🛠️ Update various documentation, scripts, and configuration templates to enhance clarity, functionality, and maintainability across the project

resources/prompts/audit-copilot-usage.prompt.md

@@ -0,0 +1,17 @@
+---
+name: "audit-copilot-usage"
+description: "Run the local Copilot reuse audit and review candidates for promotion into the shared resource repo."
+agent: "agent"
+tools: [read, search, execute]
+argument-hint: "days=<default 30> workspace=<optional path filter> sources=<optional csv>"
+---
+
+Run the repository audit workflow by using the shared audit script.
+
+Requirements:
+
+- Prefer `resources/scripts/audit-copilot-usage.sh` over ad hoc searching.
+- Keep the audit itself read-only.
+- Review `audit-summary.md`, `candidates-report.tsv`, `selection-manifest.tsv`, and `pattern-details/`.
+- Recommend promotion only for portable resources; keep machine-specific or
+  repo-specific patterns in audit notes.

resources/prompts/prepare-audit-promotions.prompt.md

@@ -0,0 +1,16 @@
+---
+name: "prepare-audit-promotions"
+description: "Generate draft resources or staging notes from approved audit manifest rows."
+agent: "agent"
+tools: [read, search, execute]
+argument-hint: "audit=<optional audit directory>"
+---
+
+Generate promotion drafts from the audit approval surface by using the shared repository script.
+
+Requirements:
+
+- Prefer `resources/scripts/prepare-audit-promotions.sh` over manual translation.
+- Treat `promote-*`, `template-only`, and `docs-only` rows as approved when their `decision` column is non-empty.
+- Treat `discard`, `needs-sanitization`, and empty decisions as non-approved.
+- Review `promotion-summary.md` after the script runs and summarize the generated draft resources and staging notes.

resources/prompts/review-audit-candidates.prompt.md

@@ -0,0 +1,20 @@
+---
+name: "review-audit-candidates"
+description: "Review an audit shortlist one candidate at a time and update the selection manifest with decisions and notes."
+agent: "agent"
+tools: [read, search, edit]
+argument-hint: "audit=<optional audit directory>"
+---
+
+Review the audit shortlist one candidate at a time.
+
+Requirements:
+
+- Resolve the target audit directory from `audit=<path>` when provided, otherwise use the latest run under `.local/audits/`.
+- Read `selection-manifest.tsv` first, then open each pending candidate's `detail_file` before asking for a decision.
+- Summarize the candidate's purpose, expected benefit, audit context, and why it scored as it did before asking for a decision.
+- Mention transcript prompt-cost proxy fields when they are present, but label them as cost signals rather than exact token counts.
+- If the detail bundle has weak or missing evidence, say that explicitly and explain what the user would be deciding on anyway.
+- Use `vscode_askQuestions` to present decision choices as buttons and allow freeform text for `review_note`.
+- Update `selection-manifest.tsv` after each answer so review progress survives interruptions.
+- Stop only when every row has a non-empty `decision` or the user tells you to stop.