cawplan-test-execution-progress

CawPlan Test Execution Progress — A3 Execution Progress Query

cawplan skill check

Hard Gate (MUST)

Read-only query: A1/A2/A4 write operations prohibited,cawplan api, direct TestRail connection,failures --test-id(belongs to A4).
summary defaults to --refresh; omitted only when the user explicitly requests accepting the cache.
Pass quality = (passed+auto_passed+passed_with_issue)/executed；prohibitedto display API pass_rate。
Pending closurebased on the deduplicated failures queue,prohibitedto use aggregated.failed/blocked or raw items.length。
User presentation: friendly field names (ux.md §Glossary）；result_id/UUID hidden by default; Agent on its own AskUserQuestion(not written to allowed-tools）。
Drill-down gate: when the full 4-layer report has been output,do not pop up; pops up only for a health summary or when the user says "just take a quick look" (ux.md §Prompts）。
A4 handoff: row-level (user clicks a row or says to register a defect) + one sentence at the end;do notforce a pop-up, do not proactively execute A4.

Timing	Read	prohibited
Opening	no	one-time Read of all references
query / metrics / failures strategy	`execution.md`	—
interaction, output template, A4 handoff	`ux.md`	—

Step	Action	Detail
0	parse product/version; scope gate (when unspecified)	`execution.md §0` · `ux.md §Prompts`
1	`execution summary`	`execution.md §Summary`
2	risky Runs separately `failures`; deduplicate and sort	`execution.md §Failures`
3	four-layer report (see `ux.md §Output`）	full 4 layers → do not pop updrill-down gate
4	user clicks a row/says to register a defect → row-level A4 guidance	`ux.md §A4`

Prohibitedto treat the A3 summary as an A5 risk conclusion.

Four layers: ① execution health ② pending items ③ progress by ticket ④ execution composition (+ appendix as needed). Template see ux.md §Output。

Fixed one-sentence A4 prompt at the end (no pop-up). If the user wants "technical details", then expand result_id、plan_mapping_id etc.