cash-analyze
SkillDev toolsAnalyze artifact consistency for a change. Use when a named change's proposal, design, specs, and tasks may conflict before implementation.
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the cash-analyze skill
What this skill tells your AI
The instructions your AI receives, as published by cashwu/iphonelocationmove in .agents/skills/cash-analyze/SKILL.md and read by ahel’s review.
Project-local Cash CLI bootstrap
執行任何 Cash artifact command 前,MUST 先從目前目錄解析並驗證 Git root,再使用該 root 下的 absolute launcher;不得依賴 PATH 或外部 runtime:
cash_root="$(git rev-parse --show-toplevel)" || exit 1
cash_cli="$cash_root/.cash-skills/bin/cash"
test -x "$cash_cli" || exit 1
同一段 workflow 後續每個 artifact command MUST 使用 "$cash_cli"。
Analyze artifact consistency for a change. Can be invoked directly or triggered automatically when all artifacts are complete.
Input: Optionally specify a change name (e.g., $cash-analyze add-auth). If omitted, infer from conversation context or auto-select if only one active change exists.
Prerequisites: This skill requires the Cash CLI. If any Cash command fails with "command not found" or similar, report the error and STOP.
Response language: All user-facing responses in this workflow MUST be written in Traditional Chinese unless the user explicitly requests another language. Keep shell commands, file paths, code identifiers, schema field names, and quoted source text verbatim.
Steps
-
Determine change name
If not provided, infer from context or run
"$cash_cli" list --jsonto auto-select. -
Run programmatic analysis
"$cash_cli" analyze <change-name> --jsonThis returns structured JSON with:
dimensions: Array of{ dimension, status, finding_count }for Coverage, Consistency, Ambiguity, Gapsfindings: Array of{ id, dimension, severity, location, summary, recommendation }artifacts_analyzed/artifacts_missing: Which artifacts were available
-
Run mandatory semantic analysis
Read the available artifacts and apply the Semantic traceability contract below to every supported dimension. Semantic analysis is required: check mutually exclusive decisions, orphan tasks, risk dispositions and cross-artifact path/behavior contradictions. Preserve schema-aware skipped/insufficient dimensions and every current validation failure.
-
Present one consolidated report
Only after CLI and semantic analysis finish, output one summary table for Coverage, Consistency, Ambiguity and Gaps, then group the merged findings by severity with locations, causal evidence and concrete recommendations. Do not present the structural report early or treat semantic checks as optional.
-
Recommend next steps
- If CRITICAL findings: "Found N issue(s) worth addressing. Want to fix these before implementing?"
- If only warnings/suggestions: Note them briefly, then recommend proceeding with
$cash-apply - If clean: "Artifacts look consistent" and suggest
$cash-apply
Passive Trigger
When "$cash_cli" status --change "<name>" --json shows isComplete: true, run this analysis automatically before recommending $cash-apply.
Guardrails
Semantic traceability contract
Run the CLI analysis and semantic checks as one bounded analysis, then emit one consolidated report. Classify each semantic finding into exactly one dimension: mutually exclusive decisions are Warning, and a decision that blocks implementation is Critical; a task isolated from the declared scope is Suggestion; a design risk without a traceable disposition is Warning; and a contradiction across artifact paths or observable behavior is Warning. Every finding records its artifact scope or location, the causal evidence, and the affected requirement or task.
An effective risk disposition is one of a requirement or scenario, a concrete mitigation, an executable verification task, or the user's explicit acceptance. An agent-authored intentional note is not user acceptance. A resolved risk is removed from uncovered-risk only after its disposition is traceable; existing safety exceptions and mutually exclusive decisions remain visible. For no-spec changes, mark spec mapping and scenario/example checks as not applicable while retaining design, tasks, execution evidence and current validation failures.
- Read-only: NEVER modify files
- Keep output concise - this runs inline, not as a separate workflow
- If AskUserQuestion tool is not available, ask the same questions as plain text and wait for the user's response
Signals
- GitHub stars
- 62
- Forks
- 10
- Last commit
- Sep 2026
Others that do the same job
Advanced
- Catalog kind
- skill
- Gateway key
cash-analyze- Source
- github.com/cashwu/iphonelocationmove