perf: bound repair plan PR hydration

CHANGELOG.md

@@ -105,6 +105,8 @@ checkpoint, and status-only commits are intentionally omitted.
   tokens exist, keeping untargeted fix prompts cheaper to build.
 - Requested 100-item REST pages for paginated GitHub list calls, reducing
   review and repair API page fan-out on large issues and pull requests.
+- Bounded repair cluster PR file and commit hydration to the context carried
+  into generated plans, avoiding full pagination for very large pull requests.
 - Compacted review prompt context lazily so large comment, timeline, file, and
   commit lists no longer process entries that are omitted from Codex input.
 - Scoped every sweep workflow status write to the active target repository so

src/repair/github-cli.ts

@@ -48,12 +48,19 @@ export function ghJsonBestEffort<T = JsonValue>(
 }
 
 export function githubPaginatedPath(apiPath: string): string {
-  const [basePart, query = ""] = apiPath.split("?", 2);
-  const base = basePart ?? apiPath;
-  const params = new URLSearchParams(query);
-  if (!params.has("per_page")) params.set("per_page", "100");
-  const serialized = params.toString();
-  return serialized ? `${base}?${serialized}` : base;
+  return githubPathWithQueryDefaults(apiPath, { per_page: "100" });
+}
+
+export function githubLimitedPagePath(apiPath: string, limit: number, page = 1): string {
+  const normalizedLimit = Number.isFinite(limit) ? Math.floor(limit) : 1;
+  const normalizedPage = Number.isFinite(page) ? Math.floor(page) : 1;
+  const pageSize = Math.max(1, Math.min(100, normalizedLimit));
+  const pageNumber = Math.max(1, normalizedPage);
+  return githubPathWithQueryDefaults(
+    apiPath,
+    { per_page: String(pageSize), page: String(pageNumber) },
+    { override: true },
+  );
 }
 
 export function ghPaged<T = JsonValue>(apiPath: string, options: GhRunOptions = {}): T[] {
@@ -89,6 +96,28 @@ export async function ghPagedWithRetryAsync<T = JsonValue>(
   return pages.flatMap((page: JsonValue) => (Array.isArray(page) ? (page as T[]) : []));
 }
 
+export function ghPagedLimit<T = JsonValue>(
+  apiPath: string,
+  limit: number,
+  options: GhRunOptions = {},
+): T[] {
+  const max = Number.isFinite(limit) ? Math.max(0, Math.floor(limit)) : 0;
+  if (max <= 0) return [];
+
+  const perPage = Math.min(100, max);
+  const out: T[] = [];
+  for (let page = 1; out.length < max; page += 1) {
+    const entries = ghJson<JsonValue[]>(
+      ["api", githubLimitedPagePath(apiPath, perPage, page)],
+      options,
+    );
+    if (!Array.isArray(entries) || entries.length === 0) break;
+    out.push(...(entries as T[]));
+    if (entries.length < perPage) break;
+  }
+  return out.slice(0, max);
+}
+
 export function ghText(ghArgs: string[], options: GhRunOptions = {}): string {
   const text = execFileSync("gh", ghArgs, {
     cwd: options.cwd ?? repoRoot(),
@@ -234,6 +263,21 @@ function resolveRetryOptions(options: GhRetryOptions | number): GhRetryOptions {
   return options;
 }
 
+function githubPathWithQueryDefaults(
+  apiPath: string,
+  defaults: Record<string, string>,
+  { override = false }: { override?: boolean } = {},
+): string {
+  const [basePart, query = ""] = apiPath.split("?", 2);
+  const base = basePart ?? apiPath;
+  const params = new URLSearchParams(query);
+  for (const [key, value] of Object.entries(defaults)) {
+    if (override || !params.has(key)) params.set(key, value);
+  }
+  const serialized = params.toString();
+  return serialized ? `${base}?${serialized}` : base;
+}
+
 function bufferLikeToString(value: unknown): string {
   if (Buffer.isBuffer(value)) return value.toString("utf8");
   return String(value ?? "");

