the annotations: the variable is inherited by every child, so a nested run would otherwise append its fixtures' results to the parent's job page.

🩹 Follow-up fix in this branch

The coverage percentage in the summary was read before the hit records were finalized, and before a --parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a lower-than-console figure sequentially. The coverage numbers are now computed before the summary is written; the console and file reports still render where they did. Covered by test_the_coverage_percentage_matches_the_console_report.

✅ Build & Verify

The earlier block on this PR is gone: the standalone size budget was raised to 544 KiB on main in #1046. The artifact built from this branch is 523306 bytes, inside the budget.

Every existing report format targets a machine, and the HTML one has to be uploaded and downloaded to be read. Nothing targeted the page a developer actually looks at first. The summary carries the verdict, a counts table and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists. Failure messages go inside a fence with ANSI stripped so they render verbatim, and test names are escaped so a name containing a pipe, asterisk, underscore or backtick cannot break the table. With GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended there. Appended, never written: that file belongs to the whole job, so truncating it would discard the other steps' output. Only the outermost run writes it, for the same reason the annotations do -- the variable is inherited by every child, so a nested run would append its own fixtures' results to the parent's job page. Generation moved ahead of print_profile_and_reset, which deletes the profile records the summary reads, and load_spooled moved with it so a --parallel run still reports the rows its workers spooled (#1004). Closes#1015

GITHUB_STEP_SUMMARY is set on every GitHub Actions runner, which makes the Markdown summary a third reason to collect rows. These tests ask whether a file report was configured, so the ambient CI environment has to be pinned off for the answer to mean anything.

…nsole nor parallel The summary read the coverage percentage before the hit records were finalized, and before a parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a low figure sequentially. Compute the coverage numbers before the summary is written; the reports still render where they did.

Chemaclass · 2026-08-10T18:36:20Z

🤔 Background

Related #1015

Every existing report format targets a machine, and the HTML one has to be uploaded as an artifact and downloaded to be read. Nothing targeted the page a developer looks at first: the pull request.

💡 Changes

--report-md <file> writes a Markdown summary: verdict, counts table, and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists.
Inside GitHub Actions there is nothing to configure: with GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended to it and renders on the job page. Appended, never written, since that file belongs to the whole job.
Failure messages are fenced and ANSI-stripped; test names are escaped so |, *, _ or a backtick cannot break the table.
Only the outermost run writes the step summary, matching the annotations: the variable is inherited by every child, so a nested run would otherwise append its fixtures' results to the parent's job page.

🩹 Follow-up fix in this branch

The coverage percentage in the summary was read before the hit records were finalized, and before a --parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a lower-than-console figure sequentially. The coverage numbers are now computed before the summary is written; the console and file reports still render where they did. Covered by test_the_coverage_percentage_matches_the_console_report.

✅ Build & Verify

The earlier block on this PR is gone: the standalone size budget was raised to 544 KiB on main in #1046. The artifact built from this branch is 523306 bytes, inside the budget.

Every existing report format targets a machine, and the HTML one has to be uploaded and downloaded to be read. Nothing targeted the page a developer actually looks at first. The summary carries the verdict, a counts table and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists. Failure messages go inside a fence with ANSI stripped so they render verbatim, and test names are escaped so a name containing a pipe, asterisk, underscore or backtick cannot break the table. With GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended there. Appended, never written: that file belongs to the whole job, so truncating it would discard the other steps' output. Only the outermost run writes it, for the same reason the annotations do -- the variable is inherited by every child, so a nested run would append its own fixtures' results to the parent's job page. Generation moved ahead of print_profile_and_reset, which deletes the profile records the summary reads, and load_spooled moved with it so a --parallel run still reports the rows its workers spooled (#1004). Closes#1015

GITHUB_STEP_SUMMARY is set on every GitHub Actions runner, which makes the Markdown summary a third reason to collect rows. These tests ask whether a file report was configured, so the ambient CI environment has to be pinned off for the answer to mean anything.

…nsole nor parallel The summary read the coverage percentage before the hit records were finalized, and before a parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a low figure sequentially. Compute the coverage numbers before the summary is written; the reports still render where they did.

