Kelly disclosure tracker
Cross-Entity Disclosure Tracker App-in-Skill — a Busabase review workspace that helps a compliance/IR team assemble and track a standardized disclosure package per financing vehicle (fund/SPV), across a generic onshore origination entity, an offshore fund-manager entity, and a listing/exchange venue. Use when the user invokes $kelly-disclosure-tracker or /kelly-disclosure-tracker, wants to review disclosure checklists, vehicle readiness, cross-entity reconciliation flags, or reviewer notes across multiple financing vehicles. Review workspace only — it never files anything or calls any external system.From its SKILL.md
npx -y skills add mr-kelly/skills --skill kelly-disclosure-trackerAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 5 stars5 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
10.9 KB, as published. Nobody here has run it
Cross-Entity Disclosure Tracker
Overview
Use this skill as a Busabase-backed review-workspace desk for a compliance/IR team assembling a standardized disclosure package per financing vehicle (fund or SPV). Each vehicle's package spans three generic entity roles:
- Origination entity — the onshore entity that originates/services the underlying assets.
- Fund-manager entity — the offshore entity that manages the vehicle.
- Listing venue — the exchange/listing venue where the vehicle's notes or units are listed.
This is a generic, brand-free tool: no real company, regulator, or exchange is referenced anywhere in the skill, its data, or its UI. All vehicle and entity names in seed data are synthetic placeholders ("SPV Alpha 12", "Onshore Originator A", "Exchange One", and so on).
Default behavior is AirApp-first. Unless the user explicitly asks only for
explanation, ensure the mock portfolio exists (run the seed script below if the
vehicles Base is empty) and give the user the clickable AirApp URL (or the
local preview URL when local preview is explicitly requested). Use chat-only
mode only when the user says "纯聊天", "chat only", "不要打开 UI", or similar.
This is a workspace/review-queue hybrid: the human works through a checklist per vehicle rather than approving a linear queue, but the underlying mechanics (statuses, decisions, writing straight to Busabase) follow the same App-in-Skill review model used elsewhere in this batch.
App UI Screenshots
<table> <tr> <td width="50%"><img src="assets/screenshots/overview.webp" alt="Disclosure Tracker overview"></td> <td width="50%"><img src="assets/screenshots/vehicle-detail.webp" alt="Disclosure Tracker vehicle detail"></td> </tr> <tr> <td><strong>Overview</strong><br>Portfolio-level summary (ready / blocked / in-progress vehicles) plus the vehicle grid.</td> <td><strong>Vehicle detail</strong><br>Checklist grouped by role (origination / fund-manager / listing venue) with a decision panel: verified, needs source, or flag inconsistent, plus a reviewer note.</td> </tr> <tr> <td colspan="2"><img src="assets/screenshots/flagged.webp" alt="Disclosure Tracker flagged items"></td> </tr> <tr> <td colspan="2"><strong>Flagged</strong><br>Cross-entity reconciliation mismatches (e.g. a figure that doesn't reconcile between the fund-manager's AUM statement and the listing venue's filing) and reviewer-flagged inconsistencies, in one list.</td> </tr> </table>Mandatory Dependencies
- Read and follow
$kelly-app-skill-creatorfor product behavior, visual quality, responsive layout, and the complete canonicalcontent/kelly-disclosure-tracker-app/artifact. - Read and follow
$busabasefor connection, target Space, node discovery, ChangeRequests, review, and merge behavior. - Read and follow
$busabase-app-creatorfor resource modeling, AirApp runtime limits, security, validation, and deployment.
If a dependency is unavailable, preserve this skill's local artifact and product contracts, stop before the unavailable Busabase operation, and report the exact missing dependency. Do not invent a second data backend.
Boundary
- Review workspace only. The skill reads and writes its own Busabase Bases; it never calls any external system, filing portal, or exchange API.
- NEVER file, submit, or transmit anything to a real regulator, fund administrator, or exchange. There is no filing path in this skill by design.
- The AirApp reads and writes its own three Busabase Bases only.
- Treat all vehicle/entity data as sensitive by convention, even though the bundled seed data is synthetic.
