refactor(tools): isolate GitHub access discovery (#17144)

## Thinking Path

> - Paperclip manages AI agents and their access to external tools.
> - Tool connections discover repositories through GitHub account
grants.
> - The tool-access service mixes GitHub response parsing with grant
authorization and persistence.
> - These responsibilities have different inputs and failure rules.
> - This PR puts GitHub discovery in one provider module.
> - The service keeps current authority checks and stored grant updates.

## Linked Issues or Issue Description

**What existing behavior does this improve?**

The ownership and review of GitHub access discovery in the tool-access
service.

**Current behavior**

The 21,525-line service contains GitHub pagination, account metadata
validation, and grant lifecycle code.

**Proposed behavior**

A GitHub module returns normalized access metadata. Existing service
exports remain compatible. This is the first PR in a three-part refactor
stack. No behavior change is intended. A search found no existing
tool-access refactor PR or issue.

Stack order: this PR, #17146, then #17147. Each PR targets its
predecessor.

## What Changed

- Move six functions into `tool-access/github-access-discovery.ts`.
Their bodies are unchanged.
- Keep grant selection, credential resolution, refresh retry, membership
locks, revision checks, and webhook updates in the service.
- Add public-entry characterization tests for HTTP error classification
and invalid pagination.
- Record five candidate boundaries and the initial three-PR scope in the
dated plan.
- Reduce `tool-access.ts` from 21,525 lines / 829,519 bytes to 21,237
lines / 820,116 bytes. The new module is 294 lines / 9,722 bytes.

## Verification

- Before and after: `pnpm exec vitest run
server/src/__tests__/github-grant-metadata.test.ts
server/src/__tests__/project-repositories.test.ts` — 13 tests pass.
- Before and after: `pnpm exec vitest run
server/src/__tests__/tool-access-service.test.ts -t 'GitHub|github'` —
11 selected tests pass with disposable PostgreSQL; 455 cases are outside
this selection.
- `pnpm --filter @paperclipai/server typecheck` passes.
- Exact text comparison confirms all six moved bodies match the base.
`git diff --check` passes.
- The initial baseline attempt failed because the plugin SDK had not
been built. The normal workspace prerequisite build resolved it before
behavioral validation.
- Final-stack local verification: `pnpm -r --workspace-concurrency=2
typecheck` and `pnpm build` pass. CI is green at
`07e78dd1b1c26e0bd4234006766444c71e234451`. Greptile is 5/5 with no
actionable findings.

- Full local-suite limitation: `pnpm test:run` was interrupted after 79
minutes during the serial general-server phase, after CI had completed
the full test matrix. Before interruption, it flagged the unchanged
`company-search-service.test.ts` suite (26 database cases skipped). One
isolated rerun passed all 28 cases without code changes. The original
log and a pre-interruption copy are preserved. The original failure
cause was not reproduced. The full local suite and its later groups are
not claimed as passed; the complete CI matrix is green on this stack.

## Risks

- Import wiring could break existing consumers. Existing imports remain
supported and all four production call sites are unchanged.
- Provider requests retain their ordering, timeout, fixed GitHub origin,
and response redaction. The new module has no database dependency or
shared mutable state.
- Tests simulate GitHub responses. They do not qualify live GitHub
access.

## Model Used

OpenAI Codex, GPT-6 family, with code execution and repository tools.
The exact serving model ID and context-window size are not exposed in
this session.

## Checklist


- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] All Paperclip CI gates are green
- [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge

Co-authored-by: Paperclip <noreply@paperclip.ing>
This commit is contained in:
DottaandPaperclip authored and GitHub committed 2026-10-21 22:21:40 -05:00
1 parent 0081863463
commit 9db3a5726e
4 files changed
+365 -290