src/repair/lib.ts

@@ -377,9 +377,14 @@ function compactPlanItem(item: LooseRecord) {
           branch_writable: pull.branch_writable,
           branch_write_reason: pull.branch_write_reason,
           changed_files: pull.changed_files,
+          files_hydrated: pull.files_hydrated,
+          files_truncated: pull.files_truncated,
           additions: pull.additions,
           deletions: pull.deletions,
           files: (pull.files ?? []).slice(0, 40),
+          commits_count: pull.commits_count,
+          commits_hydrated: pull.commits_hydrated,
+          commits_truncated: pull.commits_truncated,
           commits: (pull.commits ?? []).slice(0, 10),
           reviews: (pull.reviews ?? []).slice(0, 12),
           review_comments_count: pull.review_comments_count,

src/repair/plan-cluster.ts

@@ -11,13 +11,15 @@ import {
   repoRoot,
   validateJob,
 } from "./lib.js";
-import { ghJson, ghPaged, ghText } from "./github-cli.js";
+import { ghJson, ghPaged, ghPagedLimit, ghText } from "./github-cli.js";
 import { hasSecurityRepairOptInLabel } from "./security-boundary.js";
 
 const MAX_LINKED_REFS = Number(process.env.CLAWSWEEPER_MAX_LINKED_REFS ?? 0);
 const HYDRATE_COMMENTS = process.env.CLAWSWEEPER_HYDRATE_COMMENTS === "1";
 const MAX_COMMENTS_PER_ITEM = Number(process.env.CLAWSWEEPER_MAX_COMMENTS_PER_ITEM ?? 30);
 const MAX_REVIEW_COMMENTS_PER_PR = Number(process.env.CLAWSWEEPER_MAX_REVIEW_COMMENTS_PER_PR ?? 50);
+const MAX_FILES_PER_PR = Number(process.env.CLAWSWEEPER_MAX_FILES_PER_PR ?? 80);
+const MAX_COMMITS_PER_PR = Number(process.env.CLAWSWEEPER_MAX_COMMITS_PER_PR ?? 80);
 const MAINTAINER_AUTHOR_ASSOCIATIONS = new Set(["OWNER", "MEMBER", "COLLABORATOR"]);
 const REVIEW_BOT_PATTERN =
   /\b(greptile|codex|asile|coderabbit|code rabbit|copilot|reviewdog|sonar|deepsource|codecov|github-actions)\b/i;
@@ -141,6 +143,8 @@ const plan = {
     hydrate_comments: HYDRATE_COMMENTS,
     max_comments_per_item: MAX_COMMENTS_PER_ITEM,
     max_review_comments_per_pr: MAX_REVIEW_COMMENTS_PER_PR,
+    max_files_per_pr: MAX_FILES_PER_PR,
+    max_commits_per_pr: MAX_COMMITS_PER_PR,
   },
   items: itemList.map((item: JsonValue) => summarizeItem(item, job)),
   canonical_candidates: canonicalCandidates(itemList, job),
@@ -184,10 +188,16 @@ function hydrateItem(repo: string, number: JsonValue) {
   }
   const comments = HYDRATE_COMMENTS ? ghPaged(`repos/${repo}/issues/${number}/comments`) : [];
   const pullRequest = issue.pull_request ? ghJson(["api", `repos/${repo}/pulls/${number}`]) : null;
-  const files = pullRequest ? ghPaged(`repos/${repo}/pulls/${number}/files`) : [];
-  const commits = pullRequest ? ghPaged(`repos/${repo}/pulls/${number}/commits`) : [];
+  const files = pullRequest
+    ? ghPagedLimit(`repos/${repo}/pulls/${number}/files`, MAX_FILES_PER_PR)
+    : [];
+  const commits = pullRequest
+    ? ghPagedLimit(`repos/${repo}/pulls/${number}/commits`, MAX_COMMITS_PER_PR)
+    : [];
   const reviews = pullRequest ? ghPaged(`repos/${repo}/pulls/${number}/reviews`) : [];
   const reviewComments = pullRequest ? ghPaged(`repos/${repo}/pulls/${number}/comments`) : [];
+  const changedFilesCount = countValue(pullRequest?.changed_files, files.length);
+  const commitsCount = countValue(pullRequest?.commits, commits.length);
   const checks = pullRequest ? ghPrChecks(repo, number) : [];
 
