@tangle-network/agent-app 0.43.11 → 0.43.12

package/.claude/skills/eval-campaign/SKILL.md

@@ -36,7 +36,7 @@ import {
 } from '@tangle-network/agent-app/eval-campaign'
 ```
 
-> Requires `@tangle-network/agent-eval >= 0.95.0` (peer). The scaffold composes the substrate downward; never import a product package from agent-eval (layering rule).
+> Requires `@tangle-network/agent-eval >= 0.100.0` (peer; current published is 0.107.0). The scaffold composes the substrate downward; never import a product package from agent-eval (layering rule).
 
 ## Minimal wiring (copy, then fill the three blanks)
 
@@ -89,7 +89,7 @@ if (result.gate.decision === 'ship') await ship(result.winnerSurface)
 | `baselineSurface` | — (required) | the surface the loop optimizes; start `''` |
 | `mutationPrimitives` | gepaProposer's own | your optimization levers (additive directives) |
 | `proposer` | `gepaProposer` | pass `evolutionaryProposer({ mutator })` for blind addendum rotation |
-| `gate` | `defaultProductionGate` (Δ 0.05) | `paretoSignificanceGate` for multi-objective; tune `deltaThreshold` for your rubric scale |
+| `gate` | `defaultProductionGate` (Δ 0.05) | `paretoSignificanceGate` for multi-objective; tune `deltaThreshold` for your rubric scale . To prove a held-out lift is real CONTENT and not just added prompt/mount FOOTPRINT, compose `neutralizationGate` (agent-eval >= 0.107.0, from `@tangle-network/agent-eval/campaign`) — a footprint-matched placebo arm. It needs `ctx.neutralizedJudgeScores` from `runImprovementLoop({ neutralize })`; `selfImprove`/this scaffold do not surface that option yet, so wire it at the `runImprovementLoop` level or re-export it in the scaffold first. |
 | `budget` | 3 gens × pop 2, 0.25 holdout | `budget.reps` (replicates → tighter CIs), `budget.promoteTopK`, `budget.holdoutScenarios` (explicit split), `budget.dollars` (cost cap) |
 | `expectUsage` | **`'assert'`** | the fail-loud backend-integrity guard. Leave at `'assert'` for real runs (a stub cell throws); set `'off'` ONLY for a deterministic offline/replay run |
 | `labeledStore` | off | capture every artifact + judge score (the dataset you ship + few-shot corpus); set `captureSource` (default `'eval-run'`) |

package/.claude/skills/improve-conductor/SKILL.md

@@ -11,7 +11,7 @@ You delegate the building to an agent holding `eval-architect` + `surface-evolut
 
 ## Invariant (non-negotiable)
 
-1. **Never promise or report a lift you cannot measure with valid paired evidence.** Surface the honest verdict: `ship` / `hold` / `need-more-data` / `invalid`. "Invalid" (incomplete or unpaired evidence) is a first-class outcome — say it plainly, never paper over it with a survivor-mean number.
+1. **Never promise or report a lift you cannot measure with valid paired evidence.** Surface the honest verdict: `ship` / `hold` / `need-more-data` / `invalid`. A paired, significant lift is still not shippable until a footprint-matched placebo shows the gain comes from the CONTENT, not from added prompt/mount footprint — the substrate's `neutralizationGate` / `runImprovementLoop({ neutralize })` (`@tangle-network/agent-eval/campaign`). A lift that a neutralized twin reproduces is footprint, not improvement — refuse it. Route this check through `measurement-validation`. "Invalid" (incomplete or unpaired evidence) is a first-class outcome — say it plainly, never paper over it with a survivor-mean number.
 2. **Refuse below the data threshold, and say why** — "I have N real outcomes; I won't optimize below M. Here's how to get to M." A refusal with a reason builds more trust than a fabricated win.
 3. **Route correctly.** Improvable by surface-tuning → dispatch `surface-evolution`. Needs a new capability or architecture → escalate and say so; don't pretend tuning will fix a structural gap.
 4. **No optimization spend before the target is confirmed and the measurement is real.** If there is no improvement infrastructure yet, you do NOT improvise a metric and start spending — you dispatch `eval-bootstrap` to BUILD a validated, externally-grounded harness first. The gate between "build the apparatus" and "spend optimizing" is yours to hold.

package/.claude/skills/measurement-validation/SKILL.md

@@ -13,7 +13,7 @@ Held by both the orchestrator (`improve-conductor`) and the builder (`eval-archi
 
 1. **Refuse to optimize if CV(metric) > the target delta.** If the run-to-run noise is bigger than the effect you're paying to move, the metric *cannot* validate the change — raise reps or fix the metric first. Do not tune against noise.
 2. **Refuse to report a lift over INCOMPLETE or UNPAIRED evidence.** Every held-out scenario must have a non-errored cell on *both* the baseline and the candidate side. Below the paired-n floor (≥3), the run is **invalid**, not a verdict. A lift computed over survivors is worse than no number.
-   > **Enforced by** `trustVerdicts` from `@tangle-network/agent-app/eval-campaign` — the after-gate: IRR floor + per-item rater spread (within-item, never pooled) + survivor floor, with each failed check named in `trustReasons`.
+   > **Enforced by** `trustVerdicts` from `@tangle-network/agent-app/eval-campaign` (rater-trust dimension: IRR floor + per-item rater spread, within-item never pooled, + survivor floor, each failure named in `trustReasons`), composed with the agent-eval statistical gates from `@tangle-network/agent-eval/campaign`: `powerPreflight` (before-gate for Invariant 1 — refuse to greenlight spend when the metric is underpowered vs the target delta), `heldOutGate` / `heldoutSignificance` (paired-bootstrap CI over the held-out split — the after-gate that would have caught the +47 by refusing an unpaired lift), and `neutralizationGate` + `neutralizeText` (footprint-matched placebo gate — a held-out lift that does not survive `neutralizeText` came from added prompt/mount FOOTPRINT, not content, and must not be believed).
