Skip to content

docs(gates): document neutralizationGate — the causal placebo control (0.107.0) #327

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
drewstone merged 1 commit into from
Jul 8, 2026
Merged

docs(gates): document neutralizationGate — the causal placebo control (0.107.0) #327

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -122,7 +122,7 @@ Each example: `README.md` + a single `index.ts` runnable via `pnpm tsx`. Prints

| Subpath | What it gives you |
|---|---|
| `…/contract` | **The headline, frozen surface — new code starts here.** `defineAgentEval`, `selfImprove`, `analyzeRuns`, `runEval`, `runCampaign`, `runImprovementLoop`, `diffRuns`; intake adapters (`fromFeedbackTable`, `fromOtelSpans`); proposers (`gepaProposer`, `evolutionaryProposer`); gates (`defaultProductionGate`, `heldOutGate`, `paretoSignificanceGate`, `composeGate`); the deployment-outcome store; storage; and the five core types `Scenario` / `Dispatch` / `JudgeConfig` / `SurfaceProposer` / `Gate`. |
| `…/contract` | **The headline, frozen surface — new code starts here.** `defineAgentEval`, `selfImprove`, `analyzeRuns`, `runEval`, `runCampaign`, `runImprovementLoop`, `diffRuns`; intake adapters (`fromFeedbackTable`, `fromOtelSpans`); proposers (`gepaProposer`, `evolutionaryProposer`); gates (`defaultProductionGate`, `heldOutGate`, `paretoSignificanceGate`, `neutralizationGate`, `composeGate`); the deployment-outcome store; storage; and the five core types `Scenario` / `Dispatch` / `JudgeConfig` / `SurfaceProposer` / `Gate`. |
| `…/hosted` | `createHostedClient` / `hostedClientFromEnv` + the wire types to ship eval-run events + trace spans to a hosted orchestrator (ours or your own implementation of the spec) |
| `…/adapters/otel` | `createOtelBridge` — forwards OpenTelemetry-shape spans into the hosted-tier ingest, no `@opentelemetry/*` dependency |
| `…/adapters/langchain` | Wrap any LangChain `Runnable` as a `Dispatch` (or `JudgeConfig`), no `@langchain/core` peer dep |
Expand Down
6 changes: 5 additions & 1 deletion docs/improvement-glossary.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -196,4 +196,8 @@ Every entrant is re-scored on the **same** holdout with the **same** judges, so
- Do not put eval logic inside a proposer — scoring lives in `dispatch` + `judges`, proposing lives in the proposer.
- Do not let a proposer read held-out judge scores — `ProposeContext` makes that a compile error on purpose; a proposer that games the acceptance axis is an oracle, not an optimizer.
- Do not read `lift` without `result.power`/MDE — a "+4" on a valset too small to detect +4 is noise wearing a number.
- Do not look for a `compositeProposer` — `fapoProposer` is the composite; `compareProposers` is the race; `runImprovementLoop`/`improve` is the plug.
- Do not look for a `compositeProposer` — `fapoProposer` is the composite; `compareProposers` is the race; `runImprovementLoop`/`improve` is the plug.

### neutralizationGate — the placebo / content-causality control

Standard gates prove a candidate *beat baseline*. `neutralizationGate` proves the candidate's **content** caused the lift, not the extra prompt bytes: it blanks the candidate's added content to byte-length-matched filler, holds everything else fixed, and requires the lift to vanish. A fully-neutralized candidate that still scores is decorative and is rejected. Exports: `neutralizationGate` (`src/campaign/gates/neutralization-gate.ts`), `neutralizeText` (`src/campaign/neutralize.ts`). Since 0.107.0. Pair it with `heldOutGate` in any loop that promotes authored artifacts (prompts, tool docs, knowledge) so a lift that is really just added prompt size cannot be shipped.
Loading