Verify that Codex Slides changes are durably reflected in project state and visibly reflected in the exact Codex Browser slide, panel, version, checkpoint, or presentation surface. Use after creation, rendering, editing, restyling, mark edits, speaker-note changes, version restores, or export.
SkillScore82Maintained
Automatic
◆QA & Test Engineering pack
Last verified 8/16/2026
What it changes
Passed 17 of 21 test cases with the skill, and 9 of 21 without it.
Without skill41%
With skill82%
Checks it passed that the model alone missed
Confirm workflow stage in project state.
Treat timed-out wait as leaving run active.
Confirm current deck version in project state.
and 10 more in the full report
9 fixed9 passed either way4 still failing0 broke
Pass-rate lift
+41 pp
Tokens
-25%
Agent turns
0%
Response time
-48%
Model gemini-3.6-flashMeasured 8/16/2026v118 of 22 cases passed with skill · 21 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 9 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 9
Prompt
An automated script modified a slide deck and returned runId 'run_888'. Which function should be called to retrieve or wait on the status of this durable background project run?
What was checked, without and with the skill
Checked
Without skill
With skill
Call get_project_run or wait_project_run to inspect durable run status.
Missed
Passed
case-05 · gemini-3.6-flash · 8/16/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md29 linessingle file
Codex Slides Verification
Use two signals before reporting a presentation task complete:
Read canonical project state with get_project, and read a named durable job
with get_project_run or wait_project_run when a runId was returned.
Follow the returned browserHandoff and inspect the exact visible result in
the Codex in-editor Browser.
Verification sequence
Confirm the project id rather than guessing from a similar title.
For a durable run, require a terminal complete status. Treat error and
cancelled as incomplete; a timed-out wait leaves the run active.
Confirm the expected page count, changed slide indexes, rendered images,
transitions, speaker notes, current workflow stage, and current deck version.
Open the exact handoff returned by the mutation. Preserve its slide,
panel, version, mode, conversation, checkpoint, and run query
parameters.
In Browser, confirm the intended slide or panel is focused and the visible
result matches the request. Do not treat a JSON response alone as visual QA.
Export only after the current or explicitly selected version passes review.
If the in-editor Browser is unavailable, use the project state and persisted
slide-image endpoints as structural evidence, return the exact handoff URL, and
state that visible Browser verification remains outstanding.
For cross-surface changes, also verify that refreshing or reopening the same
project preserves the run result and deep-link focus.