Application for Filecoin ProPGF Batch 3

Plumbline : CalibrationNet Stability Project

https://github.com/Reiers/plumbline

https://faucet.reiers.io

Nicklas Reiersen (TSE Reiersen)

The operational layer of the Filecoin CalibrationNet: one long-running, deliberately-stable Storage Provider test target (t0143103), held under 1/3 of calibration QAP on decentralization guidance so a single-operator outage can't stall F3 finality; the public dual-asset tFIL + USDFC faucet at faucet.reiers.io; daily network monitoring; and end-to-end nv-upgrade validation (Calix) run before every mainnet adoption. Used by SynapS3, the FoC integrator community, and every developer who needs repeatable conditions on calibration.

Fully Open Source

N/A - fully open source under MIT. The `plumbline` umbrella, `plumbline-faucet`, and `calix` are public; the roadmap siblings `plumbline-monitor` (status page + probes) and `plumbline-cli` (operator + builder scripts) ship public under the same license. No closed components.

**Plumbline** ([github.com/Reiers/plumbline](https://github.com/Reiers/plumbline)) is the operational stability layer for the Filecoin CalibrationNet, run by TSE Reiersen (Org. 929 074 912). This proposal funds **6 months of operational ownership** of that surface, plus the new builder on-ramp + API roadmap. It is the **successor to the Curio/Blueshift calibration side letter**, which winds down Sept 1, 2026. ### What's covered | Component | What it is | Status | |---|---|---| | **Stable SP test target** | `t0143103` - a *repeatable* SP test target maintained across nv boundaries (not an experimental SP that resets between campaigns), held at **~32% of calibration QAP (~350 TiB)**, deliberately under 1/3 on decentralization guidance so a single-operator outage can't stall F3 finality | 🟢 online & steady | | **Public faucet** | [faucet.reiers.io](https://faucet.reiers.io) - independent **tFIL + USDFC** drips, 24h per-asset cooldown, Cloudflare Turnstile gate, on-chain verification of every drip, live `/status`, public API | 🟢 live since Apr 2026 | | **Calix nv-validation** | [calix.reiers.io](https://calix.reiers.io) - every network-version upgrade exercised end-to-end (manifest match, migration audit, state integrity) before mainnet adoption | 🟢 shipped clean through nv28 "Fire Horse" | | **Daily network monitoring** | every CalibrationNet node and participating miner | 🟢 ongoing | | **Builder on-ramps + API** | self-serve on-ramp (get funded + pointed at endpoints in one flow) + a public read-only builder API - the `plumbline-monitor` and `plumbline-cli` siblings | 🟡 funded by this grant | ### A note on topology (why one SP, not two) Plumbline previously ran two SP test targets. On the calibration governance thread I was asked to reduce my network footprint for decentralization: keep any single operator under 1/3 of network QAP so that if my SP goes down, calibration F3 finality can't stall (a supermajority can't form if a >1/3 holder is offline). In response I retired the backup SP (`t0144416`) and trimmed the primary (`t0143103`) to ~32% of calibration QAP. This was compliance with network guidance, not a scope reduction on the deliverable - the stable-test-target function is intact on a single, right-sized SP. ### Who this supports Every Filecoin developer who uses calibration as their integration target: application builders integrating against `synapse-sdk` and `FilecoinPay`, storage providers testing PDP before mainnet, contract developers staging FoC upgrades, and auditors who need repeatable conditions. When calibration is hard to fund, hard to monitor, or its SP target is unstable, FoC development velocity slows for the whole ecosystem. This proposal removes that drag. ### The Filecoin-specific gap The calibration stability surface is funded today only by a residual Curio/Blueshift side letter (TSE Reiersen, Org 929 074 912) scoped to "maintain calibration infra" at $3,500/mo, which **winds down Sept 1, 2026**. The actual delivered surface - a stable SP test target, the public dual-asset faucet, daily monitoring, and full nv-upgrade validation - is broader than that line funds, with no successor after August. A bounded ProPGF commitment formalises the surface with public SLOs and public reporting, funds the builder on-ramp + API roadmap, and carries the surface past the side-letter wind-down without a gap. ### End-of-grant outcome Calibration stays funded, monitored, and stable across the next 6 months. The SP test target stays online, steady, and decentralization-compliant. Every `nv` upgrade gets a green Calix validation publicly within 24h of activation. The faucet holds its published uptime + drip-latency SLOs on both assets. The builder on-ramp tooling and public builder API ship and carry real builder traffic. Operations runbooks are maintained continuously (not back-loaded), and a 6-month sustainability report closes the grant out so a successor operator could pick up the suite from public docs alone. > **Note on provenance (externally verifiable).** > > * **The SP is on-chain on CalibrationNet:** [t0143103](https://calibration.filfox.info/en/address/t0143103), with long-running power history and the recent decentralization-driven reduction visible on-chain. > * **[faucet.reiers.io](https://faucet.reiers.io) has been live since April 2026.** Drip transactions are on-chain on calibration; the dispenser wallet's history pre-dates this application by ~6 weeks. > * **[calix.reiers.io](https://calix.reiers.io) shipped end-to-end during the nv28 "Fire Horse" calibration activation on 2026-05-07** - panel green automatically post-activation, 41 messages migrated, 0 failures; the activation event is on-chain and timestamped. > * The umbrella repo [Reiers/plumbline](https://github.com/Reiers/plumbline) has been public since 2026-04-29. > * **External user proof:** SynapS3 adopted Plumbline as a default funding endpoint ("Perfect, that's exactly what I needed. I'll integrate the endpoint into SynapS3").

