You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Tracking issue for launch. Launch means: the page at filecoin.cloud/ipfs2filecoin captures a pasted CID list, shows an estimate, and hands the user's agent the runbook (migrate.md); the agent drives the CLI end to end with no cap; the hosted console covers small self-service runs within its cap. All of it documented well enough that nobody needs to ask us how to use it. This follows the 2026-07-28 product decision: the page is a teaser and router, the CLI is the migration engine.
Cap rationale, measured 2026-07: sequential 32-piece AddPieces commits ran about a minute each including confirmation, and the one measured provider pull ran ~0.18 MiB/s, so a run finishing under 10 minutes is roughly 500 CIDs. Shipped as 500 items / 1 GiB.
hosted run cap: 500 items structural; byte wall demoted to a 1 GiB ceiling (piece/upload unit) + a latched >10-minute projected-time advisory steering to the CLI
per-piece pull status readable from the CLI (status shows the local pull record; report stays chain truth)
user guide in docs/, README cleanup
wording pass over app and docs (product owns the message and CTAs; naming decision pending from the same pass)
web analytics via Plausible: pageview, run-completed, cli-steer, and contact events wired in app/src/analytics.ts, reporting into the existing filecoin.cloud site.. so there is no separate site to create and the old 404 is gone
CLI install metrics via @scarf/scarf (dependency + README opt-outs on main; scarf.sh registered; data starts with the next npm release)
operational telemetry: error tracking + step/abandon funnel behind the same hosted-only gate; BetterStack source ipfs2foc dApp in FOC WG, wired via repo variables into the Pages build
CLI stage telemetry: one anonymous event per command (name and outcome only) to the same BetterStack source; DO_NOT_TRACK and SCARF_ANALYTICS=false honored; disclosed in --help and in migrate.md (744c250; ships with the next npm release)
flip the deployed app default network to mainnet (2861cba; the launch-invariants deploy gate now asserts mainnet on every deploy)
contact-us / feedback affordance: persistent footer link + tails on all four boundary notices, pointing at https://filecoin.cloud/contact, clicks tracked
intake validation feedback: invalid-CID count inline at paste time (invalidCidStrings + InvalidCidNote)
landing fit statement: caps, wallet + USDFC/FIL prerequisites, and the CLI pointer sit in the lede before any wallet step; the caps sentence is asserted by the launch-invariants deploy gate
aggregate failure summary above the per-row detail: count of unfetchable items with retry, gateway, CLI, and contact options (FailureSummary)
storage-duration choice at the cost step: a "store for N months" control driving the estimate, aligned with the self-funding epic's funded-until concept
scripted CLI-flow rehearsal on calibration (scripts/rehearse-calibration.sh): probe, analyze, payments gate, upload, report ran end to end with a funded wallet on 2026-08-24; 6/6 CIDs committed on both provider copies (data sets 25443 and 25442), zero discrepancies. The dry runs also surfaced and fixed three exit-code bugs (upload aa307a7, probe 2dc8d5d)
console-surface rehearsal: session grant, byte-cap steer, submit ETA (the app surfaces the audit couldn't reach; the CLI rehearsal above does not exercise them)
host at filecoin.cloud/ipfs2filecoin (the ipfsto.filecoin.cloud subdomain is dropped), update the official Filecoin developer docs to point at it
Not in launch scope
Fast-follows, tracked separately: p2p retrieval sources for console prepare (#59), pause/resume (#60), foc2foc (#56), provider reassignment (#49), aggregation via PoDSI (#61, parked on upstream), pinning-service API ingestion (no issue yet).
Description
Tracking issue for launch. Launch means: the page at filecoin.cloud/ipfs2filecoin captures a pasted CID list, shows an estimate, and hands the user's agent the runbook (migrate.md); the agent drives the CLI end to end with no cap; the hosted console covers small self-service runs within its cap. All of it documented well enough that nobody needs to ask us how to use it. This follows the 2026-07-28 product decision: the page is a teaser and router, the CLI is the migration engine.
Cap rationale, measured 2026-07: sequential 32-piece AddPieces commits ran about a minute each including confirmation, and the one measured provider pull ran ~0.18 MiB/s, so a run finishing under 10 minutes is roughly 500 CIDs. Shipped as 500 items / 1 GiB.
Tasks
statusshows the local pull record;reportstays chain truth)ipfs2foc dAppin FOC WG, wired via repo variables into the Pages build--helpand in migrate.md (744c250; ships with the next npm release)invalidCidStrings+InvalidCidNote)FailureSummary)Not in launch scope
Fast-follows, tracked separately: p2p retrieval sources for console prepare (#59), pause/resume (#60), foc2foc (#56), provider reassignment (#49), aggregation via PoDSI (#61, parked on upstream), pinning-service API ingestion (no issue yet).