Evaluate landing page

Landing Page Eval — Orchestrator

<!-- BUDGET_EXCEPTION: Eval skills carry artifact-schema-as-contract (8 body sections + 8-col results row + cross-stack consumer contract) that is load-bearing and cannot move to references/. Cycle ledger discipline requires the schema be visible in the SKILL.md body. ~300 tokens over the standard cap is the legitimate cost. -->

Evaluation skill. Converts launched landing-page evidence into a cycle snapshot, a ledger row, and narrowly-scoped next action inside an existing eval loop.

Core Question: "Did this landing-page cycle create measurable signal strong enough to keep, discard, watch, or block, and what should the next strategy/execution skill know?"

Why, philosophy, methodology, when NOT to use, history: references/playbook.md [PLAYBOOK].

Critical Gates

1. Existing eval loop required. If .forsvn/loops/[slug]/program.md + context.md absent → NEEDS_CONTEXT, recommend /run-pipeline. This skill does not create loops.
2. Measurement evidence required. Not a generic heuristic audit. Require ≥1 metric source, measurement window, and current value for the loop's primary metric.
3. One primary metric decides the ledger row. Secondary metrics and qualitative evidence explain diagnosis; they do not override the primary metric unless program.md defines an explicit guardrail failure.
4. No fabricated analytics. Unknown values stay unknown. Manual notes only when labeled operator-supplied + tied to date/window/source.
5. Attribution confidence must be explicit. Every verdict includes sample size or traffic volume when available, baseline comparability, confounders, and confidence: high | medium | low | blocked.
6. Evaluation does not redesign. Recommend next changes; route actual brief/revision work to brief-landing-page.

Responsibility Split

/run-pipeline owns loop setup, program.md, context.md, results.tsv schema, durable learning ledger. This skill owns post-launch landing-page evidence snapshots for a loop cycle. /brief-landing-page owns new-page and redesign briefs after an eval identifies what should change.

Inputs

Required: loop slug/path · page URL/route · measurement window · primary metric value + source.

Recommended/optional: baseline or prior cycle row · traffic/sample size · guardrail metrics (bounce, form completion, qualified lead rate, revenue, page speed, spend efficiency) · experiment notes (variant split, assignment, test integrity) · qualitative evidence (heatmaps, recordings, comments, sales notes) · strategy/execution artifacts (what was changed).

Outputs

Primary artifact: .forsvn/loops/[slug]/evals/YYYY-MM-DD-cycle-N.md.

Side effects: append one row to .forsvn/loops/[slug]/results.tsv via scripts/append-loop-result.ts; update learnings.md ONLY for high-confidence keep/discard reusable lessons (critic-gated); run manifest-sync.ts.

Agent Manifest + Dispatch

4 sub-agents: Layer 1 parallel (Metric Ingest + Diagnosis) → Layer 2 (Recommendation) → Layer 3 (Critic). Critic FAIL → revise once; still FAIL → no ledger row + return BLOCKED. Cycle number = last results.tsv cycle + 1 unless operator names a cycle that has no existing eval artifact. Full per-agent focus + per-layer dispatch tables + critic revision-cycle semantics + side-effects ALL-OR-NOTHING on FAIL: references/procedures/dispatch-mechanics.md [PROCEDURE].

Pre-Dispatch

Canonical: references/_shared/pre-dispatch-protocol.md + _shared/eval-loop-spec.md. Hard-blocks: missing program.md/context.md → NEEDS_CONTEXT + recommend /run-pipeline; no measurement evidence → BLOCKED; custom 10+ col results.tsv → warn + require hand-edit. Cold Start: 5 bundled questions (loop slug · page URL · window · primary metric value/baseline · what changed). Full read-order + warm/cold templates + --fast behavior: references/procedures/pre-dispatch.md [PROCEDURE].

Artifact Contract

• Path: .forsvn/loops/[slug]/evals/YYYY-MM-DD-cycle-N.md. Lifecycle: evaluation.
• Body sections (8): Title · Verdict · Evidence (6-col table) · What Changed · Diagnosis · Next Cycle Recommendation · Results Row · Learning Promotion.
• Results Row (8 cols): cycle date artifact primary_metric value baseline status description. status ∈ keep|discard|watch|blocked (Critic Hard Fail #6 otherwise).
• Cross-stack contract: consumed by future lp-eval cycles, brief-landing-page --rev=N+1, human loop reviewers. Schema changes require atomic update across _shared/eval-loop-spec.md + downstream callers.

Full byte-identical template + frontmatter spec + Evidence/Results/Learning formats + custom-schema limitation: references/format-conventions.md [PROCEDURE].

Results Row Helper

bun scripts/append-loop-result.ts "<loop slug>" \
  --artifact evals/YYYY-MM-DD-cycle-N.md \
  --metric "<primary metric>" --value "<current>" --baseline "<baseline>" \
  --status "<keep|discard|watch|blocked>" --description "<one sentence>"

Do not append on Critic FAIL — return BLOCKED instead.

Critic Override Protocol

When operator ships a cycle despite critic FAIL — log the override BEFORE writing artifact or ledger row: bun scripts/log-critic-override.ts --skill evaluate-landing-page …. Full protocol (invocation, 3-override escalation, two rules an override never relaxes): references/_shared/critic-override-protocol.md [PROCEDURE].

Anti-Patterns

references/anti-patterns.md [ANTI-PATTERN] — 10 lp-eval patterns + 4 cross-cutting marketing-stack rows. Most common: scope drift to redesign (Gate 6 + Critic "Boundary Control"), confidence inflation (Gate 5 + Critic "Attribution Honesty"), low-confidence learning promotion (Critic Hard Fail #7), missing primary metric source-window (Critic Hard Fail #2).

Worked Example

Landing-page CRO cycle (section-level bottleneck diagnosis from conversion evidence, critic PASS): references/examples/landing-page-eval-cycle-walkthrough.md.

Durable Rules (protected)

<!-- SLOW_UPDATE_START --> <!-- No pinned rules yet. Populate via the slow-update workflow (see references/slow-update-fence.md). Each pinned rule must (a) be procedural not instance-specific, (b) be earned from a regression or critic-flagged failure, (c) cite the artifact / decision record that justified pinning. --> <!-- SLOW_UPDATE_END -->

Completion Status

• DONE — eval artifact written, ledger row appended, critic passed
• DONE_WITH_CONCERNS — artifact + row written, but confidence low/medium or confounders material
• NEEDS_CONTEXT — missing loop or required metric evidence
• BLOCKED — contradictory data, filesystem failure, or critic failed after revision

References

• references/playbook.md, format-conventions.md, anti-patterns.md
• references/procedures/{pre-dispatch, dispatch-mechanics}.md
• references/_shared/{eval-loop-spec, before-starting-check, manifest-spec, mode-resolver, pre-dispatch-protocol, anti-sycophancy, artifact-contract-template, thin-critic-rubric, critic-override-protocol}.md
• references/_shared/execution-policy.md — session execution profile (single-vs-multi)
• 4 sub-agents in agents/; critic-agent.md holds the 6-dimension 0-10 rubric (Loop Fit / Metric Integrity / Attribution Honesty / Decision Discipline / Boundary Control / Ledger Correctness) + 4-tier Verdict + 7 Hard Fails
• Sibling coordination: brief-landing-page (this skill routes recommendations TO it but does not produce briefs); run-pipeline (owns loop scaffolding + ledger schema + durable learnings)