$17,000.00

Indirect

Indirect

Indirect

**Output** -> a continuously-funded, monitored, human-on-call CalibrationNet stability surface. A stable, decentralization-compliant SP test target (`t0143103`, under 1/3 QAP) stays online and steady. Plumbline keeps developers funded with both tFIL and USDFC. Calix validates every `nv` upgrade end-to-end and surfaces the result publicly within 24h of activation. New builder on-ramp tooling + a public builder API let teams self-onboard to calibration funding in one flow. **Outcome** -> FoC integration work (`synapse-sdk`, FilecoinPay, PDPVerifier client integrations, flagship-client onboarding rehearsals) proceeds without the friction of a broken or under-funded calibration stage. Developers don't lose hours per week chasing faucet drips that didn't land, hunting a stable SP to test PDP against, or triaging a calibration upgrade with no public validation. **Impact** -> shorter cycle time from "we built a paid-deal integration" to "it's on mainnet." The KPI move is **upstream** - we don't add a paid deal directly; we make every team *trying* to add one go faster by removing the calibration-stage tax, and we keep `nv` upgrades from regressing on the way to mainnet.

| Metric | Data source | How it's measured | Target (end of grant) | |---|---|---|---| | Plumbline monthly uptime | Public status page at faucet.reiers.io/status | 1 - (minutes-down / minutes-in-month) | **>= 99%** each month | | Plumbline drip success rate | Server-side logs (monthly public summary) | successful drips / total drip requests | **>= 98%** | | Calix nv validation within 24h of activation | Public Calix timestamps + on-chain nv activation block | hours between activation block and dashboard-green | **<= 24h, every nv** | | SP test-target: availability + decentralization | On-chain power on calibration.filfox.info + status page | % of grant period t0143103 online & steady, held < 1/3 of calibration QAP | **>= 99% uptime, < 1/3 QAP** | | Builder on-ramp adoption | Public builder API request logs (monthly summary) | # of distinct builders self-onboarding via the on-ramp | **report monthly, growing** | | SynapS3 + downstream success via Plumbline | SynapS3 release-notes / dev-channel citation of faucet.reiers.io | citation present: yes/no | **Yes, retained** |

1. **Andrew Jackson (@snadrus)** - Curio core team (FilOz). Familiar with the Calix + nv28 "Fire Horse" rollout and the broader operational record. Contact via Slack (@snadrus). 2. **LexLuthr** - Curio core team (FilOz). Familiar with the CalibrationNet operational surface from the Curio team's own use of it. Contact via Slack (@LexLuthr). 3. **Lee - SynapS3 integrator.** Direct downstream consumer of Plumbline: "Perfect, that's exactly what I needed. I'll integrate the endpoint into SynapS3." First real public adopter of the dual-asset (tFIL + USDFC) faucet pattern, and the named first consumer of the M5 mainnet preview.

