Frontend/improve report view

[yvesbou]

No description provided.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
defiscan-v2-defiscan-frontend	Ready	Preview, Comment	Mar 6, 2026 0:14am
defiscan-v2-docs	Ready	Preview, Comment	Mar 6, 2026 0:14am
defiscan-v2-tool-frontend	Ready	Preview, Comment	Mar 6, 2026 0:14am

[claude]

test

[claude]

Logic error: governance-tagged human admins excluded from admin count

The humanControlled filter uses isHumanType(a) && !a.isGovernance, so a governance-tagged EOA/Multisig/Timelock is excluded from this array but captured by governance.

The narrative sentence (around line 146) uses {humanControlled.length} for the displayed count. If a protocol has only governance-tagged human admins (e.g., a Multisig with isGovernance: true), the text will say "We identified 0 admins" while that admin visibly renders under the Governance heading on the same screen.

Reference:

defiscan-v2/packages/defiscan-frontend/src/pages/review/views/report/AdminCards.tsx

Lines 104 to 112 in 7dce5b1

	const humanControlled = sorted.filter(
	(a) => isHumanType(a) && !a.isGovernance,
	)
	const governance = sorted.filter((a) => a.isGovernance)
	const noHumanControl =
	humanControlled.length === 0 && governance.length === 0

Suggested fix: Either use humanControlled.length + governance.length in the narrative count, or redefine governance as only non-human-type contracts tagged as governance (which is the more common semantic -- a governance contract is a smart contract, not an EOA).

---

Note: This review was intended as an inline comment on packages/defiscan-frontend/src/pages/review/views/report/AdminCards.tsx at diff position 26.

[claude]

Deleting this test comment - posted in error

[claude]

debug test

[claude]

Code review

1 issue found.

In AdminCards.tsx, the humanControlled filter excludes governance-tagged human admins (isHumanType(a) && !a.isGovernance). If a protocol only has governance-tagged EOAs/Multisigs, the narrative text reads "We identified 0 admins" while simultaneously showing those admins in the Governance section. See inline comment below referencing AdminCards.tsx at diff position 26.

---

Inline comment for packages/defiscan-frontend/src/pages/review/views/report/AdminCards.tsx (diff position 26):

Logic error: governance-tagged human admins excluded from admin count

The humanControlled filter uses isHumanType(a) && !a.isGovernance, so a governance-tagged EOA/Multisig/Timelock is excluded from this array but captured by governance.

The narrative sentence (around line 146) uses {humanControlled.length} for the displayed count. If a protocol has only governance-tagged human admins (e.g., a Multisig with isGovernance: true), the text will say "We identified 0 admins" while that admin visibly renders under the Governance heading on the same screen.

Reference:

defiscan-v2/packages/defiscan-frontend/src/pages/review/views/report/AdminCards.tsx

Lines 104 to 112 in 7dce5b1

	const humanControlled = sorted.filter(
	(a) => isHumanType(a) && !a.isGovernance,
	)
	const governance = sorted.filter((a) => a.isGovernance)
	const noHumanControl =
	humanControlled.length === 0 && governance.length === 0

Suggested fix: Either use humanControlled.length + governance.length in the narrative count, or redefine governance as only non-human-type contracts tagged as governance (which is the more common semantic — a governance contract is a smart contract, not an EOA).

[claude]

Code review

3 issues found:

Bug 1 — total prop unused in DistributionChart bar width (FundCards.tsx line ~199)

DistributionChart destructures total but computes percentages using maxValue (the largest single entry). The largest bar always renders at 100% width, misrepresenting the distribution. total is correctly passed at both call sites. Fix: use total instead of maxValue in the percentage calculation.

defiscan-v2/packages/defiscan-frontend/src/pages/review/views/report/FundCards.tsx

Lines 196 to 201 in 82259e9

	<div className="space-y-3">
	{entries.map((entry, index) => {
	const percentage =
	maxValue > 0 ? (entry.value / maxValue) * 100 : 0
	const expandKey = `${keyPrefix}-${entry.fund.address}`
	const isExpanded = expandedSet.has(expandKey)

Bug 2 — Timelock missing from HUMAN_ADMIN_TYPES in LandingPage.tsx line 13

compile-data.ts and AdminCards.tsx both treat Timelock as a human admin type, but LandingPage.tsx does not. A protocol whose only admin is a Timelock will have its token value counted in totalTokenValueAtRisk globally, but the landing page table will show 0 human admins for that protocol. Fix: add 'Timelock' to the set on line 13.

defiscan-v2/packages/defiscan-frontend/src/pages/landing/LandingPage.tsx

Lines 12 to 15 in 82259e9

	const HUMAN_ADMIN_TYPES = new Set(['EOA', 'EOAPermissioned', 'Multisig'])
	function computeHumanAdminCount(review: CompiledReview): number {

Bug 3 — funds.length count includes token-only entries excluded from Fund Distribution chart (FundCards.tsx line 115)

funds.length counts all fund entries, but totalFunds is computed from balances/positions only — token-only entries (with tokenInfo but balances: null) contribute $0 and are excluded from chartData. For Liquity V1 (added in this PR): 6 entries counted but only 4 have balance data. Fix: replace funds.length with chartData.length.