   return {
@@ -240,13 +250,18 @@ function hydrateItem(repo: string, number: JsonValue) {
             .filter(Boolean),
           additions: pullRequest.additions,
           deletions: pullRequest.deletions,
-          changed_files: pullRequest.changed_files,
+          changed_files: changedFilesCount,
+          files_hydrated: files.length,
+          files_truncated: Math.max(0, changedFilesCount - files.length),
           files: files.map((file: JsonValue) => ({
             filename: file.filename,
             status: file.status,
             additions: file.additions,
             deletions: file.deletions,
           })),
+          commits_count: commitsCount,
+          commits_hydrated: commits.length,
+          commits_truncated: Math.max(0, commitsCount - commits.length),
           commits: commits.map((commit: JsonValue) => ({
             sha: commit.sha,
             message: firstLine(commit.commit?.message),
@@ -302,6 +317,11 @@ function unavailableItem(repo: string, number: JsonValue, error: JsonValue) {
   };
 }
 
+function countValue(value: JsonValue, fallback = 0) {
+  const number = Number(value);
+  return Number.isFinite(number) && number >= 0 ? Math.floor(number) : fallback;
+}
+
 function summarizeItem(item: LooseRecord, job: LooseRecord) {
   return {
     repo: item.repo,
@@ -356,9 +376,14 @@ function summarizeItem(item: LooseRecord, job: LooseRecord) {
           requested_reviewers: item.pull_request.requested_reviewers,
           requested_teams: item.pull_request.requested_teams,
           changed_files: item.pull_request.changed_files,
+          files_hydrated: item.pull_request.files_hydrated,
+          files_truncated: item.pull_request.files_truncated,
           additions: item.pull_request.additions,
           deletions: item.pull_request.deletions,
           files: item.pull_request.files,
+          commits_count: item.pull_request.commits_count,
+          commits_hydrated: item.pull_request.commits_hydrated,
+          commits_truncated: item.pull_request.commits_truncated,
           commits: item.pull_request.commits,
           reviews: item.pull_request.reviews,
           review_comments_count: item.pull_request.review_comments.length,

test/repair/github-cli.test.ts

@@ -1,7 +1,7 @@
 import assert from "node:assert/strict";
 import test from "node:test";
 
-import { githubPaginatedPath } from "../../dist/repair/github-cli.js";
+import { githubLimitedPagePath, githubPaginatedPath } from "../../dist/repair/github-cli.js";
 
 test("githubPaginatedPath requests maximum REST page size by default", () => {
   assert.equal(
@@ -17,3 +17,22 @@ test("githubPaginatedPath requests maximum REST page size by default", () => {
     "repos/openclaw/openclaw/issues?per_page=50&state=open",
   );
 });
+
+test("githubLimitedPagePath caps one REST page and preserves existing filters", () => {
+  assert.equal(
+    githubLimitedPagePath("repos/openclaw/openclaw/pulls/123/files", 80),
+    "repos/openclaw/openclaw/pulls/123/files?per_page=80&page=1",
+  );
+  assert.equal(
+    githubLimitedPagePath(
+      "repos/openclaw/openclaw/pulls/123/files?state=open&per_page=100",
+      250,
+      3,
+    ),
+    "repos/openclaw/openclaw/pulls/123/files?state=open&per_page=100&page=3",
+  );
+  assert.equal(
+    githubLimitedPagePath("repos/openclaw/openclaw/pulls/123/files", 0, 0),
+    "repos/openclaw/openclaw/pulls/123/files?per_page=1&page=1",
+  );
+});

test/repair/plan-cluster.test.ts

@@ -222,9 +222,71 @@ test("plan-cluster treats same-repo PR branches as writable despite raw maintain
   assert.match(pull.branch_write_reason, /same-repo head branch/);
 });
 
