From 204481292e0bd2793c108260512891c234b578cb Mon Sep 17 00:00:00 2001
From: Myke Matos <123015330+shishiv@users.noreply.github.com>
Date: Wed, 26 Aug 2026 13:10:05 -0300
Subject: [PATCH 1/3] feat(pstack): give gh-stack inspect an ordered member
 list

Delivery membership is proven parent-first tuples from gh stack view JSON.
Graphite is removed from delivery contracts so stack proof has one backend.
---
 README.md                                     |   7 +-
 extensions/poteto-mode.ts                     |  23 +-
 skills/poteto-mode/SKILL.md                   |   7 +-
 .../poteto-mode/playbooks/autopilot-stack.md  |   2 +-
 skills/poteto-mode/playbooks/opening-a-pr.md  |   2 +-
 skills/poteto-mode/playbooks/shipping.md      |   8 +-
 src/delivery/adapters.ts                      | 169 ++++++--------
 src/delivery/authorization.ts                 |   3 +-
 src/delivery/types.ts                         |  36 ++-
 test/delivery/delivery.test.mjs               | 219 ++++++++++--------
 test/resources/resources.test.mjs             |   1 -
 11 files changed, 243 insertions(+), 234 deletions(-)

diff --git a/README.md b/README.md
index 58f63b5..063efdd 100644
--- a/README.md
+++ b/README.md
@@ -14,7 +14,7 @@ O `pi-pstack` preserva o método verification-first do projeto original: context
 - `/poteto-mode` persistente durante a sessão ativa.
 - Feature maps, artifacts de navegador e receipts vinculados ao `HEAD` exato.
 - Evals cegos entre modelos, com hard assertions que o judge não pode ignorar.
-- Stacked PRs com `gh stack` por padrão e Graphite como backend opcional.
+- Stacked PRs com `gh stack`.
 - Benny em modo draft-only: ele pode preparar uma draft PR, mas nunca faz merge ou deploy.
 
 ## Requisitos
@@ -26,8 +26,7 @@ O `pi-pstack` preserva o método verification-first do projeto original: context
 - Bun para as ferramentas locais que o utilizam.
 - `portless` para fluxos locais que expõem serviços.
 - Playwright Chromium para verificação de navegador.
-- `github/gh-stack` como backend padrão de stacked PRs.
-- Graphite `gt` somente quando o backend opcional for selecionado.
+- `github/gh-stack` como backend de stacked PRs.
 - Um provider externo de Slack e tracker para executar Benny contra serviços reais.
 
 ## Instalação
@@ -84,7 +83,7 @@ O modo sticky vale somente para a sessão ativa. O Pi restaura o estado pelo his
 
 ## Stacked PRs
 
-`gh stack` é o backend padrão. O adapter Graphite só aparece quando `gt` está instalado. O pacote traduz operações para os CLIs oficiais e não mantém um segundo grafo de branches.
+`gh stack` é o backend de delivery. O pacote traduz operações para o CLI oficial e não mantém um segundo grafo de branches.
 
 O merge atômico exige um receipt por PR até o alvo. Gere cada receipt no checkout limpo do respectivo `HEAD` e mantenha os artifacts em caminhos imutáveis disponíveis durante a validação final. Se um digest de uma camada anterior não estiver disponível no checkout atual, o merge falha fechado.
 
diff --git a/extensions/poteto-mode.ts b/extensions/poteto-mode.ts
index 20a9ac5..b7fb320 100644
--- a/extensions/poteto-mode.ts
+++ b/extensions/poteto-mode.ts
@@ -245,7 +245,7 @@ export default function potetoModeExtension(pi: ExtensionAPI): void {
     description:
       "Hash local verification, review, and eval artifacts into a receipt for the current repository HEAD.",
     parameters: Type.Object({
-      backend: Type.Union([Type.Literal("gh-stack"), Type.Literal("graphite")]),
+      backend: Type.Literal("gh-stack"),
       featureMapPath: Type.String({ minLength: 1 }),
       skillPath: Type.String({ minLength: 1 }),
       reviewPath: Type.String({ minLength: 1 }),
@@ -303,9 +303,9 @@ export default function potetoModeExtension(pi: ExtensionAPI): void {
     name: "pstack_delivery",
     label: "Run gated pstack delivery",
     description:
-      "Inspect a stack or run a receipt-gated gh-stack or Graphite mutation. Auto-merge also verifies the live PR head and checks.",
+      "Inspect a stack or run a receipt-gated gh-stack mutation. Auto-merge also verifies the live PR head and checks.",
     parameters: Type.Object({
-      backend: Type.Union([Type.Literal("gh-stack"), Type.Literal("graphite")]),
+      backend: Type.Literal("gh-stack"),
       operation: Type.Union([
         Type.Literal("inspect"),
         Type.Literal("prepare"),
@@ -344,8 +344,6 @@ export default function potetoModeExtension(pi: ExtensionAPI): void {
             paths.map((path) => loadEvidenceReceipt(path, ctx.cwd)),
           );
           if (params.operation === "auto-merge") {
-            if (params.backend !== "gh-stack")
-              return deliveryRejected("Graphite auto-merge lacks stack-wide receipt verification");
             if (!params.pullRequest)
               return deliveryRejected("auto-merge requires a pull request number");
             const stackResult = await pi.exec("gh", ["stack", "view", "--json"], {
@@ -414,18 +412,8 @@ export default function potetoModeExtension(pi: ExtensionAPI): void {
             return { exitCode: result.code, stdout: result.stdout, stderr: result.stderr };
           },
         };
-        const availableCommands = ["gh"];
-        if (params.backend === "graphite") {
-          const gt = await pi.exec("gt", ["--version"], { cwd: ctx.cwd, signal });
-          if (gt.code === 0) availableCommands.push("gt");
-        }
-        const backends = createDeliveryBackends({
-          runner,
-          availableCommands,
-        });
-        const backend = params.backend === "graphite" ? backends.graphite : backends.ghStack;
-        if (!backend)
-          return deliveryRejected("Graphite is unavailable; install and authenticate gt first");
+        const backends = createDeliveryBackends({ runner });
+        const backend = backends.ghStack;
         const operation = deliveryOperation(params);
         const result = await backend.execute(operation);
         return {
@@ -692,6 +680,7 @@ function deliveryOperation(params: {
     case "rebase":
       return { kind: "rebase" };
     case "auto-merge":
+      if (!params.pullRequest) throw new Error("auto-merge requires pull request");
       return { kind: "auto-merge", pullRequest: params.pullRequest };
     default:
       throw new Error(`unsupported delivery operation: ${params.operation}`);
diff --git a/skills/poteto-mode/SKILL.md b/skills/poteto-mode/SKILL.md
index fb80b18..b99fff7 100644
--- a/skills/poteto-mode/SKILL.md
+++ b/skills/poteto-mode/SKILL.md
@@ -17,9 +17,8 @@ managed worktrees isolate their edits. Resolve models through Pi profiles and
 agent roles, never through provider-specific slugs embedded in a skill.
 