Busabase Resources
Three Bases under one application Folder (kelly-disclosure-tracker),
declared in content/kelly-disclosure-tracker-app/app/js/config.js and the generated template sidecars under content/:
vehicles: one row per financing vehicle (fund/SPV) — the origination entity, fund-manager entity, listing venue, base currency, and target close date. Per-vehicle checklist completeness/readiness is never stored — it is recomputed client-side from theitemsBase on every read.items: one row per standardized disclosure checklist item, scoped to a vehicle and one of the three entity roles. Carries the raw checklist fields, an optional cross-entity reconciliation record, the reviewer's decision (decision-action/decision-comment/decided-at/override-reconciliation), and — oncescripts/execute_decisions.mjsruns — an execution marker, all written directly onto the same row. Status is never stored — it is recomputed client-side from decision + reconciliation on every read.settings: up to two rows, keyed byrecord-id/kind:config(reviewer name) andrun(the current batch's batch id + generated-at).
Resources provision lazily through an idempotent Busabase ChangeRequest the
first time the app runs in a Space; see references/ui-schema.md for exact
field shapes.
First Run And Onboarding
On invocation, check the vehicles Base. If it's empty, run the trusted seed
script to generate the fixed mock portfolio:
node skills/kelly-disclosure-tracker/scripts/generate_batch.mjs --apply
There are no credentials to collect beyond Busabase itself — onboarding is
just confirming the reviewer's display name in the settings Base's config
row (defaults to "Unassigned reviewer" if omitted).
Local App
Default behavior is AirApp-first — give the user the clickable AirApp URL.
Start pnpm --dir content/kelly-disclosure-tracker-app dev only when local preview/debugging is explicitly
requested.
Required app views (hash routes):
#/vehicles: portfolio summary (ready / blocked / in-progress vehicles) plus the vehicle grid;?filter=needs_review|changes_requested|blocked|readynarrows the grid.#/vehicles/<vehicle_id>: checklist grouped by role, with per-vehicle metrics.#/vehicles/<vehicle_id>/<item_id>: item detail + decision panel (verified / needs source / flag inconsistent) with a reviewer note. Decisions write directly onto the item record throughbusabase-sdk.#/flagged: every item currently flagged, across all vehicles, with the reconciliation detail that triggered the flag.#/settings: sanitized setup summary — data provider, reviewer name, onboarding state.
Demo Mode
?demo=1opens a deterministic, fully offline mock portfolio (9 vehicles, 6 items each) for documentation and screenshots. Demo mode never reads or writes Busabase.lang=enorlang=zhforces UI chrome language for screenshots.
UI language: support English and Chinese (zh-CN) chrome with Auto default.
Workflow
node scripts/generate_batch.mjs --apply(dry run without--apply) writes the fixed 9-vehicle / 54-item mock portfolio to thevehiclesanditemsBases (resetting every item's decision fields to the standard pre-seeded mix) and refreshes thesettingsBase'srunrow — run it once at setup, and again any time the demo portfolio needs resetting.- Open the app. Vehicles shows the portfolio summary and the vehicle grid; Flagged lists every cross-entity reconciliation mismatch or reviewer-flagged inconsistency in one place.
- For each vehicle, open its checklist grouped by role (origination /
fund-manager / listing venue), select an item, and record
Verified/Needs source/Flag inconsistentwith an optional reviewer note — written straight onto the item record. A "Verified" decision on an item with an unresolved reconciliation mismatch is held at "Awaiting source" until the reviewer explicitly acknowledges the mismatch. node scripts/execute_decisions.mjs --apply(dry run without--apply) re-reads Busabase and writes an execution marker (execution-status,execution-detail,executed-at) onto every item, reporting which are settled (verified/done or flagged/blocked) vs still awaiting a human decision. It performs no external side effect — no filing, no submission, no transmission to a regulator, fund administrator, or exchange.
Read references/ui-schema.md before editing the app, scripts, or
content/kelly-disclosure-tracker-app/app/js/tracker-model.js.
The Domain Model (not a filing system)
content/kelly-disclosure-tracker-app/app/js/tracker-model.js documents and implements the entire domain
model: the fixed seed portfolio (buildSeedData()), the per-item decision ->
status mapping (computeItemStatus(), including the reconciliation-mismatch
guardrail), and the per-vehicle/portfolio rollups
(computeVehicleMetrics()/computeReadiness()/computeBatchMetrics()).
Every function is pure and deterministic — same inputs always produce the
same output — so a human reviewer can audit every status and count by hand.
