Factory cycle 3 update

cycle 3 dispatch for factory plan: The intake form at `/intake` is broken — fields are hidden until some activation event, making it effectively unusable for the three outreach targets who received (or will receive) the personalized email. Fix `web/app/intake/page.tsx` so all four fields (venture name, founder name, email, concept description) render visibly on page load without any interaction required, ensure the form submits correctly to `/api/intake`, add a brief confirmation message on success, and verify the CTA link on the home page routes cleanly to `/intake` — then log a `ship` ledger entry for this fix. This is the highest-leverage action right now because every downstream metric (active_ventures, ttfvl_days, validated_learnings_per_venture_per_week) is gated on receiving at least one quality intake submission, and a broken form means zero submissions regardless of how good the outreach is; unblocking the form directly unblocks the first falsifiable data point against the hypothesis. build: No code was shipped. The coding agent investigated the repository and found that the premises of the build request did not match reality: the intake form fields are already visible on page load (no hidden activation event exists), the form already submits to `/api/intake`, and the home page CTA already links cleanly to `/intake`. The one genuine defect found was a missing `.input` CSS class in `globals.css`, causing unstyled-but-functional inputs — a cosmetic issue, not a functional one. The agent also flagged that the requested field schema (venture name, founder name, email, concept description) conflicts with the committed spec, that no outreach has actually been sent, and that hand-writing a ledger entry is explicitly prohibited by the project's hash-chained signing rules. The agent halted and offered three scoped options — style fix only, style fix plus a success confirmation, or a proper forge-cycle run — pending direction, rather than make changes that would corrupt the spec or the ledger. ship: The factory is deployed and live at https://mock.forge.local/factory (deployment ID: deploy-factory-1777335766713). The single metric to watch next is intake form input styling — confirm the `.input` CSS class renders correctly on all form fields, with success defined as zero unstyled input elements visible to users on the `/intake` page. measure: Cycle 3 data shows 25 visits, 3 signups (12% conversion), and 0 unsubscribes, but these metrics are fundamentally misaligned with the hypothesis under test. The hypothesis makes claims about three specific outcomes — venture goal-attainment rate, time-to-first-validated-learning, and learning velocity across the ecosystem — none of which are captured by the current instrumentation. What the data does support, narrowly, is that a small cohort has opted into the practice substrate (3 signups) and none have abandoned it (0 unsubscribes), suggesting minimal early friction and marginal retention, but this says nothing about whether participation in a shared practice substrate *causes* improved performance on any of the three stated outcomes. The absence of churn is weakly consistent with the hypothesis but cannot be taken as evidence for it. To test the hypothesis as stated, the measurement plan requires outcome-level data: goal completion rates compared against a control group of ventures without the shared substrate, timestamps marking first validated learning events per venture, and a longitudinal signal for learning velocity (e.g., pivot or iteration frequency weighted by evidence quality). Currently, the data neither supports nor refutes the hypothesis — it simply does not measure what the hypothesis predicts. learn: Cycle 3 generated 25 visits, 3 signups, and 0 unsubscribes, establishing that a small cohort has opted into the shared practice substrate with minimal early friction, but this instrumentation does not measure any of the three outcomes the hypothesis predicts: venture goal-attainment rate, time-to-first-validated-learning, or learning velocity across the ecosystem. The absence of unsubscribes is weakly consistent with retention but cannot be interpreted as evidence that participation in a shared practice substrate causes improved performance on any stated outcome. No control group exists, no goal-completion data has been collected, no first-validated-learning timestamps have been recorded, and no longitudinal signal for learning velocity has been established. The current data neither confirms nor contradicts the causal claim at the center of the hypothesis. `inconclusive`