Chemaclass · 2026-08-10T18:36:20Z

🤔 Background

Related #1015

Every existing report format targets a machine, and the HTML one has to be uploaded as an artifact and downloaded to be read. Nothing targeted the page a developer looks at first: the pull request.

💡 Changes

--report-md <file> writes a Markdown summary: verdict, counts table, and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists.
Inside GitHub Actions there is nothing to configure: with GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended to it and renders on the job page. Appended, never written, since that file belongs to the whole job.
Failure messages are fenced and ANSI-stripped; test names are escaped so |, *, _ or a backtick cannot break the table.
Only the outermost run writes the step summary, matching the annotations: the variable is inherited by every child, so a nested run would otherwise append its fixtures' results to the parent's job page.

🩹 Follow-up fix in this branch

The coverage percentage in the summary was read before the hit records were finalized, and before a --parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a lower-than-console figure sequentially. The coverage numbers are now computed before the summary is written; the console and file reports still render where they did. Covered by test_the_coverage_percentage_matches_the_console_report.

✅ Build & Verify

The earlier block on this PR is gone: the standalone size budget was raised to 544 KiB on main in #1046. The artifact built from this branch is 523306 bytes, inside the budget.

Every existing report format targets a machine, and the HTML one has to be uploaded and downloaded to be read. Nothing targeted the page a developer actually looks at first. The summary carries the verdict, a counts table and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists. Failure messages go inside a fence with ANSI stripped so they render verbatim, and test names are escaped so a name containing a pipe, asterisk, underscore or backtick cannot break the table. With GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended there. Appended, never written: that file belongs to the whole job, so truncating it would discard the other steps' output. Only the outermost run writes it, for the same reason the annotations do -- the variable is inherited by every child, so a nested run would append its own fixtures' results to the parent's job page. Generation moved ahead of print_profile_and_reset, which deletes the profile records the summary reads, and load_spooled moved with it so a --parallel run still reports the rows its workers spooled (#1004). Closes#1015

GITHUB_STEP_SUMMARY is set on every GitHub Actions runner, which makes the Markdown summary a third reason to collect rows. These tests ask whether a file report was configured, so the ambient CI environment has to be pinned off for the answer to mean anything.

…nsole nor parallel The summary read the coverage percentage before the hit records were finalized, and before a parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a low figure sequentially. Compute the coverage numbers before the summary is written; the reports still render where they did.

Chemaclass · 2026-08-10T18:36:20Z

🤔 Background

Related #1015

Every existing report format targets a machine, and the HTML one has to be uploaded as an artifact and downloaded to be read. Nothing targeted the page a developer looks at first: the pull request.

💡 Changes

--report-md <file> writes a Markdown summary: verdict, counts table, and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists.
Inside GitHub Actions there is nothing to configure: with GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended to it and renders on the job page. Appended, never written, since that file belongs to the whole job.
Failure messages are fenced and ANSI-stripped; test names are escaped so |, *, _ or a backtick cannot break the table.
Only the outermost run writes the step summary, matching the annotations: the variable is inherited by every child, so a nested run would otherwise append its fixtures' results to the parent's job page.

🩹 Follow-up fix in this branch

The coverage percentage in the summary was read before the hit records were finalized, and before a --parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a lower-than-console figure sequentially. The coverage numbers are now computed before the summary is written; the console and file reports still render where they did. Covered by test_the_coverage_percentage_matches_the_console_report.

✅ Build & Verify

The earlier block on this PR is gone: the standalone size budget was raised to 544 KiB on main in #1046. The artifact built from this branch is 523306 bytes, inside the budget.