defiscan-v2/packages/defiscan-frontend/src/pages/review/views/report/FundCards.tsx

Lines 113 to 118 in 82259e9

	{formatUsdValue(totalFunds)}
	</span>{' '}
	across {funds.length} contract{funds.length !== 1 ? 's' : ''}.
	{totalTokenValue > 0 && tokenNames.length > 0 && (
	<>
	{' '}

[claude]

Removing test comment

[claude]

inline test body

[claude]

Code review comment on packages/config/src/projects/liquity-v1/compiled-review.json (line 1): Wrong output location. CLAUDE.md documents compiled-review.json output exclusively to packages/defiscan-frontend/public/data/slug/. The correctly placed copy already exists in this PR at packages/defiscan-frontend/public/data/liquity-v1/compiled-review.json. No other project has compiled-review.json in its config project folder, and reviewCompiler.ts has no code path that writes here. Please remove packages/config/src/projects/liquity-v1/compiled-review.json from this PR.

[claude]

Code review comment on packages/defiscan-frontend/src/pages/landing/LandingPage.tsx (line 14): HUMAN_ADMIN_TYPES is missing Timelock. This set is inconsistent with compile-data.ts and AdminCards.tsx which both include Timelock. Timelock-type admins are not counted in the landing page Admins column but are counted as human-controlled in compile-data.ts when computing totalTokenValueAtRisk. A protocol with only Timelock admins would show 0 in the table Admins column while still having its token value counted in global stats. Fix: const HUMAN_ADMIN_TYPES = new Set(['EOA', 'EOAPermissioned', 'Multisig', 'Timelock'])

[claude]

Code review comment on packages/defiscan-frontend/src/pages/review/views/report/AdminCards.tsx (line 146): Admin count and capital figure are drawn from inconsistent populations. The intro sentence says We identified N admins controlling X in locked funds where N = humanControlled.length + governance.length (human-type + governance-tagged admins only) but X = totals.totalCapitalAtRisk (capital reachable by ALL admins including Immutable, plain Contract, and Revoked types). For protocols with immutable contracts controlling funds, X will be larger than what the N named admins actually control, making the sentence misleading. The humanTotal and govTotal variables are already computed just above this JSX. Using humanTotal + govTotal would make count and capital consistent.

[yvesbou]

@claude review the changes to the explorer view pls

[claude]

Claude finished @yvesbou's task in 1m 59s —— View job

---

Explorer View Code Review

• Read explorer view changed files
• Analyze changes and identify issues
• Post review feedback

---

Overall the explorer view is well-structured: sortable tables, expandable rows, and the getHumanAdmins utility cleanly shared across all tabs. A few issues worth addressing:

Bug 1 — Double-counting in Admin Type Distribution bar (OverviewTab.tsx line 37–51)

governanceAdmins is built from humanAdmins.filter((a) => a.isGovernance). A Multisig or Timelock with isGovernance: true lands in both its type bucket (Multisig/Timelock) and the Governance bucket. totalAdminCount is the sum of all buckets, so it exceeds humanAdmins.length in that case. The bar segments then use count / totalAdminCount * 100% and their widths won't add up to 100%.

Additionally, revokedAdmins is pulled from all admins (admins.filter(...) at line 26) rather than from humanAdmins. If revoked admins exist alongside human admins, they inflate totalAdminCount further.

Suggested fix: Filter governance to only non-human-type contracts (!isHumanAdminType(a.adminType) && a.isGovernance) so each admin appears in exactly one bucket; exclude Revoked from the breakdown or base it on humanAdmins only. Fix this →

Bug 2 — "positions" mislabeled as "Market Cap" in FundsTab (FundsTab.tsx)

In the fund data model, positions holds DeFi protocol positions (Aave supply, Morpho vault positions, etc.) — not the protocol token's market cap. The actual market cap is tokenInfo.tokenValue. However:

• Chart tooltip formatter (line ~153): name === 'balances' ? 'Total Value Locked (TVL)' : 'Market Cap' → calls positions "Market Cap"
• Table column header (line ~220): <SortHeader field="positions" label="Market Cap" ...> → same mislabeling
• Chart <Bar> name (line ~175): name="Market Cap" for the positions bar

This makes the "Market Cap Token" column (which correctly uses tokenInfo) redundant/confusing — two columns appear to show market cap. Fix this →

Bug 3 — funds.length count includes token-only holders (FundsTab.tsx line 101)

The summary text "N fund holders" uses funds.length, but chartData filters out entries with no balances or positions (token-only holders with only tokenInfo). A holder with market cap data but $0 TVL still increments the count but shows nothing in the bar chart — users see "6 fund holders" but only 4 bars. (Same issue as previously noted for FundCards.tsx.) Fix this →

Observation — CapitalFlowDiagram.tsx: function circles disconnected from fund lines

The center column renders a circle per admin showing the permissioned function count, with a straight arrow line from the admin box to the circle. However the bezier connections paths (lines 201–218) draw directly from the admin's right edge to the fund holder's left edge — they visually pass through the function column without connecting to the circles. This makes the center column appear decorative: the curved lines bypass the circles entirely.

