← the planner · for judges · no login, no key, no setup

For the judge

Everything below is verifiable in the repo or on the live app. No account, no wallet, no key needed to read this page or to click the path.

The claim

You have to sell a token you never meant to own. Glidepath turns it into a dated selling calendar sized to the market's organic demand — so you are never the biggest seller on a day the pros are exiting.

Nansen labels decide what counts as organic. It plans; it never trades.

The 30-second path

Open https://glidepath-lilac.vercel.app and click PEPE · 12B (a $40K donation) — a two-tranche calendar in ≈3 s cold (12 live Nansen credits; 0 on a cache hit). Read the Dump today line: position, estimated impact, and what share of a full day's organic buys you would be.
Click BONK · 20B on solana — when exchanges are net-receiving BONK, today is red: the first tranche is halved and the reason is printed. Costs come from a real trade/quote route, labelled as such.
Click Every Nansen call (11) under the plan — the provenance drawer: every Nansen call, the fields used, credits from the response headers, cached or live, and milliseconds.
Click ICS — one calendar event per tranche with the go/no-go rule inside; CSV for the spreadsheet.
Open the share page /p?chain=solana&token=BONK&amount=20000000000 — server-rendered, with an OG image at /api/og.

Receipts (real runs, not estimates)

Hero plan	PEPE · 12,000,000,000 on ethereum — 12 credits · 8 calls · 1.5 s, plan `688239ac1d8c`, recorded 2026-09-16 14:18 UTC (`fixtures/0X6982508145--ETHEREUM.json`)
Bench	7 tokens × 3 runs, 2026-09-16 14:22 UTC — cold p50 3.47 s / p95 5.95 s, warm p50 5 ms / p95 356 ms, 13.0 credits/plan, 9.3 calls/plan, 2.0 who-bought-sold pages/plan, 7 of 195 calls failed (6 deterministic: `tgm/flows` refuses stablecoins), 272 credits total
Credits spent building	≈ 800 credits · ≈ 700 live Nansen calls (header balance 62,320 → 59,360 across the build; table in DEMO.md)
Nansen surface	7 endpoints · 11 calls per plan · 12 credits on EVM chains, 15 on solana/base, 0 on a cache hit
Tests	221 vitest tests (unit, end-to-end on a fake Nansen, one replay per fixture, key-boundary) · 13/13 recorded plans reproduce offline byte-for-byte
Property-based verification	60,000 generated cases (fast-check, 6 properties × 10,000) on the tranche planner: Σ tranches + remainder = amount held; every tranche ≤ k × organic/day and ≤ 1% of liquidity; a red today halves tranche 1 and only tranche 1; consecutive UTC dates, ≤ 90 of them, expectedDays ≥ days ≥ 1; the decision hash is deterministic and invariant to timing and context fields. It found one real defect (a zero-token tranche emitted 90 empty rows) — fixed and pinned.
E2E	4 Playwright suites (home, planner flow, responsive 320/375/768/1440, this page) — run in CI with no key; the built app is asserted to never contain an `nsn_` key
Clean clone	≈ 15 s from `git clone` to the first live plan (independent reviewer, 2026-09-16 23:51 UTC, warm npm cache) — budget 10 minutes on a cold cache

Reproduce

The real path — live Nansen, real credits:

git clone https://github.com/edycutjong/glidepath && cd glidepath && npm install
export NANSEN_API_KEY=nsn_...     # https://app.nansen.ai/api
npm run glidepath -- PEPE --chain ethereum --amount 12000000000 --explain   # 12 credits, ~2–3 s
npm run bench                                                              # ~270 credits, 7 tokens × 3 runs

CI / deterministic replay — no key, no network, 0 credits (this is proof the engine is a pure function of the recorded responses, not the product):

npm run verify        # 13/13 plans reproduced offline (NANSEN_OFFLINE=1 inside the script)
npm test              # 221 tests incl. 60,000 property cases
npm run check         # submission-readiness audit

Honest limits

The impact model is an approximation. Constant-product on Nansen's liquidity_usd as one pool with the token on one side — optimistic for a single thin pool, no MEV, no gas. Only solana and base get a real routed trade/quote; EVM chains show the model, labelled.
The pro share is small on real tokens. Across 11 tokens, Smart Money / funds / whales / exchanges were 0–3.6 % of DEX buys, so the headline label filter moves “organic” by a few percent. Glidepath shows the split instead of dramatising it; the red-day rule (cohort flows) is where Nansen data visibly changes the calendar.
A slow Nansen day is visible. Cold plans take 2–10 s; the organic-breadth page has a 12 s ceiling and timed out once in review (plan took 14.8 s with a red “failed: timeout” provenance row, concentration not applied). Failures are never cached and never invented — the term is null and the warning names the endpoint.

Links

Live app Repository DEMO.md — reproduce every number SCORING.md — formulas with the PEPE numbers Release v1.1.0 @edycutjong — the demo clip is posted here at submission

Built for the Nansen Meridian Buildathon (14–27 Sep 2026). Not financial advice.