It backs the trusted seed script (scripts/generate_batch.mjs), the live
Busabase read path (content/kelly-disclosure-tracker-app/app/js/providers/busabase-provider.js), and the
offline ?demo= scenario (content/kelly-disclosure-tracker-app/app/js/providers/demo-provider.js), so all
three always agree on the portfolio and its statuses.
Safety
- Review workspace only: never file, submit, or transmit anything to a real regulator, fund administrator, or exchange — there is no filing path in this skill by design.
- Do not invent reconciliation figures beyond the deterministic demo/seed
data; real usage should have the skill populate the
vehicles/itemsBases from actual source documents before asking the human to review. - Redact anything that looks like a real credential or account number in logs, reports, and UI state (none are expected in this skill's data model).
- Keep the seed portfolio minimal and use stable item ids so repeated seeds
stay idempotent (the seed script upserts by
vehicle-id/item-id, never duplicating rows).
Useful Commands
node skills/kelly-disclosure-tracker/scripts/generate_batch.mjs --apply
node skills/kelly-disclosure-tracker/scripts/execute_decisions.mjs --apply
pnpm --dir skills/kelly-disclosure-tracker/content/kelly-disclosure-tracker-app dev
What ships with it: 41 files
140.8 KB alongside SKILL.md, 19 of them executable
agents/
- openai.yaml274 B
app/
- accent-theme.css7.5 KB
- accent-theme.jsruns7.3 KB
- app.jsruns20.8 KB
- i18n/messages.jsruns5.5 KB
- index.html4.4 KB
- server/demo.tsruns9.2 KB
- server/hono.tsruns4.1 KB
- server/index.tsruns806 B
- server/launcher.tsruns2.9 KB
- server/paths.tsruns989 B
- server/setup.tsruns6.3 KB
- server/store.tsruns8.0 KB
- server/types.tsruns3.4 KB
- setup-gate.css4.5 KB
- setup-gate.jsruns8.8 KB
- start.shruns478 B
- styles.css18.6 KB
assets/
- screenshots/flagged.webp130 B
- screenshots/flagged-zh-CN.webp130 B
- screenshots/overview.webp130 B
- screenshots/overview-zh-CN.webp130 B
- screenshots/thumbs/flagged.webp130 B
- screenshots/thumbs/flagged-zh-CN.webp129 B
- screenshots/thumbs/overview.webp130 B
- screenshots/thumbs/overview-zh-CN.webp130 B
- screenshots/thumbs/vehicle-detail.webp130 B
- screenshots/thumbs/vehicle-detail-zh-CN.webp130 B
- screenshots/vehicle-detail.webp130 B
- screenshots/vehicle-detail-zh-CN.webp130 B
lib/
- data-provider/index.tsruns655 B
- data-provider/local-file-provider.tsruns1.1 KB
- data-provider/provider-interface.tsruns1.1 KB
references/
- ui-schema.md3.6 KB
scripts/
- execute_decisions.tsruns2.2 KB
- generate_batch.tsruns9.7 KB
- config.example.json131 B
- .gitignore181 B
- package.json754 B
- README.md3.3 KB
1 more file not listed here. See all 41 in the repository.
Gives 0 of the 12 instructions most review quality skills give
Counted across 1,273 of the 2,403 authors here whose files we hold, read 2026-09-06
- Ask one question at a timein 63 of 1273, across 62 files
- Provide a recommended answer for each questionin 47 of 1273, across 45 files
- Rank findings by severityin 44 of 1273
- Use parameterized queries for database accessin 38 of 1273, across 20 files
- Validate all user input with schemasin 33 of 1273, across 15 files
- Store secrets in environment variablesin 32 of 1273, across 14 files
- Explore the codebase to answer questionsin 31 of 1273, across 29 files
- Store tokens in httpOnly cookiesin 30 of 1273, across 12 files
- Implement rate limiting on API endpointsin 30 of 1273, across 12 files
- Sanitize user-provided HTMLin 29 of 1273, across 11 files
- Return generic error messages to usersin 28 of 1273, across 10 files
- Cite file and line for every findingin 28 of 1273, across 25 files
Said here and by no other author read
- read and follow the kelly-app-skill-creator dependency
- read and follow the busabase dependency
- check the vehicles base on invocation
- run the seed script if the vehicles base is empty
- provide the clickable airapp url to the user
- recompute status client-side from decision and reconciliation
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.