If the intent is to show the admin → function → fund flow, the lines should route through the function circles (i.e., from admin right → circle center → fund left, split into two segments). If the intent is just to annotate function counts per admin, the column label "FUNCTIONS" is misleading. Fix this →

Minor — AdminsTab.tsx footer totals are global, not filtered-admin totals (lines 194–219)

The table footer displays totals.totalCapitalAtRisk under "Direct Funds" and totals.permissionedFunctionCount under "Functions". These are protocol-wide totals from the compiled review, not sums over the displayed humanAdmins rows. If immutable contracts hold funds or non-human admins control functions, the footer will show a higher number than the sum of visible rows — which looks like an addition error. The table could instead sum humanAdmins.reduce((s, a) => s + a.totalDirectCapital, 0) for the Direct Funds total and sum humanAdmins.reduce((s, a) => s + a.functions.length, 0) for the Functions total.

---

[yvesbou]

@claude review again

[claude]

Claude finished @yvesbou's task in 3m 45s —— View job

---

Code Review

• Read all changed files
• Analyze report view changes
• Analyze explorer view changes
• Analyze other changes
• Post review

---

What's been fixed since the last review

• AdminCards.tsx: Admin count now uses humanControlled.length + governance.length ✓
• OverviewTab.tsx: Admin type breakdown bar — each admin now lands in exactly one bucket (!a.isGovernance guard on EOA/Multisig/Timelock buckets, so governance-tagged human-types fall only into the Governance bucket). No double-counting. ✓
• FundsTab.tsx: "Positions" column is no longer mislabeled "Market Cap" ✓
• FundsTab.tsx: Summary now uses FundsSummaryLabel with separate tvlCount / tokenCount, not funds.length ✓
• packages/config/src/projects/liquity-v1/compiled-review.json: No longer present in the wrong location ✓

---

Remaining issues

Bug — DistributionChart bars relative to maxValue, not total (FundCards.tsx line 198–199)

const maxValue = Math.max(...entries.map((e) => e.value), 0)
// ...
const percentage = maxValue > 0 ? (entry.value / maxValue) * 100 : 0

The total prop is passed to DistributionChart but never used for percentages. The largest bar always renders at 100% width, making each bar show "how much relative to the largest entry" rather than "what share of the total." The first bar will always be 100% wide, and a fund holding 10% of TVL next to one holding 9% will look nearly identical. Fix: const percentage = total > 0 ? (entry.value / total) * 100 : 0.

Fix this →

---

Bug — Timelock missing from HUMAN_ADMIN_TYPES in LandingPage.tsx line 13

const HUMAN_ADMIN_TYPES = new Set(['EOA', 'EOAPermissioned', 'Multisig'])

admins.ts (the shared utility), compile-data.ts, and AdminCards.tsx all treat Timelock as a human admin type. LandingPage.tsx has its own local copy that excludes Timelock. A protocol whose only active admin is a Timelock without isGovernance: true will show 0 in the "Admins" column of the landing page table, even though that same protocol reports a Timelock admin in its detail view. Fix: add 'Timelock' to the set.

Fix this →

---

Bug — AdminsTab.tsx footer totals don't match visible rows (lines 194–217)

The "Direct Funds" footer cell shows totals.totalCapitalAtRisk (line 196) and "Functions" shows totals.permissionedFunctionCount (line 216). Both are protocol-wide compiled totals. The table only shows humanAdmins rows. If immutable contracts hold funds or non-human admins control functions, the footer will show a larger number than the sum of the visible rows — which looks like an arithmetic error to the reader.

Fix: replace with sums over the displayed humanAdmins array:

humanAdmins.reduce((s, a) => s + a.totalDirectCapital, 0)  // for Direct Funds
humanAdmins.reduce((s, a) => s + a.functions.length, 0)    // for Functions

Fix this →

---

Minor / Observation

AdminCards.tsx narrative capital vs admin count mismatch (line 154–155)

The intro sentence says "N admins controlling X in locked funds" where X = totals.totalCapitalAtRisk. This total includes funds controlled by all admin types — Immutable, Contract, Revoked — not just the N human+governance admins displayed. The humanTotal + govTotal variables are already computed directly above and would give a consistent number. Low severity since Immutable contracts typically aren't the dangerous case, but could mislead readers of protocols with significant immutable-locked TVL.

CapitalFlowDiagram.tsx — function circles disconnected from fund flow (design observation)

The bezier connections paths draw directly from the admin box's right edge to the fund holder's left edge (M${adminColX + boxW},${adminCenterY} ... ${fundColX},${fundCenterY}). The function count circles in the center column are connected to admins via a short arrow, but have no outgoing connection to fund holders — the curved lines bypass them entirely. The column label "FUNCTIONS" and the circles are decorative under the current rendering. Either route lines through the circles (admin→circle→fund), or remove the center column and label.

---