No files matched your search

@@ -92,4 +92,27 @@ describe("GitHub grant metadata", () => {
details: expect.objectContaining({ code: "github_access_check_failed" }),
});
});
it.each([
[401, "oauth_reauthorization_required"],
[503, "github_access_check_failed"],
])("preserves the retry decision for HTTP %s without reflecting provider text", async (status, code) => {
const request = vi.fn<typeof fetch>().mockResolvedValue(new Response("private provider detail", { status }));
await expect(loadGitHubGrantMetadata("ghu_secret", request)).rejects.toMatchObject({
status: 422,
message: "GitHub could not verify this account's installation access",
details: { code },
});
expect(request).toHaveBeenCalledTimes(1);
});
it("rejects an empty next page before reporting an incomplete access snapshot", async () => {
const request = vi.fn<typeof fetch>()
.mockResolvedValueOnce(json({ id: 42, login: "octocat" }))
.mockResolvedValueOnce(json({ installations: [] }, true));
await expect(loadGitHubGrantMetadata("ghu_secret", request)).rejects.toMatchObject({
details: { code: "github_bad_response" },
});
expect(request).toHaveBeenCalledTimes(2);
});
});
+2 -290
View File
@@ -1,3 +1,5 @@
import { loadGitHubGrantMetadata, loadGitHubTokenRepositories } from "./tool-access/github-access-discovery.js";
export { loadGitHubGrantMetadata, loadGitHubTokenRepositories } from "./tool-access/github-access-discovery.js";
import { getAppDefinitionForApplicationKey, isConnectorFeatureEnabled } from "@paperclipai/shared";
import { stableHash, descriptorHash } from "./tool-catalog-hash.js";
import { assertProfileEntryInputs } from "./tool-profile-entry-validation.js";
@@ -2602,296 +2604,6 @@ function managedConnectorProfile(value: string | undefined): {
return null;
}
function gitHubRepositoryOwnerType(repository: Record<string, unknown>) {
const type = recordValue(repository.owner) ? repository.owner.type : undefined;
return type === "Organization" ? "organization" as const
: type === "User" ? "personal" as const : undefined;
}
export async function loadGitHubTokenRepositories(
headers: Record<string, string>,
request: typeof fetch = fetch,
) {
const repositories: Array<{
id: string;
fullName: string;
ownerType?: "personal" | "organization";
private?: boolean;
}> = [];
for (let page = 1; ; page += 1) {
const response = await request(
`https://api.github.com/user/repos?per_page=100&page=${page}`,
{
headers: {
...headers,
accept: "application/vnd.github+json",
"user-agent": "Paperclip",
"x-github-api-version": "2022-11-28",
},
signal: AbortSignal.timeout(15_000),
},
);
if (!response.ok)
throw unprocessable(
"Could not load GitHub repositories. Reconnect GitHub and try again.",
);
const rows: unknown = await response.json();
if (!Array.isArray(rows))
throw unprocessable("GitHub returned invalid repositories");
for (const row of rows) {
if (
!recordValue(row) ||
!githubId(row.id) ||
typeof row.full_name !== "string" ||
!/^[A-Za-z0-9][A-Za-z0-9-]*\/(?!\.{1,2}$)[A-Za-z0-9_.-]+$/.test(
row.full_name,
)
) {
throw unprocessable("GitHub returned invalid repository metadata");
}
repositories.push({
id: githubId(row.id)!,
fullName: row.full_name,
...(gitHubRepositoryOwnerType(row) ? { ownerType: gitHubRepositoryOwnerType(row) } : {}),
...(typeof row.private === "boolean" ? { private: row.private } : {}),
});
}
if (!/;\s*rel="next"/.test(response.headers.get("link") ?? ""))
return repositories;
if (!rows.length) throw unprocessable("GitHub returned invalid pagination");
}
}
export async function loadGitHubGrantMetadata(
accessToken: string,
request: typeof fetch = fetch,
appSlug?: string,
): Promise<{
userId: string;
login: string;
avatarUrl?: string;
installationCount: number;
repositoryCount: number;
repositorySelection: "all" | "selected" | "mixed" | "none";
installationIds: string[];
installationOwnerLogins: string[];
repositories: Array<{
id: string;
fullName: string;
installationId: string;
ownerType?: "personal" | "organization";
private?: boolean;
}>;
installationUrl: string;
managementUrl: string;
appSlug?: string;
accessRevision: string;
lastAccessRefreshAt: string;
webhookHealth: "pending";
}> {
let resolvedAppSlug = appSlug;
const accessRefreshStartedAt = new Date().toISOString();
const github = async (
path: string,
): Promise<{ data: Record<string, unknown>; hasNext: boolean }> => {
const response = await request(`https://api.github.com${path}`, {
headers: {
accept: "application/vnd.github+json",
authorization: `Bearer ${accessToken}`,
"user-agent": "Paperclip",
"x-github-api-version": "2022-11-28",
},
signal: AbortSignal.timeout(15_000),
});
