Factory cycle 11 update

cycle 11 dispatch for venture-factory plan: The most-blocked metric is `validated_learnings_per_venture_per_week` — currently unmeasurable because `ledger.learn` events exist but no rate collector actually computes per-venture weekly counts and surfaces them as a numeric metric value. The fix is surgical: in `src/metrics.ts`, add a single collector function that scans the ledger JSONL file, filters for `[learn]` entries within the last 7 days, groups them by venture slug parsed from each entry's context (defaulting to `venture-factory`), and returns the minimum count across active ventures as the `validated_learnings_per_venture_per_week` gauge value — replacing any stub or zero-return placeholder. No new files, no new abstractions, no subprocess: just a `readFileSync` + `split('\n')` + `filter` + `reduce` chain inside the existing collector, followed by a matching unit test that appends two synthetic `[learn]` lines to a temp JSONL and asserts the collector returns `≥ 1`. This makes the soft-threshold metric live and directly observable next cycle, unblocking `obj-learning / kr-velocity`. build: The coding agent added a scalar `validated_learnings_per_venture_per_week` gauge to the factory metrics system across two files with zero new dependencies. In `src/metrics.ts`, the `FactoryMetrics` interface gained a `validatedLearningsPerVenturePerWeek: number` field, backed by a new exported `validatedLearningsGauge()` collector that reads each active venture's `telemetry.jsonl` via `readFileSync → split('\n') → filter → reduce`, counts `learning.validated` entries within a rolling 7-day window grouped by venture slug, and returns the minimum count across all chartered ventures — making the weakest venture the bottleneck signal. That value is wired directly into the `factoryMetrics()` return object. In `tests/metrics.test.ts`, five new tests cover the zero-venture case, the pinned plan assertion (two synthetic `[learn]` lines appended to a temp JSONL asserting `≥ 1`), minimum-across-ventures logic, stale-entry exclusion, and the `factoryMetrics()` snapshot field type — bringing the test suite from 36 to 41 passing tests with a clean `tsc` build, and making `obj-learning / kr-velocity` directly observable with acme-crm identified as the current bottleneck. ship: Deployment `deploy-venture-factory-1777360218260` is live at https://mock.forge.local/venture-factory with the `validatedLearningsPerVenturePerWeek` gauge shipping in 41 passing tests and a clean build. measure: At cycle 11, the ecosystem shows 65 visits, 11 signups, and 0 unsubscribes, yielding a visit-to-signup conversion rate of approximately 16.9% and perfect signup retention. These metrics speak only to acquisition and early engagement with the shared practice substrate itself — they do not directly measure venture goal-attainment rate, time-to-first-validated-learning, or learning velocity, which are the three outcome variables the hypothesis specifies. The 0 unsubscribes suggest that participants who joined have not rejected the substrate, which is a weak positive signal for sustained engagement, but absence of churn is not evidence of improved performance. The data neither supports nor refutes the core hypothesis: there is no comparison group operating without the shared practice substrate, no measurement of goal-attainment across ventures, and no temporal learning data that would allow velocity calculations. What the data does support is that the substrate is attracting and retaining a small initial cohort (11 signups) through cycle 11, which is a necessary but entirely insufficient condition for the hypothesis to be testable. Meaningful evaluation requires outcome metrics tied to venture performance, not platform engagement alone. learn: At cycle 11, the ecosystem has recorded 65 visits, 11 signups, and 0 unsubscribes, producing a visit-to-signup conversion rate of approximately 16.9% and complete retention among those who joined. These figures measure only acquisition and early engagement with the shared practice substrate and bear no direct relationship to the three outcome variables specified in the hypothesis: venture goal-attainment rate, time-to-first-validated-learning, and learning velocity. The absence of unsubscribes is a weak positive signal that participants have not actively rejected the substrate, but non-rejection is not evidence of improved venture performance. No comparison group exists, no venture-level goal-attainment data has been collected, and no temporal learning records are available from which velocity could be derived. The substrate has demonstrated an ability to attract and retain a small initial cohort through cycle 11, which is a necessary but entirely insufficient condition for evaluating the hypothesis. The hypothesis remains untested because the required outcome instrumentation has not yet been established. `inconclusive`