 3. **Every metric ties to a product-value claim** — "if this number moves, *this* user-visible outcome moves with it." No claim → it's a proxy → don't optimize it.
 4. **Below the data threshold of real outcomes, refuse to optimize** — state N and say why. You cannot improve what you have not yet observed enough of.
 

package/.claude/skills/skill-evolution/SKILL.md

@@ -24,7 +24,7 @@ Each skill's *judgment* surface is itself an evolvable surface, optimized by the
 
 > *Did following this skill produce an eval that yielded real held-out lift, with no critical-dimension regression?*
 
-Above a data threshold of real runs, the skill proposes revisions to its own judgment — gated identically (held-out, critical-dimension floor, paired-n ≥ floor). **Invariants are the frozen surface; judgment is the evolvable surface.** A skill improving itself is just `surface-evolution` pointed inward.
+Above a data threshold of real runs, the skill proposes revisions to its own judgment — gated identically (held-out, critical-dimension floor, paired-n ≥ floor, and a footprint-matched placebo — the neutralization gate — that proves the lift came from content, not from added prompt/mount footprint). **Invariants are the frozen surface; judgment is the evolvable surface.** A skill improving itself is just `surface-evolution` pointed inward.
 
 ## The north-star: the agent builder, closed-loop
 
@@ -39,4 +39,4 @@ The fleet — **legal, tax, gtm, creative, insurance** — is the training distr
 - A skill that lists steps but has **no self-test** — you can't tell if following it worked.
 - An "invariant" that's really a **judgment call in disguise** — over-constraining; it should live in Judgment so the agent can adapt it per product.
 - A judgment surface with **no evolves-by hook** — it will rot, and nothing will notice.
-- A reported lift with **no held-out or no paired-n** — the slot machine. This is the one that ends the product.
+- A reported lift with **no held-out or no paired-n, or no footprint-matched placebo (neutralization)** — the slot machine. This is the one that ends the product.

package/.claude/skills/surface-evolution/SKILL.md

@@ -7,12 +7,12 @@ description: Optimize ONE evolvable surface (a prompt section / tool config) aga
 
 You are a closed-loop controller for agent quality. **Sensor** = the eval (built by `eval-architect`, certified by `measurement-validation`). **Controller** = the proposer that proposes surface rewrites. **Actuator** = promotion (writing the surface to the live agent). **Safety interlock** = the gate. The interlock is the entire point: it prefers *under-promotion* to Goodhart. A loop that ships every apparent gain is worthless; a loop that ships only evidence-backed gains is the product.
 
-The engine exists in the substrate (`@tangle-network/agent-eval/contract` `selfImprove` / `runImprovementLoop`, `gepaProposer`, `defaultProductionGate`) — re-exported via `@tangle-network/agent-app/eval-campaign`. **Do not rebuild it.** This skill is how you wire and run it safely.
+The engine exists in the substrate (`@tangle-network/agent-eval/contract` `selfImprove` / `runImprovementLoop`, `gepaProposer`, `defaultProductionGate`, `neutralizationGate`, `composeGate`) — re-exported via `@tangle-network/agent-app/eval-campaign`. **Do not rebuild it.** This skill is how you wire and run it safely.
 
 ## Invariant (non-negotiable)
 
 1. **Optimize exactly ONE surface that production renders identically.** The artifact you mutate offline must be the artifact the live agent loads — one source, rendered both places (e.g. an evolvable prompt section materialized from a single file into the live system prompt). If offline and online diverge, the lift is fictional the moment it ships.
-2. **Gate promotion on a held-out split AND a critical-dimension floor.** Never promote a net composite gain that regresses a guarded dimension (safety, hallucination, the regulated invariant). A +10 composite that loses 30 on hallucination is a regression, not a win.
+2. **Gate promotion on a held-out split AND a critical-dimension floor.** Never promote a net composite gain that regresses a guarded dimension (safety, hallucination, the regulated invariant). A +10 composite that loses 30 on hallucination is a regression, not a win. **AND gate on a footprint placebo.** A held-out lift proves the candidate beat baseline; it does NOT prove the lift came from the surface's CONTENT rather than the extra prompt/mount FOOTPRINT it added. Compose `neutralizationGate` (from `@tangle-network/agent-eval/campaign`, wired via `runImprovementLoop({ neutralize })`) AFTER the significance gate — `composeGate(heldOutGate({...}), neutralizationGate({...}))` — so a candidate whose content-blanked, footprint-matched twin reproduces more than `maxDecorativeFraction` (default 0.5) of the lift is HELD as decorative, no matter how large or significant its raw lift.
 3. **Budget is a hard ceiling and cost-aware** — skip cells beyond the ceiling, never abort. The user's spend maps to generations × candidates × reps: $0.20 buys one quick generation; $50 buys a wide search with tight CIs.
 4. **Never evolve a frozen surface.** The regulated invariants — human-in-the-loop, the compliance gate, auth/RBAC — are off-limits. Declare exactly what is evolvable; everything else the loop must not touch.
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tangle-network/agent-app",
-  "version": "0.43.11",
+  "version": "0.43.12",
   "packageManager": "pnpm@10.33.4",
   "description": "Application-shell framework for Tangle agent products: a bounded tool loop, the structured agent→app tool side channel, integration-hub client, per-workspace billing, and crypto — composed over the Tangle agent substrate through typed seams.",
   "keywords": [