**Lead engineer / operator: Nicklas Reiersen (TSE Reiersen, Org. 929 074 912).** 8+ years in the Filecoin ecosystem. SAFT participant 2017; community contributor since 2019. Protocol Labs 2021-2023 (Lotus + Boost technical-support engineer). Curio core team 2023-2026 (through engagement close-out May 2026). Active member of all MinerX cohorts; first-name relationships with a substantial fraction of the current mainnet SP set. Marketing degree alongside the engineering. [github.com/Reiers](https://github.com/Reiers) · [linkedin.com/in/nicklas-reiersen](https://linkedin.com/in/nicklas-reiersen) · [reiers.io](https://reiers.io/) **Infrastructure footprint (Norwegian datacenter):** - Production fleet hosting the CalibrationNet SP test target (`t0143103`), the public faucet, and the Calix validation dashboard. When decentralization guidance called for a smaller footprint, the fleet was rebalanced accordingly (backup SP retired, primary trimmed under 1/3 QAP) - the capacity to run more exists; the deliberate choice is to stay decentralization-compliant. - Multiple independent internet providers for redundancy at the colocation site - this underpins the SP, faucet, and Calix uptime story. - Datacenter operations partner maintains the physical fleet on a long-standing colocation contract; software-side operations sit with TSE Reiersen. **On key-person risk (addressing it directly):** this is a solo engineering operation, so the mitigations are built into the milestones, not deferred. Everything is public; a living operations runbook is maintained from M1 (not back-loaded to M6); a successor-operability dry-run happens mid-grant (M5); and two Curio-core advisors with deep context (LexLuthr, @snadrus) exist. The M6 goal is that a successor operator could pick up the whole suite from public docs alone. > **Same operator as our Curio Core + Lantern application** ($130k, submitted). That application funds the engineering salary line and explicitly scopes the calibration-ops bandwidth - Plumbline, Calix, the mainnet SP operations - to this proposal at proposal-specific rate. The two reconcile by design; neither cross-funds the other. **Technical advisors (named, non-binding):** LexLuthr (Curio team) · Andrew Jackson / @snadrus (Curio team).

**External dependencies** - **USDFC mainnet supply** for the M5 mainnet-preview drip - gated on modest USDFC liquidity on the dispenser address. *Mitigation:* M5 is explicitly capped (e.g. 5 USDFC/address/week, hard monthly cap); the working capped endpoint (with a named first consumer, SynapS3) is the deliverable, not unconditional supply. - **Calibration upgrade cadence** - we can't predict exactly when the next `nv` lands; M3-M6 nv-validation is tied to "whatever lands in the window." *Mitigation:* Calix is reusable across upgrades; if no `nv` lands in a milestone window, that milestone delivers a dry-run validation report against the latest manifest. - **Datacenter / colocation reliability** for the SP test target and public services. *Mitigation:* long-standing operator relationship; multiple independent uplinks; the SP has stayed online & steady across nv boundaries to date. **Technical risks** - **Faucet abuse / Sybil** - public-path drips are gated by Cloudflare Turnstile + 24h per-asset per-address cooldown, but determined attackers can rotate addresses. *Mitigation:* per-asset hard caps on the dispenser balance, alerting on drip-rate anomalies, key-based auth for high-volume consumers (CI, SDK integrations). The same capped-exposure machinery is what the M5 mainnet preview hardens. - **Builder API / on-ramp scope is net-new build** (M2 onward). *Mitigation:* the umbrella README already names `plumbline-monitor` and `plumbline-cli` in the build order; this is execution of an existing roadmap, not a greenfield design. **Team risks** - **Solo operator (key-person risk).** *Mitigation:* everything is public; a living operations runbook is maintained continuously from M1 (not back-loaded); a mid-grant successor-operability dry-run (M5) verifies a second operator could stand up / recover each service from docs alone; advisors with deep context (LexLuthr, @snadrus) exist. - **Single operator across both TSE proposals** (App 1 Curio Core + Lantern, App 2 this one). *Mitigation:* the proposals are intentionally separable; if reviewers fund only one, neither cross-funds the other.

Three things worth surfacing for reviewers: **(1) This is the successor to an existing, cleanly-delivered grant - not a new ask out of nowhere.** The CalibrationNet stability surface is funded today by a Curio/Blueshift side letter (TSE Reiersen, Org 929 074 912) at $3,500/mo for June, July and August 2026, for "maintaining Filecoin calibration infra work." That side letter winds down Sept 1, 2026 unless a successor is executed by Aug 15. This proposal IS that successor. **(2) No double-funding - the dollars don't overlap.** The target rate for the full surface (the stable SP test target, the public tFIL+USDFC faucet, daily monitoring, nv-upgrade validation, and the new builder on-ramp + API) is $4,000/mo. For the two months the side letter still covers (Jul + Aug), ProPGF is asked for only the $500/mo delta on top of the side letter's $3,500. From September, when the side letter has sunset, ProPGF funds the full $4,000/mo. The milestone table reflects this exactly: $500 + $500 + $4,000 x 4 = $17,000 over six months. ProPGF only ever pays for dollars the side letter does not. **(3) The single-SP topology is intentional and decentralization-driven.** Plumbline previously ran two SP test targets. On calibration governance guidance I reduced footprint - retired the backup SP and trimmed the primary under 1/3 of network QAP - so a single-operator outage can't stall calibration F3 finality. The stable-test-target function is fully intact on one right-sized SP; the change is compliance with network guidance, not a reduction in what this grant delivers.

Well-structured form.