if (!response.ok) {
throw unprocessable(
"GitHub could not verify this account's installation access",
{
code:
response.status === 401
? "oauth_reauthorization_required"
: "github_access_check_failed",
},
);
}
const value = (await response.json()) as unknown;
if (!recordValue(value))
throw unprocessable("GitHub returned invalid account metadata", {
code: "github_bad_response",
});
return {
data: value,
hasNext: /;\s*rel="next"/.test(response.headers.get("link") ?? ""),
};
};
const list = async (
path: string,
key: string,
): Promise<Record<string, unknown>[]> => {
const items: Record<string, unknown>[] = [];
for (let page = 1; ; page += 1) {
const { data, hasNext } = await github(
`${path}?per_page=100&page=${page}`,
);
const batch = data[key];
if (
!Array.isArray(batch) ||
!batch.every(recordValue) ||
(hasNext && batch.length === 0)
) {
throw unprocessable("GitHub returned invalid access metadata", {
code: "github_bad_response",
});
}
items.push(...batch);
if (!hasNext) return items;
}
};
const { data: user } = await github("/user");
const userId = githubId(user.id);
const login = typeof user.login === "string" ? user.login : null;
if (!userId || !login)
throw unprocessable("GitHub returned invalid account metadata", {
code: "github_bad_response",
});
const installations = await list("/user/installations", "installations");
const installationIds: string[] = [];
const owners = new Set<string>();
const selections = new Set<"all" | "selected">();
const managementUrls = new Set<string>();
const repositories = new Map<
string,
{ id: string; fullName: string; installationId: string; ownerType?: "personal" | "organization"; private?: boolean }
>();
for (const installation of installations) {
const installationId = githubId(installation.id);
if (!installationId) continue;
installationIds.push(installationId);
// Older grants predate the broker's appSlug field. GitHub's installation
// response identifies this token's app without choosing an environment.
if (
!resolvedAppSlug &&
typeof installation.app_slug === "string" &&
/^[a-z0-9-]{1,100}$/.test(installation.app_slug)
) {
resolvedAppSlug = installation.app_slug;
}
if (
installation.repository_selection === "all" ||
installation.repository_selection === "selected"
) {
selections.add(installation.repository_selection);
}
const account = recordValue(installation.account)
? installation.account
: null;
if (typeof account?.login === "string") owners.add(account.login);
const managementUrl = githubInstallationManagementUrl(
installation.html_url,
);
if (managementUrl) managementUrls.add(managementUrl);
for (const repository of await list(
`/user/installations/${installationId}/repositories`,
"repositories",
)) {
const id = githubId(repository.id);
const fullName =
typeof repository.full_name === "string" ? repository.full_name : "";
if (
!id ||
!/^[A-Za-z0-9][A-Za-z0-9-]*\/(?!\.{1,2}$)[A-Za-z0-9_.-]+$/.test(
fullName,
)
) {
throw unprocessable("GitHub returned invalid repository metadata", {
code: "github_bad_response",
});
}
repositories.set(id, {
id,
fullName,
installationId,
...(gitHubRepositoryOwnerType(repository) ? { ownerType: gitHubRepositoryOwnerType(repository) } : {}),
...(typeof repository.private === "boolean"
? { private: repository.private }
: {}),
});
}
}
const repositoryCount = repositories.size;
if (installationIds.length === 0 || repositoryCount === 0) {
const installationUrl = resolvedAppSlug
? `https://github.com/apps/${resolvedAppSlug}/installations/new`
: "https://github.com/settings/installations";
throw unprocessable(
"GitHub access is required. Install Paperclip and grant at least one repository before refreshing access.",
{
code: "github_installation_required",
installationUrl,
managementUrl: "https://github.com/settings/installations",
},
);
}
const installationUrl = resolvedAppSlug
? `https://github.com/apps/${resolvedAppSlug}/installations/new`
: "https://github.com/settings/installations";
return {
userId,
login,
...(typeof user.avatar_url === "string"
? { avatarUrl: user.avatar_url }
: {}),
installationCount: installationIds.length,
repositoryCount,
repositorySelection:
selections.size > 1
? "mixed"
: (selections.values().next().value ?? "none"),
installationIds,
installationOwnerLogins: [...owners],
repositories: [...repositories.values()].sort((a, b) =>
a.fullName.localeCompare(b.fullName),
),
installationUrl,
managementUrl:
managementUrls.size === 1
? managementUrls.values().next().value!
: "https://github.com/settings/installations",
...(resolvedAppSlug ? { appSlug: resolvedAppSlug } : {}),
accessRevision: randomUUID(),
lastAccessRefreshAt: accessRefreshStartedAt,
