Readiness · Task evidence
What did the agent do?
These records show read-only browser tasks on the owned Readiness site and on four local test pages. Read the steps, outcome, time, and cost of each run. Task results are separate from the static readiness score.
One owned site, three fixed tasks, two runs per task. No independent or general agent-success benchmark. No model weight was changed from these results.
0 published runs · model 27 · capture profile 28 · revision cd4e33b
No task record is published with an approved source audit.
Does a documentation check predict task success?
Four owned local fixture pages; one fixed read-only task, repeated twice per case.
Predict documentation task success only when public-api-docs is pass.
4 test cases · 8 browser runs · model 27
- Pass prediction; task passed
- 2
- Pass prediction; task failed
- 2
- Failure prediction; task passed
- 2
- Failure prediction; task failed
- 2
A documentation link can open an empty page. A working button can have no source anchor. These cases show why a link check cannot guarantee that a task will pass.
These purpose-built controls test a task proxy, not the truth of the narrower link check or the accuracy of the full score. They are not a random or independent site sample.
Keep model 27 unchanged. A source link can exist without usable documentation, and a native button can work without a source anchor. A link check must not be presented as a task-success guarantee. This small control set does not justify new score weights.
Download the checks, page source, and task records
Working documentation link · succeeded · true pass
Task: Find documentation for GET /api/v1/widgets and its response format.
Expected outcome: succeeded
Observed: The required GET operation and response format are present.
2.35 seconds of browser execution · 2026-09-30T19:51:42 UTC
public-api-docs: pass. The page links to 1 API or documentation target(s).
-
Open the owned fixture
linked-good.html1151 ms -
Follow the named documentation control
docs.html783 ms
Task runner paid API cost: $0.00
Total cost: not measured. Total task cost is not measured. The local browser runner makes no paid model or browser API call. Codex planning and hosting are outside this measure.
Link to an empty page · failed · false pass
Task: Find documentation for GET /api/v1/widgets and its response format.
Expected outcome: failed
Observed: The page is readable, but the required operation and response format are absent.
2.25 seconds of browser execution · 2026-09-30T19:51:44 UTC
public-api-docs: pass. The page links to 1 API or documentation target(s).
-
Open the owned fixture
linked-placeholder.html1094 ms -
Follow the named documentation control
placeholder.html748 ms
Task runner paid API cost: $0.00
Total cost: not measured. Total task cost is not measured. The local browser runner makes no paid model or browser API call. Codex planning and hosting are outside this measure.
No documentation control · failed · true failure
Task: Find documentation for GET /api/v1/widgets and its response format.
Expected outcome: failed
Observed: No documentation link or control is available.
1.61 seconds of browser execution · 2026-09-30T19:51:50 UTC
public-api-docs: warning. The page mentions an API, but no API documentation link was classified.
-
Open the owned fixture
no-docs.html1000 ms
Task runner paid API cost: $0.00
Total cost: not measured. Total task cost is not measured. The local browser runner makes no paid model or browser API call. Codex planning and hosting are outside this measure.
Working documentation link · succeeded · true pass
Task: Find documentation for GET /api/v1/widgets and its response format.
Expected outcome: succeeded
Observed: The required GET operation and response format are present.
2.22 seconds of browser execution · 2026-09-30T19:51:52 UTC
public-api-docs: pass. The page links to 1 API or documentation target(s).
-
Open the owned fixture
linked-good.html985 ms -
Follow the named documentation control
docs.html831 ms
Task runner paid API cost: $0.00
Total cost: not measured. Total task cost is not measured. The local browser runner makes no paid model or browser API call. Codex planning and hosting are outside this measure.
Link to an empty page · failed · false pass
Task: Find documentation for GET /api/v1/widgets and its response format.
Expected outcome: failed
Observed: The page is readable, but the required operation and response format are absent.
2.24 seconds of browser execution · 2026-09-30T19:51:54 UTC
public-api-docs: pass. The page links to 1 API or documentation target(s).
-
Open the owned fixture
linked-placeholder.html1104 ms -
Follow the named documentation control
placeholder.html725 ms
Task runner paid API cost: $0.00
Total cost: not measured. Total task cost is not measured. The local browser runner makes no paid model or browser API call. Codex planning and hosting are outside this measure.
No documentation control · failed · true failure
Task: Find documentation for GET /api/v1/widgets and its response format.
Expected outcome: failed
Observed: No documentation link or control is available.
1.62 seconds of browser execution · 2026-09-30T19:51:59 UTC
public-api-docs: warning. The page mentions an API, but no API documentation link was classified.
-
Open the owned fixture
no-docs.html1003 ms
Task runner paid API cost: $0.00
Total cost: not measured. Total task cost is not measured. The local browser runner makes no paid model or browser API call. Codex planning and hosting are outside this measure.