Every existing report format targets a machine, and the HTML one has to be uploaded and downloaded to be read. Nothing targeted the page a developer actually looks at first. The summary carries the verdict, a counts table and each failure with its file:line and message. Coverage and slowest-tests sections appear only when that data exists. Failure messages go inside a fence with ANSI stripped so they render verbatim, and test names are escaped so a name containing a pipe, asterisk, underscore or backtick cannot break the table. With GITHUB_STEP_SUMMARY set and no explicit path, the summary is appended there. Appended, never written: that file belongs to the whole job, so truncating it would discard the other steps' output. Only the outermost run writes it, for the same reason the annotations do -- the variable is inherited by every child, so a nested run would append its own fixtures' results to the parent's job page. Generation moved ahead of print_profile_and_reset, which deletes the profile records the summary reads, and load_spooled moved with it so a --parallel run still reports the rows its workers spooled (#1004). Closes#1015

GITHUB_STEP_SUMMARY is set on every GitHub Actions runner, which makes the Markdown summary a third reason to collect rows. These tests ask whether a file report was configured, so the ambient CI environment has to be pinned off for the answer to mean anything.

…nsole nor parallel The summary read the coverage percentage before the hit records were finalized, and before a parallel run's per-worker data was aggregated, so it reported 0% for a covered run under --parallel and a low figure sequentially. Compute the coverage numbers before the summary is written; the reports still render where they did.

Chemaclass added the enhancement New feature or request label Aug 10, 2026

Chemaclass self-assigned this Aug 10, 2026

Chemaclass added 3 commits August 10, 2026 21:33

Chemaclass force-pushed the feat/1015-report-md branch from 14d5cde to 341da2aCompare August 10, 2026 19:50

Chemaclass merged commit eab6b9a into mainAug 10, 2026
37 checks passed

Chemaclass deleted the feat/1015-report-md branch August 10, 2026 20:02

This was referenced Aug 10, 2026
feat(reports): --report-md and automatic $GITHUB_STEP_SUMMARY output #1015
Closed
feat(reports): --report-md and automatic GITHUB_STEP_SUMMARY output #1049
Closed
test(reports): keep fixture summaries off the real job page #1050
Merged

Chemaclass added the enhancement New feature or request label Aug 10, 2026

Chemaclass self-assigned this Aug 10, 2026

Chemaclass added 3 commits August 10, 2026 21:33

Chemaclass force-pushed the feat/1015-report-md branch from 14d5cde to 341da2aCompare August 10, 2026 19:50

Chemaclass merged commit eab6b9a into mainAug 10, 2026
37 checks passed

Chemaclass deleted the feat/1015-report-md branch August 10, 2026 20:02

This was referenced Aug 10, 2026
feat(reports): --report-md and automatic $GITHUB_STEP_SUMMARY output #1015
Closed
feat(reports): --report-md and automatic GITHUB_STEP_SUMMARY output #1049
Closed
test(reports): keep fixture summaries off the real job page #1050
Merged

Chemaclass added the enhancement New feature or request label Aug 10, 2026

Chemaclass self-assigned this Aug 10, 2026

Chemaclass added 3 commits August 10, 2026 21:33

Chemaclass force-pushed the feat/1015-report-md branch from 14d5cde to 341da2aCompare August 10, 2026 19:50

Chemaclass merged commit eab6b9a into mainAug 10, 2026
37 checks passed

Chemaclass deleted the feat/1015-report-md branch August 10, 2026 20:02

This was referenced Aug 10, 2026
feat(reports): --report-md and automatic $GITHUB_STEP_SUMMARY output #1015
Closed
feat(reports): --report-md and automatic GITHUB_STEP_SUMMARY output #1049
Closed
test(reports): keep fixture summaries off the real job page #1050
Merged

Chemaclass added the enhancement New feature or request label Aug 10, 2026

Chemaclass self-assigned this Aug 10, 2026

Chemaclass added 3 commits August 10, 2026 21:33

Chemaclass force-pushed the feat/1015-report-md branch from 14d5cde to 341da2aCompare August 10, 2026 19:50

Chemaclass merged commit eab6b9a into mainAug 10, 2026
37 checks passed

Chemaclass deleted the feat/1015-report-md branch August 10, 2026 20:02

This was referenced Aug 10, 2026
feat(reports): --report-md and automatic $GITHUB_STEP_SUMMARY output #1015
Closed
feat(reports): --report-md and automatic GITHUB_STEP_SUMMARY output #1049
Closed
test(reports): keep fixture summaries off the real job page #1050
Merged

Chemaclass added the enhancement New feature or request label Aug 10, 2026

Chemaclass self-assigned this Aug 10, 2026

Chemaclass added 3 commits August 10, 2026 21:33

Chemaclass force-pushed the feat/1015-report-md branch from 14d5cde to 341da2aCompare August 10, 2026 19:50