webhookHealth: "pending",
};
}
function githubInstallationManagementUrl(value: unknown): string | null {
if (typeof value !== "string" || value.length > 2_000) return null;
try {
const url = new URL(value);
if (
url.protocol !== "https:" ||
url.hostname.toLowerCase() !== "github.com"
)
return null;
return url.pathname.includes("/settings/installations/")
? url.toString()
: null;
} catch {
return null;
}
}
function githubId(value: unknown): string | null {
if (typeof value === "number" && Number.isSafeInteger(value) && value > 0)
return String(value);
if (typeof value === "string" && /^[1-9][0-9]{0,30}$/.test(value))
return value;
return null;
}
function recordValue(value: unknown): value is Record<string, unknown> {
return typeof value === "object" && value !== null && !Array.isArray(value);
}
/**
@@ -0,0 +1,294 @@
import { randomUUID } from "node:crypto";
import { unprocessable } from "../../errors.js";
function gitHubRepositoryOwnerType(repository: Record<string, unknown>) {
const type = recordValue(repository.owner) ? repository.owner.type : undefined;
return type === "Organization" ? "organization" as const
: type === "User" ? "personal" as const : undefined;
}
export async function loadGitHubTokenRepositories(
headers: Record<string, string>,
request: typeof fetch = fetch,
) {
const repositories: Array<{
id: string;
fullName: string;
ownerType?: "personal" | "organization";
private?: boolean;
}> = [];
for (let page = 1; ; page += 1) {
const response = await request(
`https://api.github.com/user/repos?per_page=100&page=${page}`,
{
headers: {
...headers,
accept: "application/vnd.github+json",
"user-agent": "Paperclip",
"x-github-api-version": "2022-11-28",
},
signal: AbortSignal.timeout(15_000),
},
);
if (!response.ok)
throw unprocessable(
"Could not load GitHub repositories. Reconnect GitHub and try again.",
);
const rows: unknown = await response.json();
if (!Array.isArray(rows))
throw unprocessable("GitHub returned invalid repositories");
for (const row of rows) {
if (
!recordValue(row) ||
!githubId(row.id) ||
typeof row.full_name !== "string" ||
!/^[A-Za-z0-9][A-Za-z0-9-]*\/(?!\.{1,2}$)[A-Za-z0-9_.-]+$/.test(
row.full_name,
)
) {
throw unprocessable("GitHub returned invalid repository metadata");
}
repositories.push({
id: githubId(row.id)!,
fullName: row.full_name,
...(gitHubRepositoryOwnerType(row) ? { ownerType: gitHubRepositoryOwnerType(row) } : {}),
...(typeof row.private === "boolean" ? { private: row.private } : {}),
});
}
if (!/;\s*rel="next"/.test(response.headers.get("link") ?? ""))
return repositories;
if (!rows.length) throw unprocessable("GitHub returned invalid pagination");
}
}
export async function loadGitHubGrantMetadata(
accessToken: string,
request: typeof fetch = fetch,
appSlug?: string,
): Promise<{
userId: string;
login: string;
avatarUrl?: string;
installationCount: number;
repositoryCount: number;
repositorySelection: "all" | "selected" | "mixed" | "none";
installationIds: string[];
installationOwnerLogins: string[];
repositories: Array<{
id: string;
fullName: string;
installationId: string;
ownerType?: "personal" | "organization";
private?: boolean;
}>;
installationUrl: string;
managementUrl: string;
appSlug?: string;
accessRevision: string;
lastAccessRefreshAt: string;
webhookHealth: "pending";
}> {
let resolvedAppSlug = appSlug;
const accessRefreshStartedAt = new Date().toISOString();
const github = async (
path: string,
): Promise<{ data: Record<string, unknown>; hasNext: boolean }> => {
const response = await request(`https://api.github.com${path}`, {
headers: {
accept: "application/vnd.github+json",
authorization: `Bearer ${accessToken}`,
"user-agent": "Paperclip",
"x-github-api-version": "2022-11-28",
},
signal: AbortSignal.timeout(15_000),
});
if (!response.ok) {
throw unprocessable(
"GitHub could not verify this account's installation access",
{
code:
response.status === 401
? "oauth_reauthorization_required"
: "github_access_check_failed",
},
);
}
const value = (await response.json()) as unknown;
if (!recordValue(value))
throw unprocessable("GitHub returned invalid account metadata", {
code: "github_bad_response",
});
return {
data: value,
hasNext: /;\s*rel="next"/.test(response.headers.get("link") ?? ""),
};
};
const list = async (
path: string,
key: string,
): Promise<Record<string, unknown>[]> => {
const items: Record<string, unknown>[] = [];
for (let page = 1; ; page += 1) {
const { data, hasNext } = await github(
`${path}?per_page=100&page=${page}`,
);
const batch = data[key];
if (
!Array.isArray(batch) ||
!batch.every(recordValue) ||
(hasNext && batch.length === 0)