 Delivery resources prepare evidence and receipts only. The default named
-backend is `gh stack`; Graphite is an optional named backend. Executable
-delivery behavior belongs to the later `src/delivery` adapters, not these
-skills.
+backend is `gh stack`. Executable delivery behavior belongs to the later
+`src/delivery` adapters, not these skills.
 
 ## Non-negotiables
 
@@ -139,7 +138,7 @@ A large or cross-cutting effort (a migration across many call sites, an ambitiou
 - **Authoring or modifying a skill.** Writing or editing a SKILL.md. `playbooks/authoring-a-skill.md`.
 - **Eval.** Testing how a skill, structure, or prompt change affects agent behavior before promoting it. `playbooks/eval.md`.
 - **Babysit.** Driving a PR or a stack to merge-ready: conflicts, review threads, CI. `playbooks/babysit.md`.
-- **Shipping.** The half after Babysit. Independently verifying a green stack, then handing the contiguous verified run to the default `gh stack` adapter (Graphite is optional). `playbooks/shipping.md`.
+- **Shipping.** The half after Babysit. Independently verifying a green stack, then handing the contiguous verified run to the `gh stack` adapter. `playbooks/shipping.md`.
 - **Autonomous run.** A long task to drive to completion without stopping ("run until done", "wait until X"). `playbooks/autonomous-run.md`.
 - **Orchestrate.** A standing project handed to one coordinator chat: multi-day, many stacked PRs, dozens to hundreds of subagents, minimal human turns ("run this whole project", "own this migration until it lands"). Distinct from Autonomous run, which drives one task to a predicate; work one agent could finish inside the session's budget routes there, not here, however program-shaped the phrasing sounds. `playbooks/orchestrate.md`.
 - **Autopilot-full.** A queue of independent PRs run to merged with full autonomy: one owner per PR carries build through merge, and the root swarm-verifies each merge-ready head before its owner merges ("autopilot this queue", "full autopilot", one-owner-per-PR programs). `playbooks/autopilot-full.md`.
diff --git a/skills/poteto-mode/playbooks/autopilot-stack.md b/skills/poteto-mode/playbooks/autopilot-stack.md
index c1ef8af..7a95f90 100644
--- a/skills/poteto-mode/playbooks/autopilot-stack.md
+++ b/skills/poteto-mode/playbooks/autopilot-stack.md
@@ -2,7 +2,7 @@
 
 **You own the stack, never the landing. Build and verify the queue with full
 autonomy, then hand the operator one linear reviewed stack.** `gh stack` is
-the default named backend; Graphite is optional. For "autopilot-stack", "stack
+the delivery backend. For "autopilot-stack", "stack
 them, don't ship", "build the stack, I'll land it". The sibling of
 **Autopilot-full**. The owner loop and verification gate are the same; only
 the terminal differs. Nothing auto-ships from this skill.
diff --git a/skills/poteto-mode/playbooks/opening-a-pr.md b/skills/poteto-mode/playbooks/opening-a-pr.md
index ebfd4b5..340b322 100644
--- a/skills/poteto-mode/playbooks/opening-a-pr.md
+++ b/skills/poteto-mode/playbooks/opening-a-pr.md
@@ -20,7 +20,7 @@ Invoked at the end of every other playbook.
 
 After these sections, attach videos or screenshots when they prove a claim. Do not use `## Summary` or `## Test plan` boilerplate. A commit body does not restate its subject.
 
-**Size and stacks.** Prefer narrow PRs to one large PR. Use the configured delivery adapter, with `gh stack` as the default and Graphite as an explicit option. Keep the ordered stack visible to reviewers. Branch from main only for independent work. Rebase on `main` before substantial stack work.
+**Size and stacks.** Prefer narrow PRs to one large PR. Use the `gh stack` delivery adapter. Keep the ordered stack visible to reviewers. Branch from main only for independent work. Rebase on `main` before substantial stack work.
 
 **Readiness.** Open every PR ready, never as a draft. Set `draft: false` on every PR creation call. If a PR still opens as a draft, run the host's ready command, such as `gh pr ready <number>`. Run `gh pr view <number>` before you refer to PR status.
 
diff --git a/skills/poteto-mode/playbooks/shipping.md b/skills/poteto-mode/playbooks/shipping.md
index 4cc2fe1..11ef875 100644
--- a/skills/poteto-mode/playbooks/shipping.md
+++ b/skills/poteto-mode/playbooks/shipping.md
@@ -4,17 +4,15 @@
 
 This is the half after `playbooks/babysit.md`. Babysit makes a stack mergeable.
 Shipping decides what is actually safe to merge and hands a verified receipt to
-the selected delivery adapter. `gh stack` is the default backend; Graphite is
-an optional named backend. These resources do not implement backend state.
+the selected delivery adapter. `gh stack` is the delivery backend. These resources do not implement backend state.
 Green is not safe, and the gap between those two words is where this playbook
 lives.
 
 1. **Verify every PR independently before arming anything.** One subagent per PR, not batched, each a managed-worktree subagent, each exercising the real surface through the available browser, CLI, desktop, or mobile verification skill against parent versus head. Each returns `PASS`, `PASS+NOTES` or `FAIL` and posts that verdict on its own PR so the record outlives the chat. Safe means a verdict from an agent that did not write the code. CI green is not a verdict, and an approving bot review is not a verdict.
 2. **Land only the contiguous verified run rooted at the bottom.** Walk up from the lowest unmerged PR and stop at the first one without a passing verdict, where both `PASS` and `PASS+NOTES` pass. A verified PR sitting above an unverified one is not landable, because merging it would pull the gap in underneath it. Report the ceiling as a PR number and say what breaks the chain.
 3. **Re-check that the verdicts still describe the code.** A restack rewrites every SHA above it and silently invalidates every verdict without touching a single check. Compare `git patch-id` at the verdict SHA against the current head before trusting an older verdict, and re-verify anything that actually drifted. Twenty-one verdicts went stale this way in one run with no signal at all.
-4. **Hand the receipt to the delivery adapter.** Use the default `gh stack`
-   adapter unless the operator explicitly selected the named Graphite adapter.
-   Do not invent or mutate stack state in a skill; the later
+4. **Hand the receipt to the delivery adapter.** Use the `gh stack`
+   adapter. Do not invent or mutate stack state in a skill; the later
    `src/delivery` adapter owns submission, merge-when-ready, and exact-head
    checks.
 5. **Never enable GitHub auto-merge on a stack directly.** Only the selected
diff --git a/src/delivery/adapters.ts b/src/delivery/adapters.ts
index 926aa65..ca118c1 100644
--- a/src/delivery/adapters.ts
+++ b/src/delivery/adapters.ts
@@ -1,9 +1,13 @@
 import type {
   CommandResult,
   CommandRunner,
+  GitSha,
+  ProvenMembers,
+  PullRequestNumber,
   StackActionResult,
   StackBackend,
   StackBackendName,
+  StackMember,
   StackOperation,
   StackSnapshot,
 } from "./types.js";
@@ -12,6 +16,17 @@ export interface AdapterOptions {
   runner: CommandRunner;
 }
 
+const PULL_REQUEST_PATTERN = /^[1-9][0-9]*$/;
+const GIT_SHA_PATTERN = /^[a-f0-9]{6,64}$/i;
+
+function gitSha(value: string): GitSha | undefined {
+  return GIT_SHA_PATTERN.test(value) ? (value as GitSha) : undefined;
+}
+
+function pullRequestNumber(value: string): PullRequestNumber | undefined {
+  return PULL_REQUEST_PATTERN.test(value) ? (value as PullRequestNumber) : undefined;
+}
+
 function parseOutput(result: CommandResult): unknown {
   const output = result.stdout.trim();
   if (!output) return {};
@@ -31,25 +46,55 @@ function field(value: unknown, ...names: string[]): string | undefined {
   return undefined;
 }
 
-function stackFields(value: unknown): Pick<StackSnapshot, "headSha" | "pullRequest"> {
-  if (
-    !value ||
-    typeof value !== "object" ||
-    !Array.isArray((value as { branches?: unknown }).branches)
-  )
-    return {};
-  const snapshot = value as {
-    currentBranch?: unknown;
-    branches: { name?: unknown; head?: unknown; pr?: { url?: unknown } }[];
-  };
-  const current = snapshot.branches.find(
-    (branch) =>
-      typeof snapshot.currentBranch === "string" && branch.name === snapshot.currentBranch,
-  );
-  return {
-    headSha: typeof current?.head === "string" ? current.head : undefined,
-    pullRequest: typeof current?.pr?.url === "string" ? current.pr.url : undefined,
-  };
+function parseGhStackMembership(raw: unknown): ProvenMembers | undefined {
+  if (!raw || typeof raw !== "object" || !Array.isArray((raw as { branches?: unknown }).branches))
+    return undefined;
+  const entries = (raw as { branches: unknown[] }).branches.flatMap((branch) => {
+    if (!branch || typeof branch !== "object") return [];
+    const item = branch as {
+      head?: unknown;
+      base?: unknown;
+      isMerged?: unknown;
+      pr?: { number?: unknown; state?: unknown };
+    };
+    if (item.isMerged === true || item.pr?.state === "MERGED") return [];
+    const head = typeof item.head === "string" ? gitSha(item.head) : undefined;
+    const base = typeof item.base === "string" ? gitSha(item.base) : undefined;
+    const number = item.pr?.number;
+    if (typeof number !== "number" || !Number.isInteger(number) || number <= 0) return [];
+    const pullRequest = pullRequestNumber(String(number));
+    if (!head || !base || !pullRequest) return [];
+    return [{ pullRequest, headSha: head, baseSha: base }];
+  });
+  if (entries.length === 0) return undefined;
+  const heads = new Map(entries.map((entry) => [entry.headSha, entry]));
+  if (heads.size !== entries.length) return undefined;
+  const roots = entries.filter((entry) => !heads.has(entry.baseSha));
+  if (roots.length !== 1) return undefined;
+  const ordered = [] as StackMember[];
+  const visited = new Set<string>();
+  let current: StackMember | undefined = roots[0];
+  while (current) {
+    if (visited.has(current.headSha)) return undefined;
+    visited.add(current.headSha);
+    ordered.push(current);
+    const children = entries.filter((entry) => entry.baseSha === current!.headSha);
+    if (children.length > 1) return undefined;
+    current = children[0];
+  }
+  if (ordered.length !== entries.length) return undefined;
+  if (ordered.length === 0) return undefined;
+  return ordered as unknown as ProvenMembers;
+}
+
+export function membersThrough(
+  members: ProvenMembers,
+  targetPullRequest: string,
+): ProvenMembers | undefined {
+  if (!PULL_REQUEST_PATTERN.test(targetPullRequest)) return undefined;
+  const target = members.findIndex((member) => member.pullRequest === targetPullRequest);
+  if (target < 0) return undefined;
+  return members.slice(0, target + 1) as unknown as ProvenMembers;
 }
 
 function asCommandFailure(error: unknown): CommandResult {
@@ -75,9 +120,8 @@ function branchOperand(value: string): string {
   return value;
 }
 
-function pullRequestOperand(value: string | undefined): string[] {
-  if (value === undefined) return [];
-  if (!/^[1-9][0-9]*$/.test(value)) throw new Error(`unsafe pull request operand: ${value}`);
+function pullRequestOperand(value: string): string[] {
+  if (!PULL_REQUEST_PATTERN.test(value)) throw new Error(`unsafe pull request operand: ${value}`);
   return [value];
 }
 
@@ -96,17 +140,11 @@ abstract class CliStackBackend implements StackBackend {
   public constructor(protected readonly runner: CommandRunner) {}
 
   public async inspect(): Promise<StackSnapshot> {
-    const argv = this.argv({ kind: "inspect" });
-    const result = await run(this.runner, argv);
-    const raw = parseOutput(result);
-    const stack = this.name === "gh-stack" ? stackFields(raw) : {};
-    return {
-      backend: this.name,
-      repoIdentity: field(raw, "repoIdentity", "repository", "repo", "name"),
-      headSha: stack.headSha ?? field(raw, "headSha", "head_sha", "sha", "oid"),
-      pullRequest: stack.pullRequest ?? field(raw, "pullRequest", "pull_request", "pr", "url"),
-      raw,
-    };
+    const result = await run(this.runner, this.argv({ kind: "inspect" }));
+    if (result.exitCode !== 0) return { kind: "unproven", backend: this.name };
+    const members = parseGhStackMembership(parseOutput(result));
+    if (!members) return { kind: "unproven", backend: this.name };
+    return { kind: "proven", backend: this.name, members };
   }
 
   public async execute(operation: StackOperation): Promise<StackActionResult> {
@@ -153,80 +191,17 @@ export class GhStackBackend extends CliStackBackend {
 
 export const GhStackAdapter = GhStackBackend;
 
-/** Adapter for Graphite's official `gt` CLI. It stores no stack state. */
-export class GraphiteBackend extends CliStackBackend {
-  public readonly name = "graphite" as const;
-
-  protected argv(operation: StackOperation): readonly string[] {
-    switch (operation.kind) {
-      case "inspect":
-        return ["gt", "log", "short", "--stack", "--reverse"];
-      case "prepare":
-        return ["gt", "create", branchOperand(operation.branch), "--no-interactive"];
-      case "submit":
-        return [
-          "gt",
-          "submit",
-          ...(operation.draft ? ["--draft"] : []),
-          ...(operation.draft ? [] : ["--publish"]),
-          "--no-edit",
-          "--no-interactive",
-        ];
-      case "sync":
-        return ["gt", "sync", "--no-interactive"];
-      case "rebase":
-        return ["gt", "restack", "--no-interactive"];
-      case "auto-merge":
-        return [
-          "gt",
-          "submit",
-          "--merge-when-ready",
-          "--always",
-          "--update-only",
-          "--no-edit",
-          "--no-interactive",
-        ];
-    }
-  }
-}
-
-export const GraphiteAdapter = GraphiteBackend;
-
-export function isGraphiteAvailable(
-  availableCommands: readonly (string | { name: string })[],
-): boolean {
-  return availableCommands.some(
-    (command) => (typeof command === "string" ? command : command.name) === "gt",
-  );
-}
-
 export interface DeliveryBackends {
   ghStack: GhStackBackend;
-  graphite?: GraphiteBackend;
 }
 
-/** gh is the default; Graphite is exposed only after a capability preflight. */
 export function createDeliveryBackends(options: {
   runner: CommandRunner;
   availableCommands?: readonly (string | { name: string })[];
 }): DeliveryBackends {
-  const backends: DeliveryBackends = { ghStack: new GhStackBackend(options.runner) };
-  if (options.availableCommands && isGraphiteAvailable(options.availableCommands)) {
-    backends.graphite = new GraphiteBackend(options.runner);
-  }
-  return backends;
+  return { ghStack: new GhStackBackend(options.runner) };
 }
 
 export function createGhStackBackend(options: AdapterOptions): GhStackBackend {
   return new GhStackBackend(options.runner);
 }
-
-/** Returns no adapter when the optional `gt` capability is not present. */
-export function createGraphiteBackend(options: {
-  runner: CommandRunner;
-  availableCommands: readonly (string | { name: string })[];
-}): GraphiteBackend | undefined {
-  return isGraphiteAvailable(options.availableCommands)
-    ? new GraphiteBackend(options.runner)
-    : undefined;
-}
diff --git a/src/delivery/authorization.ts b/src/delivery/authorization.ts
index 05d3a76..abd9e0f 100644
--- a/src/delivery/authorization.ts
+++ b/src/delivery/authorization.ts
@@ -38,8 +38,7 @@ export function validateEvidenceReceipt(receipt: unknown): string[] {
   }
   if (!validFileEvidence(value.featureMap)) reasons.push("invalid feature map evidence");
   if (!validFileEvidence(value.skill)) reasons.push("invalid skill evidence");
-  if (value.backend !== "gh-stack" && value.backend !== "graphite")
-    reasons.push("missing evidence: backend");
+  if (value.backend !== "gh-stack") reasons.push("missing evidence: backend");
 
   const checks = value.deterministicChecks;
   if (
diff --git a/src/delivery/types.ts b/src/delivery/types.ts
index c64d980..5db0e90 100644
--- a/src/delivery/types.ts
+++ b/src/delivery/types.ts
@@ -2,10 +2,34 @@
 export const AUTONOMY_LEVELS = ["verify", "prepare", "pr", "merge-ready", "auto-merge"] as const;
 
 export type AutonomyLevel = (typeof AUTONOMY_LEVELS)[number];
-export type StackBackendName = "gh-stack" | "graphite";
+export type StackBackendName = "gh-stack";
 export type DeliveryOrigin = "human" | "benny";
 export type ProjectReadiness = "ready" | "not-ready";
 
+export type GitSha = string & { readonly __brand: "GitSha" };
+export type PullRequestNumber = string & { readonly __brand: "PullRequestNumber" };
+
+export interface StackMember {
+  readonly pullRequest: PullRequestNumber;
+  readonly headSha: GitSha;
+  readonly baseSha: GitSha;
+}
+
+export type ProvenMembers = readonly [StackMember, ...StackMember[]] & {
+  readonly __brand: "ProvenMembers";
+};
+
+export type StackSnapshot =
+  | {
+      readonly kind: "proven";
+      readonly backend: StackBackendName;
+      readonly members: ProvenMembers;
+    }
+  | {
+      readonly kind: "unproven";
+      readonly backend: StackBackendName;
+    };
+
 export interface CommandResult {
   exitCode: number;
   stdout: string;
@@ -135,15 +159,7 @@ export type StackOperation =
   | { kind: "submit"; draft: boolean }
   | { kind: "sync" }
   | { kind: "rebase" }
-  | { kind: "auto-merge"; pullRequest?: string };
-
-export interface StackSnapshot {
-  backend: StackBackendName;
-  repoIdentity?: string;
-  headSha?: string;
-  pullRequest?: string;
-  raw: unknown;
-}
+  | { kind: "auto-merge"; pullRequest: string };
 
 export interface StackActionResult {
   backend: StackBackendName;
diff --git a/test/delivery/delivery.test.mjs b/test/delivery/delivery.test.mjs
index bdeec21..92c2e9d 100644
--- a/test/delivery/delivery.test.mjs
+++ b/test/delivery/delivery.test.mjs
@@ -5,7 +5,27 @@ import { createJiti } from "jiti";
 const jiti = createJiti(import.meta.url, { interopDefault: true });
 const delivery = await jiti.import("../../src/delivery/index.ts");
 
-function runnerFor(backend) {
+const twoPrStack = {
+  trunk: "main",
+  branches: [
+    {
+      name: "base-feature",
+      head: "111111",
+      base: "000000",
+      isMerged: false,
+      pr: { number: 41, state: "OPEN" },
+    },
+    {
+      name: "feature",
+      head: "abc123",
+      base: "111111",
+      isMerged: false,
+      pr: { number: 42, state: "OPEN" },
+    },
+  ],
+};
+
+function runnerForStack(payload, exitCode = 0) {
   const calls = [];
   return {
     calls,
@@ -13,29 +33,8 @@ function runnerFor(backend) {
       calls.push([...argv]);
       if (argv[0] === "gh" && argv.includes("view")) {
         return {
-          exitCode: 0,
-          stdout: JSON.stringify({
-            trunk: "main",
-            currentBranch: "feature",
-            branches: [
-              {
-                name: "feature",
-                head: "abc123",
-                pr: { number: 42, url: "https://github.com/acme/demo/pull/42" },
-              },
-            ],
-          }),
-          stderr: "",
-        };
-      }
-      if (argv.includes("log")) {
-        return {
-          exitCode: 0,
-          stdout: JSON.stringify({
-            repoIdentity: "acme/demo",
-            headSha: "abc123",
-            pullRequest: "42",
-          }),
+          exitCode,
+          stdout: typeof payload === "string" ? payload : JSON.stringify(payload),
           stderr: "",
         };
       }
@@ -45,35 +44,105 @@ function runnerFor(backend) {
         stderr: "",
       };
     },
-    backend,
   };
 }
 
-for (const [label, Backend, expectedCommand] of [
-  ["gh stack", delivery.GhStackBackend, "gh"],
-  ["Graphite", delivery.GraphiteBackend, "gt"],
-]) {
-  test(`${label} adapter translates argv and parses fixture output`, async () => {
-    const runner = runnerFor(label);
-    const backend = new Backend(runner);
+test("gh stack inspect returns proven parent-first members for a two-PR chain", async () => {
+  const runner = runnerForStack(twoPrStack);
+  const backend = new delivery.GhStackBackend(runner);
+  const snapshot = await backend.inspect();
+  assert.equal(snapshot.kind, "proven");
+  assert.equal(snapshot.backend, "gh-stack");
+  assert.deepEqual(
+    snapshot.members.map((member) => ({
+      pullRequest: member.pullRequest,
+      headSha: member.headSha,
+      baseSha: member.baseSha,
+    })),
+    [
+      { pullRequest: "41", headSha: "111111", baseSha: "000000" },
+      { pullRequest: "42", headSha: "abc123", baseSha: "111111" },
+    ],
+  );
+  const prefix = delivery.membersThrough(snapshot.members, "42");
+  assert.deepEqual(prefix, snapshot.members);
+});
+
+test("gh stack inspect is unproven for invalid membership topologies", async () => {
+  const cases = [
+    [
+      "fork",
+      {
+        branches: [
+          { head: "aaa", base: "000", pr: { number: 1, state: "OPEN" } },
+          { head: "bbb", base: "000", pr: { number: 2, state: "OPEN" } },
+        ],
+      },
+    ],
+    [
+      "cycle",
+      {
+        branches: [
+          { head: "aaa", base: "bbb", pr: { number: 1, state: "OPEN" } },
+          { head: "bbb", base: "aaa", pr: { number: 2, state: "OPEN" } },
+        ],
+      },
+    ],
+    [
+      "duplicate head",
+      {
+        branches: [
+          { head: "aaa", base: "000", pr: { number: 1, state: "OPEN" } },
+          { head: "aaa", base: "111", pr: { number: 2, state: "OPEN" } },
+        ],
+      },
+    ],
+    [
+      "missing base",
+      {
+        branches: [{ head: "aaa", pr: { number: 1, state: "OPEN" } }],
+      },
+    ],
+    ["exit code 1 with valid JSON", twoPrStack, 1],
+    ["empty stdout", "", 0],
+    ["malformed stdout", "not-json", 0],
+    [
+      "all merged",
+      {
+        branches: [
+          {
+            head: "aaa",
+            base: "000",
+            isMerged: true,
+            pr: { number: 1, state: "MERGED" },
+          },
+        ],
+      },
+    ],
+  ];
+  for (const [label, payload, exitCode = 0] of cases) {
+    const backend = new delivery.GhStackBackend(runnerForStack(payload, exitCode));
     const snapshot = await backend.inspect();
-    assert.equal(snapshot.backend, label === "gh stack" ? "gh-stack" : "graphite");
-    if (label === "gh stack") {
-      assert.equal(snapshot.headSha, "abc123");
-      assert.equal(snapshot.pullRequest, "https://github.com/acme/demo/pull/42");
-    }
-    const result = await backend.execute({ kind: "submit", draft: true });
-    assert.equal(result.accepted, true);
-    assert.equal(result.pullRequest, "42");
-    assert.equal(runner.calls[0][0], expectedCommand);
-    assert.ok(runner.calls.every((argv) => Array.isArray(argv)));
-  });
-}
+    assert.equal(snapshot.kind, "unproven", label);
+    assert.equal(snapshot.backend, "gh-stack");
+  }
+});
+
+test("membersThrough returns undefined for missing and malformed targets without throwing", () => {
+  const members = [
+    { pullRequest: "41", headSha: "111111", baseSha: "000000" },
+    { pullRequest: "42", headSha: "abc123", baseSha: "111111" },
+  ];
+  const proven = delivery.membersThrough(members, "99");
+  assert.equal(proven, undefined);
+  assert.doesNotThrow(() => delivery.membersThrough(members, "--admin"));
+  assert.equal(delivery.membersThrough(members, "--admin"), undefined);
+});
 
 test("gh stack adapter emits only documented non-interactive commands", async () => {
-  const runner = runnerFor("gh stack");
+  const runner = runnerForStack(twoPrStack);
   const backend = new delivery.GhStackBackend(runner);
-  await backend.inspect();
+  await backend.execute({ kind: "inspect" });
   await backend.execute({ kind: "prepare", branch: "feature" });
   await backend.execute({ kind: "submit", draft: false });
   await backend.execute({ kind: "sync" });
@@ -89,45 +158,15 @@ test("gh stack adapter emits only documented non-interactive commands", async ()
   ]);
 });
 
-test("Graphite adapter emits documented non-interactive commands", async () => {
-  const runner = runnerFor("Graphite");
-  const backend = new delivery.GraphiteBackend(runner);
-  await backend.inspect();
-  await backend.execute({ kind: "prepare", branch: "feature" });
-  await backend.execute({ kind: "submit", draft: true });
-  await backend.execute({ kind: "sync" });
-  await backend.execute({ kind: "rebase" });
-  await backend.execute({ kind: "auto-merge" });
-  assert.deepEqual(runner.calls, [
-    ["gt", "log", "short", "--stack", "--reverse"],
-    ["gt", "create", "feature", "--no-interactive"],
-    ["gt", "submit", "--draft", "--no-edit", "--no-interactive"],
-    ["gt", "sync", "--no-interactive"],
-    ["gt", "restack", "--no-interactive"],
-    [
-      "gt",
-      "submit",
-      "--merge-when-ready",
-      "--always",
-      "--update-only",
-      "--no-edit",
-      "--no-interactive",
-    ],
-  ]);
-});
-
-test("Graphite is capability-gated while gh stack remains the default", () => {
-  const runner = runnerFor("both");
-  const withoutGt = delivery.createDeliveryBackends({ runner, availableCommands: ["gh"] });
-  assert.ok(withoutGt.ghStack);
-  assert.equal(withoutGt.graphite, undefined);
-  assert.equal(delivery.isGraphiteAvailable(["gh", { name: "other" }]), false);
-  const withGt = delivery.createDeliveryBackends({ runner, availableCommands: ["gh", "gt"] });
-  assert.ok(withGt.graphite);
+test("createDeliveryBackends exposes gh stack only", () => {
+  const runner = runnerForStack(twoPrStack);
+  const backends = delivery.createDeliveryBackends({ runner, availableCommands: ["gh", "gt"] });
+  assert.ok(backends.ghStack);
+  assert.equal(backends.graphite, undefined);
 });
 
 test("stack adapters reject flag-shaped branch and pull request operands", async () => {
-  const backend = new delivery.GhStackBackend(runnerFor("gh stack"));
+  const backend = new delivery.GhStackBackend(runnerForStack(twoPrStack));
   await assert.rejects(
     backend.execute({ kind: "prepare", branch: "--help" }),
     /unsafe branch operand/,
@@ -177,7 +216,7 @@ function receipt(overrides = {}) {
   });
 }
 
-test("complete receipt authorizes only the selected backend action", () => {
+test("complete receipt authorizes gh-stack auto-merge", () => {
   const allowed = delivery.authorizeDelivery({
     receipt: receipt(),
     currentHeadSha: "abc123",
@@ -188,15 +227,11 @@ test("complete receipt authorizes only the selected backend action", () => {
   assert.equal(allowed.allowed, true);
   assert.equal(allowed.backend, "gh-stack");
   assert.equal(allowed.draftOnly, false);
-  const wrongBackend = delivery.authorizeDelivery({
-    receipt: receipt(),
-    currentHeadSha: "abc123",
-    backend: "graphite",
-    level: "auto-merge",
-    projectReadiness: "ready",
-  });
-  assert.equal(wrongBackend.allowed, false);
-  assert.match(wrongBackend.reasons.join(" "), /backend/i);
+});
+
+test("validateEvidenceReceipt rejects non-gh-stack backend values", () => {
+  const reasons = delivery.validateEvidenceReceipt(receipt({ backend: "graphite" }));
+  assert.match(reasons.join(" "), /backend/i);
 });
 
 test("negative controls fail closed for every delivery gate", () => {
diff --git a/test/resources/resources.test.mjs b/test/resources/resources.test.mjs
index cbadc26..3098ba6 100644
--- a/test/resources/resources.test.mjs
+++ b/test/resources/resources.test.mjs
@@ -148,7 +148,6 @@ test("active resources use Pi runtime contracts", async () => {
   assert.match(noComments, /agent: "comment-sicko"/);
   assert.match(recall, /PI_SESSION_FILE/);
   assert.match(shipping, /gh stack/);
-  assert.match(shipping, /Graphite/);
 });
 
 test("agent frontmatter uses valid Pi roles", async () => {

From 650922c68c24f29f5ac9db798d1f52a662b3ba58 Mon Sep 17 00:00:00 2001
From: Myke Matos <123015330+shishiv@users.noreply.github.com>
Date: Wed, 26 Aug 2026 13:11:40 -0300
Subject: [PATCH 2/3] fix(pstack): prove inspect topology with six-character
 SHAs

Short fixture SHAs never reached the fork and cycle walks. Drop the leftover availableCommands factory argument.
---
 extensions/poteto-mode.ts       |  6 +++---
 src/delivery/adapters.ts        |  5 +----
 test/delivery/delivery.test.mjs | 27 ++++++++++++++++-----------
 3 files changed, 20 insertions(+), 18 deletions(-)

diff --git a/extensions/poteto-mode.ts b/extensions/poteto-mode.ts
index b7fb320..f7b07db 100644
--- a/extensions/poteto-mode.ts
+++ b/extensions/poteto-mode.ts
@@ -366,7 +366,7 @@ export default function potetoModeExtension(pi: ExtensionAPI): void {
                 root: ctx.cwd,
                 repoIdentity: repository.stdout.trim(),
                 headSha: entry.headSha,
-                backend: params.backend as StackBackendName,
+                backend: params.backend,
                 level,
               });
               if (rejection) return deliveryRejected(rejection);
@@ -389,7 +389,7 @@ export default function potetoModeExtension(pi: ExtensionAPI): void {
               root: ctx.cwd,
               repoIdentity: repository.stdout.trim(),
               headSha: currentHead.stdout.trim(),
-              backend: params.backend as StackBackendName,
+              backend: params.backend,
               level,
             });
             if (rejection) return deliveryRejected(rejection);
@@ -397,7 +397,7 @@ export default function potetoModeExtension(pi: ExtensionAPI): void {
               receipt,
               repoIdentity: repository.stdout.trim(),
               currentHeadSha: currentHead.stdout.trim(),
-              backend: params.backend as StackBackendName,
+              backend: params.backend,
               level,
             });
             if (authorization.draftOnly && params.operation === "submit" && params.draft === false)
diff --git a/src/delivery/adapters.ts b/src/delivery/adapters.ts
index ca118c1..70886c8 100644
--- a/src/delivery/adapters.ts
+++ b/src/delivery/adapters.ts
@@ -195,10 +195,7 @@ export interface DeliveryBackends {
   ghStack: GhStackBackend;
 }
 
-export function createDeliveryBackends(options: {
-  runner: CommandRunner;
-  availableCommands?: readonly (string | { name: string })[];
-}): DeliveryBackends {
+export function createDeliveryBackends(options: { runner: CommandRunner }): DeliveryBackends {
   return { ghStack: new GhStackBackend(options.runner) };
 }
 
diff --git a/test/delivery/delivery.test.mjs b/test/delivery/delivery.test.mjs
index 92c2e9d..aa6b8b7 100644
--- a/test/delivery/delivery.test.mjs
+++ b/test/delivery/delivery.test.mjs
@@ -66,6 +66,11 @@ test("gh stack inspect returns proven parent-first members for a two-PR chain",
   );
   const prefix = delivery.membersThrough(snapshot.members, "42");
   assert.deepEqual(prefix, snapshot.members);
+  assert.deepEqual(
+    delivery.membersThrough(snapshot.members, "41")?.map((member) => member.pullRequest),
+    ["41"],
+  );
+  assert.equal(delivery.membersThrough(snapshot.members, "99"), undefined);
 });
 
 test("gh stack inspect is unproven for invalid membership topologies", async () => {
@@ -74,8 +79,8 @@ test("gh stack inspect is unproven for invalid membership topologies", async ()
       "fork",
       {
         branches: [
-          { head: "aaa", base: "000", pr: { number: 1, state: "OPEN" } },
-          { head: "bbb", base: "000", pr: { number: 2, state: "OPEN" } },
+          { head: "aaa111", base: "000000", pr: { number: 1, state: "OPEN" } },
+          { head: "bbb222", base: "000000", pr: { number: 2, state: "OPEN" } },
         ],
       },
     ],
@@ -83,8 +88,8 @@ test("gh stack inspect is unproven for invalid membership topologies", async ()
       "cycle",
       {
         branches: [
-          { head: "aaa", base: "bbb", pr: { number: 1, state: "OPEN" } },
-          { head: "bbb", base: "aaa", pr: { number: 2, state: "OPEN" } },
+          { head: "aaa111", base: "bbb222", pr: { number: 1, state: "OPEN" } },
+          { head: "bbb222", base: "aaa111", pr: { number: 2, state: "OPEN" } },
         ],
       },
     ],
@@ -92,15 +97,15 @@ test("gh stack inspect is unproven for invalid membership topologies", async ()
       "duplicate head",
       {
         branches: [
-          { head: "aaa", base: "000", pr: { number: 1, state: "OPEN" } },
-          { head: "aaa", base: "111", pr: { number: 2, state: "OPEN" } },
+          { head: "aaa111", base: "000000", pr: { number: 1, state: "OPEN" } },
+          { head: "aaa111", base: "111111", pr: { number: 2, state: "OPEN" } },
         ],
       },
     ],
     [
       "missing base",
       {
-        branches: [{ head: "aaa", pr: { number: 1, state: "OPEN" } }],
+        branches: [{ head: "aaa111", pr: { number: 1, state: "OPEN" } }],
       },
     ],
     ["exit code 1 with valid JSON", twoPrStack, 1],
@@ -111,8 +116,8 @@ test("gh stack inspect is unproven for invalid membership topologies", async ()
       {
         branches: [
           {
-            head: "aaa",
-            base: "000",
+            head: "aaa111",
+            base: "000000",
             isMerged: true,
             pr: { number: 1, state: "MERGED" },
           },
@@ -160,9 +165,9 @@ test("gh stack adapter emits only documented non-interactive commands", async ()
 
 test("createDeliveryBackends exposes gh stack only", () => {
   const runner = runnerForStack(twoPrStack);
-  const backends = delivery.createDeliveryBackends({ runner, availableCommands: ["gh", "gt"] });
+  const backends = delivery.createDeliveryBackends({ runner });
   assert.ok(backends.ghStack);
-  assert.equal(backends.graphite, undefined);
+  assert.equal("graphite" in backends, false);
 });
 
 test("stack adapters reject flag-shaped branch and pull request operands", async () => {

From c072556e9315eda3caf180f208004e358ccd831c Mon Sep 17 00:00:00 2001
From: Myke Matos <123015330+shishiv@users.noreply.github.com>
Date: Wed, 26 Aug 2026 13:13:23 -0300
Subject: [PATCH 3/3] chore(pstack): drop leftover comments in touched delivery
 files

The phase 8 diff added no comments. These lines were narration on restore, preflight, a private helper, Benny speculation, and a types banner.
---
 extensions/poteto-mode.ts     | 5 -----
 src/delivery/authorization.ts | 2 --
 src/delivery/types.ts         | 1 -
 3 files changed, 8 deletions(-)

diff --git a/extensions/poteto-mode.ts b/extensions/poteto-mode.ts
index f7b07db..d023576 100644
--- a/extensions/poteto-mode.ts
+++ b/extensions/poteto-mode.ts
@@ -51,8 +51,6 @@ export default function potetoModeExtension(pi: ExtensionAPI): void {
   }
 
   function restore(ctx: ExtensionContext): void {
-    // This closure is intentionally reset from the active branch on every
-    // session lifecycle event. It prevents state leaking across sessions.
     active = branchState(ctx);
   }
 
@@ -68,8 +66,6 @@ export default function potetoModeExtension(pi: ExtensionAPI): void {
 
   pi.on("session_start", async (_event, ctx) => {
     restore(ctx);
-    // Probe all host capability surfaces during startup without changing
-    // settings or failing print/JSON sessions. A task invocation fails closed.
     preflight(ctx);
   });
 
@@ -569,7 +565,6 @@ async function verifyPullRequestState(
   return undefined;
 }
 
-/** Flatten shell-like and structured tool payloads before applying merge gates. */
 function normalizeToolCallInput(input: unknown): string {
   const values: string[] = [];
   const seen = new WeakSet<object>();
diff --git a/src/delivery/authorization.ts b/src/delivery/authorization.ts
index abd9e0f..6c32797 100644
--- a/src/delivery/authorization.ts
+++ b/src/delivery/authorization.ts
@@ -129,8 +129,6 @@ export function authorizeDelivery(request: DeliveryAuthorizationRequest): Delive
     reasons.push("insufficient project readiness");
 
   const origin = request.origin ?? receipt?.origin;
-  // The shipped receipt tool emits human origin only. The Benny branch remains
-  // defense-in-depth for any future coordinator-issued receipt mechanism.
   const draftOnly = origin === "benny";
   if (origin === "benny" && (level === "merge-ready" || level === "auto-merge"))
     reasons.push("Benny origin is draft-only");
diff --git a/src/delivery/types.ts b/src/delivery/types.ts
index 5db0e90..3bf4f65 100644
--- a/src/delivery/types.ts
+++ b/src/delivery/types.ts
@@ -1,4 +1,3 @@
-/** The only delivery state kept by this package is evidence for one head SHA. */
 export const AUTONOMY_LEVELS = ["verify", "prepare", "pr", "merge-ready", "auto-merge"] as const;
 
 export type AutonomyLevel = (typeof AUTONOMY_LEVELS)[number];