Chemaclass merged commit eab6b9a into mainAug 10, 2026
37 checks passed

Chemaclass deleted the feat/1015-report-md branch August 10, 2026 20:02

This was referenced Aug 10, 2026
feat(reports): --report-md and automatic $GITHUB_STEP_SUMMARY output #1015
Closed
feat(reports): --report-md and automatic GITHUB_STEP_SUMMARY output #1049
Closed
test(reports): keep fixture summaries off the real job page #1050
Merged

Chemaclass added the enhancement New feature or request label Aug 10, 2026

Chemaclass self-assigned this Aug 10, 2026

Chemaclass added 3 commits August 10, 2026 21:33

Chemaclass force-pushed the feat/1015-report-md branch from 14d5cde to 341da2aCompare August 10, 2026 19:50

Chemaclass merged commit eab6b9a into mainAug 10, 2026
37 checks passed

Chemaclass deleted the feat/1015-report-md branch August 10, 2026 20:02

This was referenced Aug 10, 2026
feat(reports): --report-md and automatic $GITHUB_STEP_SUMMARY output #1015
Closed
feat(reports): --report-md and automatic GITHUB_STEP_SUMMARY output #1049
Closed
test(reports): keep fixture summaries off the real job page #1050
Merged

Chemaclass added the enhancement New feature or request label Aug 10, 2026

Chemaclass self-assigned this Aug 10, 2026

Chemaclass added 3 commits August 10, 2026 21:33

Chemaclass force-pushed the feat/1015-report-md branch from 14d5cde to 341da2aCompare August 10, 2026 19:50

Chemaclass merged commit eab6b9a into mainAug 10, 2026
37 checks passed

Chemaclass deleted the feat/1015-report-md branch August 10, 2026 20:02

This was referenced Aug 10, 2026
feat(reports): --report-md and automatic $GITHUB_STEP_SUMMARY output #1015
Closed
feat(reports): --report-md and automatic GITHUB_STEP_SUMMARY output #1049
Closed
test(reports): keep fixture summaries off the real job page #1050
Merged

Chemaclass added the enhancement New feature or request label Aug 10, 2026

Chemaclass self-assigned this Aug 10, 2026

Chemaclass added 3 commits August 10, 2026 21:33

Chemaclass force-pushed the feat/1015-report-md branch from 14d5cde to 341da2aCompare August 10, 2026 19:50

Chemaclass merged commit eab6b9a into mainAug 10, 2026
37 checks passed

Chemaclass deleted the feat/1015-report-md branch August 10, 2026 20:02

This was referenced Aug 10, 2026
feat(reports): --report-md and automatic $GITHUB_STEP_SUMMARY output #1015
Closed
feat(reports): --report-md and automatic GITHUB_STEP_SUMMARY output #1049
Closed
test(reports): keep fixture summaries off the real job page #1050
Merged

Uh oh!

Uh oh!

Conversation

Chemaclass commented Aug 10, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🤔 Background

💡 Changes

🩹 Follow-up fix in this branch

✅ Build & Verify

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Uh oh!

Conversation

Chemaclass commented Aug 10, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🤔 Background

💡 Changes

🩹 Follow-up fix in this branch

✅ Build & Verify

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Uh oh!

Conversation

Chemaclass commented Aug 10, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🤔 Background

💡 Changes

🩹 Follow-up fix in this branch

✅ Build & Verify

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Uh oh!

Conversation

Chemaclass commented Aug 10, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🤔 Background

💡 Changes

🩹 Follow-up fix in this branch

✅ Build & Verify

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Uh oh!

Conversation

Chemaclass commented Aug 10, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Chemaclass commented Aug 10, 2026•
edited
Loading

Chemaclass commented Aug 10, 2026•
edited
Loading

Chemaclass commented Aug 10, 2026•
edited
Loading

Chemaclass commented Aug 10, 2026•
edited
Loading

Chemaclass commented Aug 10, 2026•
edited
Loading

Chemaclass commented Aug 10, 2026•
edited
Loading