) {
throw unprocessable("GitHub returned invalid access metadata", {
code: "github_bad_response",
});
}
items.push(...batch);
if (!hasNext) return items;
}
};
const { data: user } = await github("/user");
const userId = githubId(user.id);
const login = typeof user.login === "string" ? user.login : null;
if (!userId || !login)
throw unprocessable("GitHub returned invalid account metadata", {
code: "github_bad_response",
});
const installations = await list("/user/installations", "installations");
const installationIds: string[] = [];
const owners = new Set<string>();
const selections = new Set<"all" | "selected">();
const managementUrls = new Set<string>();
const repositories = new Map<
string,
{ id: string; fullName: string; installationId: string; ownerType?: "personal" | "organization"; private?: boolean }
>();
for (const installation of installations) {
const installationId = githubId(installation.id);
if (!installationId) continue;
installationIds.push(installationId);
// Older grants predate the broker's appSlug field. GitHub's installation
// response identifies this token's app without choosing an environment.
if (
!resolvedAppSlug &&
typeof installation.app_slug === "string" &&
/^[a-z0-9-]{1,100}$/.test(installation.app_slug)
) {
resolvedAppSlug = installation.app_slug;
}
if (
installation.repository_selection === "all" ||
installation.repository_selection === "selected"
) {
selections.add(installation.repository_selection);
}
const account = recordValue(installation.account)
? installation.account
: null;
if (typeof account?.login === "string") owners.add(account.login);
const managementUrl = githubInstallationManagementUrl(
installation.html_url,
);
if (managementUrl) managementUrls.add(managementUrl);
for (const repository of await list(
`/user/installations/${installationId}/repositories`,
"repositories",
)) {
const id = githubId(repository.id);
const fullName =
typeof repository.full_name === "string" ? repository.full_name : "";
if (
!id ||
!/^[A-Za-z0-9][A-Za-z0-9-]*\/(?!\.{1,2}$)[A-Za-z0-9_.-]+$/.test(
fullName,
)
) {
throw unprocessable("GitHub returned invalid repository metadata", {
code: "github_bad_response",
});
}
repositories.set(id, {
id,
fullName,
installationId,
...(gitHubRepositoryOwnerType(repository) ? { ownerType: gitHubRepositoryOwnerType(repository) } : {}),
...(typeof repository.private === "boolean"
? { private: repository.private }
: {}),
});
}
}
const repositoryCount = repositories.size;
if (installationIds.length === 0 || repositoryCount === 0) {
const installationUrl = resolvedAppSlug
? `https://github.com/apps/${resolvedAppSlug}/installations/new`
: "https://github.com/settings/installations";
throw unprocessable(
"GitHub access is required. Install Paperclip and grant at least one repository before refreshing access.",
{
code: "github_installation_required",
installationUrl,
managementUrl: "https://github.com/settings/installations",
},
);
}
const installationUrl = resolvedAppSlug
? `https://github.com/apps/${resolvedAppSlug}/installations/new`
: "https://github.com/settings/installations";
return {
userId,
login,
...(typeof user.avatar_url === "string"
? { avatarUrl: user.avatar_url }
: {}),
installationCount: installationIds.length,
repositoryCount,
repositorySelection:
selections.size > 1
? "mixed"
: (selections.values().next().value ?? "none"),
installationIds,
installationOwnerLogins: [...owners],
repositories: [...repositories.values()].sort((a, b) =>
a.fullName.localeCompare(b.fullName),
),
installationUrl,
managementUrl:
managementUrls.size === 1
? managementUrls.values().next().value!
: "https://github.com/settings/installations",
...(resolvedAppSlug ? { appSlug: resolvedAppSlug } : {}),
accessRevision: randomUUID(),
lastAccessRefreshAt: accessRefreshStartedAt,
webhookHealth: "pending",
};
}
function githubInstallationManagementUrl(value: unknown): string | null {
if (typeof value !== "string" || value.length > 2_000) return null;
try {
const url = new URL(value);
if (
url.protocol !== "https:" ||
url.hostname.toLowerCase() !== "github.com"
)
return null;
return url.pathname.includes("/settings/installations/")
? url.toString()
: null;
} catch {
return null;
}
}
function githubId(value: unknown): string | null {
if (typeof value === "number" && Number.isSafeInteger(value) && value > 0)
return String(value);
if (typeof value === "string" && /^[1-9][0-9]{0,30}$/.test(value))
return value;
return null;
}
function recordValue(value: unknown): value is Record<string, unknown> {
return typeof value === "object" && value !== null && !Array.isArray(value);
}