+test("plan-cluster bounds PR file and commit hydration", () => {
+  const tmp = fs.mkdtempSync(path.join(os.tmpdir(), "clawsweeper-plan-bounded-pr-"));
+  const binDir = path.join(tmp, "bin");
+  const jobPath = path.join(tmp, "job.md");
+  const runDir = path.join(tmp, "run");
+  fs.mkdirSync(binDir);
+  fs.writeFileSync(path.join(binDir, "gh"), fakeGhScript(), { mode: 0o755 });
+
+  fs.writeFileSync(
+    jobPath,
+    [
+      "---",
+      "repo: openclaw/openclaw",
+      "cluster_id: automerge-openclaw-openclaw-74134",
+      "mode: autonomous",
+      "allowed_actions:",
+      "  - comment",
+      "  - fix",
+      "  - raise_pr",
+      "blocked_actions:",
+      "  - close",
+      "  - merge",
+      "source: pr_automerge",
+      "canonical:",
+      "  - #74134",
+      "candidates:",
+      "  - #74134",
+      "allow_fix_pr: true",
+      "allow_merge: false",
+      "security_policy: central_security_only",
+      "security_sensitive: false",
+      "---",
+      "Maintainer opted #74134 into ClawSweeper automerge.",
+      "",
+    ].join("\n"),
+  );
+
+  execFileSync(process.execPath, ["dist/repair/plan-cluster.js", jobPath, "--run-dir", runDir], {
+    cwd: process.cwd(),
+    env: {
+      ...process.env,
+      PATH: `${binDir}${path.delimiter}${process.env.PATH}`,
+      FAKE_GH_LARGE_PR: "1",
+    },
+    stdio: "pipe",
+  });
+
+  const clusterPlan = JSON.parse(fs.readFileSync(path.join(runDir, "cluster-plan.json"), "utf8"));
+  const pull = clusterPlan.items[0].pull_request;
+
+  assert.equal(pull.changed_files, 120);
+  assert.equal(pull.files_hydrated, 80);
+  assert.equal(pull.files_truncated, 40);
+  assert.equal(pull.files.length, 80);
+  assert.equal(pull.commits_count, 120);
+  assert.equal(pull.commits_hydrated, 80);
+  assert.equal(pull.commits_truncated, 40);
+  assert.equal(pull.commits.length, 80);
+});
+
 function fakeGhScript() {
   return `#!/usr/bin/env node
+const fs = require("node:fs");
 const args = process.argv.slice(2);
+if (process.env.FAKE_GH_LOG) fs.appendFileSync(process.env.FAKE_GH_LOG, args.join(" ") + "\\n");
 function write(value) {
   process.stdout.write(JSON.stringify(value));
 }
@@ -248,6 +310,10 @@ if (isPaged()) {
   write([pagedResponse(endpoint)]);
   process.exit(0);
 }
+if (/\\?(?:.*&)?per_page=/.test(endpoint)) {
+  write(pagedResponse(endpoint));
+  process.exit(0);
+}
 if (endpoint === "repos/openclaw/openclaw/issues/74134") {
   write(issue(74134, [], "Replacement PR: https://github.com/openclaw/openclaw/pull/74742"));
   process.exit(0);
@@ -282,6 +348,7 @@ function issue(number, labels, body) {
   };
 }
 function pull(number, sha) {
+  const large = process.env.FAKE_GH_LARGE_PR === "1";
   return {
     draft: false,
     merged: false,
@@ -300,11 +367,30 @@ function pull(number, sha) {
     requested_teams: [],
     additions: 1,
     deletions: 0,
-    changed_files: 1,
+    changed_files: large ? 120 : 1,
+    commits: large ? 120 : 1,
+    review_comments: 0,
   };
 }
 function pagedResponse(endpoint) {
-  if (endpoint.endsWith("/commits")) return [{ sha: "commit-sha", commit: { message: "test" }, author: { login: "contributor" } }];
+  const [endpointPath, query = ""] = endpoint.split("?");
+  const params = new URLSearchParams(query);
+  const limit = Math.max(1, Number(params.get("per_page") || 1));
+  if (endpointPath.endsWith("/files")) {
+    return Array.from({ length: limit }, (_, index) => ({
+      filename: "src/file-" + index + ".ts",
+      status: "modified",
+      additions: 1,
+      deletions: 0,
+    }));
+  }
+  if (endpointPath.endsWith("/commits")) {
+    return Array.from({ length: limit }, (_, index) => ({
+      sha: "commit-sha-" + index,
+      commit: { message: "test " + index },
+      author: { login: "contributor" },
+    }));
+  }
   return [];
 }
 `;