feat(ui): headline summary on scan completion

[GautamTalksDev]

Adds a one-line summary above the Unattributed section stating the finding in
plain language — grant count, systems, unattributed, stale, irreversible. All
figures derived from scan data; zero-value clauses omitted. Header counts
clarified: pending, held, total.

[qodo-code-review]

PR Summary by Qodo

Add completed-scan finding headline to approval queue

✨ Enhancement 🧪 Tests ⚙️ Configuration changes 🕐 10-20 Minutes

AI Description

• Summarizes completed scans with grants, systems, unattributed, stale, and irreversible findings.
• Omits empty finding clauses while preserving singular/plural and accessible summary text.
• Applies replay pacing to the demo server so completion remains understandable.

Diagram

graph TD
  D["Demo Launcher"] --> R["Replay Server"] --> S["Scan Session"] --> A["App Shell"] --> Q["Approval Queue"] --> H["Summary Helpers"] --> B["Finding Headline"]

Loading

High-Level Assessment

The client-side approach is appropriate because the approval queue already has the complete card set and scan lifecycle state. Deriving counts in shared, tested helpers avoids expanding the API contract, while using subagent identifiers correctly represents all scanned systems, including systems that produced no cards.

Files changed (5) +126 / -20

Enhancement (3) +79 / -17

App.tsxPass scanned system identifiers into the approval queue +5/-15

Pass scanned system identifiers into the approval queue

• Supplies subagent system IDs alongside cards and scan state so the queue can report every scanned system, including those without grants. Remaining changes are formatting-only.

apps/web/src/App.tsx

ApprovalQueue.tsxRender a completed-scan finding headline +35/-1

Render a completed-scan finding headline

• Computes scan-level counts and displays a compact headline only after successful completion. Optional clauses highlight unattributed, year-idle, and irreversible grants, while an aria-label exposes the complete plain-language summary.

apps/web/src/components/ApprovalQueue.tsx

format.tsAdd scan summary counting and text helpers +39/-1

Add scan summary counting and text helpers

• Introduces a typed summary model that counts cards, unique scanned systems, unattributed grants, grants idle at least one year, and irreversible actions. Adds plural-aware plain-language formatting that excludes empty finding categories.

apps/web/src/lib/format.ts

Bug fix (1) +2 / -2

demo.tsApply replay pacing before starting the demo server +2/-2

Apply replay pacing before starting the demo server

• Sets default replay speed and maximum event gap on the parent environment before launching API and UI processes. This ensures the server-side replay driver receives pacing settings and leaves the completed result visible long enough to understand.

scripts/demo.ts

Tests (1) +45 / -1

format.test.tsCover scan summary counting and clause omission +45/-1

Cover scan summary counting and clause omission

• Verifies grant and connected-system totals, notable finding categories, deterministic staleness, singular wording, and omission of zero-value clauses.

apps/web/src/lib/format.test.ts

[qodo-code-review]

Code Review by Qodo

🐞 Bugs (1) 📘 Rule violations (0) 📜 Skill insights (0)

1. Stale refresh corrupts headline 🐞 Bug ≡ Correctness

Description

A card refresh started by the previous scan can resolve after a new scan begins and overwrite the
cleared queue, while the new headline counts those old cards against the new scan's subagents.
Because that stale response can also restore status completed, users can see a completed summary
containing mismatched grant and system counts during the new scan.

Code

apps/web/src/App.tsx[71]

+          systemIds={Object.keys(session.activity.subagents)}

Evidence

The terminal UI exposes the New scan form as soon as status becomes completed, while the same
completion event asynchronously starts refreshCards; therefore a user can begin another scan
before that request resolves. scan_starting clears cards and subagents, but refreshCards carries
no scan ID and the reducer unconditionally installs its cards/status, after which the new summary
independently receives subagent keys from the current activity state.

apps/web/src/components/AgentActivity.tsx[138-144]
apps/web/src/hooks/useScanSession.ts[85-103]
apps/web/src/hooks/useScanSession.ts[132-152]
apps/web/src/hooks/useScanSession.ts[469-509]
apps/web/src/components/ApprovalQueue.tsx[123-126]
apps/web/src/components/ApprovalQueue.tsx[199-208]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
Card fetches from a prior scan can update state after a new scan starts, causing the summary to combine stale cards with the current scan's systems and status.

## Issue Context
`refreshCards` dispatches results without identifying their scan, and the reducer accepts every `cards` action. Track the request's scan ID or a session generation and ignore results that no longer belong to the active scan.

## Fix Focus Areas
- apps/web/src/hooks/useScanSession.ts[32-39]
- apps/web/src/hooks/useScanSession.ts[132-152]
- apps/web/src/hooks/useScanSession.ts[469-509]
- apps/web/src/App.tsx[69-73]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

Context sources

Review mode: ⚖️ Balanced

Tip of the day

💡 Did you know, you can enable the Remediation agent and Qodo fixes findings in a dedicated fix PR

More tips ↗ | Customize Qodo ↗ | Qodo docs ↗

[qodo-code-review]

1. Stale refresh corrupts headline 🐞 Bug ≡ Correctness

A card refresh started by the previous scan can resolve after a new scan begins and overwrite the
cleared queue, while the new headline counts those old cards against the new scan's subagents.
Because that stale response can also restore status completed, users can see a completed summary
containing mismatched grant and system counts during the new scan.

Agent Prompt

## Issue description
Card fetches from a prior scan can update state after a new scan starts, causing the summary to combine stale cards with the current scan's systems and status.

## Issue Context
`refreshCards` dispatches results without identifying their scan, and the reducer accepts every `cards` action. Track the request's scan ID or a session generation and ignore results that no longer belong to the active scan.

## Fix Focus Areas
- apps/web/src/hooks/useScanSession.ts[32-39]
- apps/web/src/hooks/useScanSession.ts[132-152]
- apps/web/src/hooks/useScanSession.ts[469-509]
- apps/web/src/App.tsx[69-73]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools