Harden remote workspace sync and restore flows (#5444)

## Thinking Path

> - Paperclip orchestrates AI agents for zero-human companies
> - When an agent runs against a remote target, Paperclip syncs the
workspace out to the remote at run start and restores changes back to
the local workspace at run end
> - The previous restore flow naïvely overwrote local files with
whatever the remote returned, so files that the remote run never touched
but had timestamp/mode drift could be needlessly rewritten — and a
single static `refs/paperclip/ssh-sync/imported` ref made concurrent SSH
workspace exports race on the same git ref
> - This pull request adds a `workspace-restore-merge` module that diffs
a pre-run snapshot against the post-run remote state and only writes
back files the remote actually changed; SSH workspace exports now use a
per-import unique ref so concurrent runs can't trample each other
> - Every adapter's execute path threads the snapshot through
`prepareAdapterExecutionTargetRuntime` so the merge has the baseline it
needs
> - The benefit is workspace restores no longer churn untouched files,
and concurrent SSH runs no longer collide on the import ref

## What Changed

- `packages/adapter-utils/src/workspace-restore-merge.{ts,test.ts}`: new
module — directory snapshot (kind/mode/sha256/symlink target) plus
snapshot-aware merge that writes only the files the remote changed
- `packages/adapter-utils/src/ssh.ts`: SSH workspace export uses a
per-import unique ref (`refs/paperclip/ssh-sync/imported/<uuid>`);
restore goes through the new merge helper; `ssh-fixture.test.ts` covers
the unique-ref + merge paths
- `packages/adapter-utils/src/sandbox-managed-runtime.ts` +
`remote-managed-runtime.ts`: thread the snapshot/merge through the
sandbox and SSH paths
- `packages/adapter-utils/src/server-utils.{ts,test.ts}` +
`execution-target.ts`: helpers for capturing the pre-run snapshot;
`prepareAdapterExecutionTargetRuntime` gains required `runId` and
optional `workspaceRemoteDir`, and returns the realized
`workspaceRemoteDir`
- Each adapter's `execute.ts` (acpx, claude, codex, cursor, gemini,
opencode, pi) takes the snapshot at run start and passes it through to
the runtime restore
- Remote execute test mocks updated to match the new
`prepareWorkspaceForSshExecution` return shape and the per-run
`${managedRemoteWorkspace}` cwd subdirectory

## Verification

- `pnpm vitest run --no-coverage --project @paperclipai/adapter-utils
--project @paperclipai/adapter-acpx-local --project
@paperclipai/adapter-claude-local --project
@paperclipai/adapter-codex-local --project
@paperclipai/adapter-cursor-local --project
@paperclipai/adapter-gemini-local --project
@paperclipai/adapter-opencode-local --project
@paperclipai/adapter-pi-local` — 196/196 passing
- `pnpm typecheck` clean across the workspace

## Risks

Medium. The restore path now writes a strict subset of what it
previously did — files the remote did not touch are no longer rewritten.
If any flow was relying on a touch-without-content-change being copied
back (timestamp or permission propagation only), that behavior is now
skipped. Snapshot capture adds an O(N-files-in-workspace) hash pass at
run start; the cost is bounded by the existing exclude list. The `runId`
parameter on `prepareAdapterExecutionTargetRuntime` is now required —
every in-tree caller is updated; out-of-tree adapter authors need to
pass it.

## Model Used

Claude Opus 4.7 (1M context)

## 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 run tests locally and they pass
- [x] I have added or updated tests where applicable — new module +
every adapter execute path covered
- [x] If this change affects the UI, I have included before/after
screenshots — N/A (no UI)
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
This commit is contained in:
Devin Foley 2026-05-07 14:44:45 -07:00 committed by GitHub
parent 824298f414
commit 12cb7b40fd
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
23 changed files with 1234 additions and 183 deletions

View file

@ -27,7 +27,7 @@ const {
})),
ensureCommandResolvable: vi.fn(async () => undefined),
resolveCommandForLogs: vi.fn(async () => "ssh://fixture@127.0.0.1:2222/remote/workspace :: claude"),
prepareWorkspaceForSshExecution: vi.fn(async () => undefined),
prepareWorkspaceForSshExecution: vi.fn(async () => ({ gitBacked: false })),
restoreWorkspaceFromSshExecution: vi.fn(async () => undefined),
syncDirectoryToSsh: vi.fn(async () => undefined),
startAdapterExecutionTargetPaperclipBridge: vi.fn(async () => ({
@ -94,6 +94,7 @@ describe("claude remote execution", () => {
const workspaceDir = path.join(rootDir, "workspace");
const alternateWorkspaceDir = path.join(rootDir, "workspace-other");
const instructionsPath = path.join(rootDir, "instructions.md");
const managedRemoteWorkspace = "/remote/workspace/.paperclip-runtime/runs/run-1/workspace";
await mkdir(workspaceDir, { recursive: true });
await mkdir(alternateWorkspaceDir, { recursive: true });
await writeFile(instructionsPath, "Use the remote workspace.\n", "utf8");
@ -116,6 +117,11 @@ describe("claude remote execution", () => {
config: {
command: "claude",
instructionsFilePath: instructionsPath,
env: {
QA_PROJECT_WORKSPACE_CWD: workspaceDir,
RANDOM_WORKSPACE_CWD: workspaceDir,
OTHER_ENV: workspaceDir,
},
},
context: {
paperclipWorkspace: {
@ -161,11 +167,11 @@ describe("claude remote execution", () => {
expect(prepareWorkspaceForSshExecution).toHaveBeenCalledTimes(1);
expect(prepareWorkspaceForSshExecution).toHaveBeenCalledWith(expect.objectContaining({
localDir: workspaceDir,
remoteDir: "/remote/workspace",
remoteDir: managedRemoteWorkspace,
}));
expect(syncDirectoryToSsh).toHaveBeenCalledTimes(1);
expect(syncDirectoryToSsh).toHaveBeenCalledWith(expect.objectContaining({
remoteDir: "/remote/workspace/.paperclip-runtime/claude/skills",
remoteDir: `${managedRemoteWorkspace}/.paperclip-runtime/claude/skills`,
followSymlinks: true,
}));
expect(runChildProcess).toHaveBeenCalledTimes(1);
@ -173,15 +179,17 @@ describe("claude remote execution", () => {
| [string, string, string[], { env: Record<string, string>; remoteExecution?: { remoteCwd: string } | null }]
| undefined;
expect(call?.[2]).toContain("--append-system-prompt-file");
expect(call?.[2]).toContain("/remote/workspace/.paperclip-runtime/claude/skills/agent-instructions.md");
expect(call?.[2]).toContain(
`${managedRemoteWorkspace}/.paperclip-runtime/claude/skills/agent-instructions.md`,
);
expect(call?.[2]).toContain("--add-dir");
expect(call?.[2]).toContain("/remote/workspace/.paperclip-runtime/claude/skills");
expect(call?.[3].env.PAPERCLIP_WORKSPACE_CWD).toBe("/remote/workspace");
expect(call?.[2]).toContain(`${managedRemoteWorkspace}/.paperclip-runtime/claude/skills`);
expect(call?.[3].env.PAPERCLIP_WORKSPACE_CWD).toBe(managedRemoteWorkspace);
expect(call?.[3].env.PAPERCLIP_WORKSPACE_WORKTREE_PATH).toBeUndefined();
expect(JSON.parse(call?.[3].env.PAPERCLIP_WORKSPACES_JSON ?? "[]")).toEqual([
{
workspaceId: "workspace-1",
cwd: "/remote/workspace",
cwd: managedRemoteWorkspace,
repoUrl: "https://github.com/paperclipai/paperclip.git",
repoRef: "main",
},
@ -193,12 +201,15 @@ describe("claude remote execution", () => {
]);
expect(call?.[3].env.PAPERCLIP_API_URL).toBe("http://127.0.0.1:4310");
expect(call?.[3].env.PAPERCLIP_API_BRIDGE_MODE).toBe("queue_v1");
expect(call?.[3].remoteExecution?.remoteCwd).toBe("/remote/workspace");
expect(call?.[3].env.QA_PROJECT_WORKSPACE_CWD).toBe(managedRemoteWorkspace);
expect(call?.[3].env.RANDOM_WORKSPACE_CWD).toBe(managedRemoteWorkspace);
expect(call?.[3].env.OTHER_ENV).toBe(workspaceDir);
expect(call?.[3].remoteExecution?.remoteCwd).toBe(managedRemoteWorkspace);
expect(startAdapterExecutionTargetPaperclipBridge).toHaveBeenCalledTimes(1);
expect(restoreWorkspaceFromSshExecution).toHaveBeenCalledTimes(1);
expect(restoreWorkspaceFromSshExecution).toHaveBeenCalledWith(expect.objectContaining({
localDir: workspaceDir,
remoteDir: "/remote/workspace",
remoteDir: managedRemoteWorkspace,
}));
});
@ -259,6 +270,7 @@ describe("claude remote execution", () => {
const rootDir = await mkdtemp(path.join(os.tmpdir(), "paperclip-claude-remote-resume-match-"));
cleanupDirs.push(rootDir);
const workspaceDir = path.join(rootDir, "workspace");
const managedRemoteWorkspace = "/remote/workspace/.paperclip-runtime/runs/run-ssh-resume/workspace";
await mkdir(workspaceDir, { recursive: true });
await execute({
@ -274,13 +286,13 @@ describe("claude remote execution", () => {
sessionId: "session-123",
sessionParams: {
sessionId: "session-123",
cwd: "/remote/workspace",
cwd: managedRemoteWorkspace,
remoteExecution: {
transport: "ssh",
host: "127.0.0.1",
port: 2222,
username: "fixture",
remoteCwd: "/remote/workspace",
remoteCwd: managedRemoteWorkspace,
},
},
sessionDisplayId: "session-123",

View file

@ -6,6 +6,7 @@ import type { RunProcessResult } from "@paperclipai/adapter-utils/server-utils";
import {
adapterExecutionTargetIsRemote,
adapterExecutionTargetRemoteCwd,
overrideAdapterExecutionTargetRemoteCwd,
adapterExecutionTargetSessionIdentity,
adapterExecutionTargetSessionMatches,
adapterExecutionTargetUsesManagedHome,
@ -35,8 +36,10 @@ import {
buildInvocationEnvForLogs,
ensureAbsoluteDirectory,
ensurePathInEnv,
refreshPaperclipWorkspaceEnvForExecution,
renderTemplate,
renderPaperclipWakePrompt,
rewriteWorkspaceCwdEnvVarsForExecution,
shapePaperclipWorkspaceEnvForExecution,
stringifyPaperclipWakePayload,
DEFAULT_PAPERCLIP_AGENT_PROMPT_TEMPLATE,
@ -152,7 +155,7 @@ async function buildClaudeRuntimeConfig(input: ClaudeExecutionInput): Promise<Cl
const effectiveWorkspaceCwd = useConfiguredInsteadOfAgentHome ? "" : workspaceCwd;
const cwd = effectiveWorkspaceCwd || configuredCwd || process.cwd();
const executionTargetIsRemote = adapterExecutionTargetIsRemote(executionTarget);
const effectiveExecutionCwd = adapterExecutionTargetRemoteCwd(executionTarget, cwd);
let effectiveExecutionCwd = adapterExecutionTargetRemoteCwd(executionTarget, cwd);
const shapedWorkspaceEnv = shapePaperclipWorkspaceEnvForExecution({
workspaceCwd: effectiveWorkspaceCwd,
workspaceWorktreePath,
@ -241,7 +244,13 @@ async function buildClaudeRuntimeConfig(input: ClaudeExecutionInput): Promise<Cl
if (runtimePrimaryUrl) {
env.PAPERCLIP_RUNTIME_PRIMARY_URL = runtimePrimaryUrl;
}
for (const [key, value] of Object.entries(envConfig)) {
const shapedEnvConfig = rewriteWorkspaceCwdEnvVarsForExecution({
env: envConfig,
workspaceCwd: effectiveWorkspaceCwd,
executionCwd: shapedWorkspaceEnv.workspaceCwd,
executionTargetIsRemote,
});
for (const [key, value] of Object.entries(shapedEnvConfig)) {
if (typeof value === "string") env[key] = value;
}
@ -351,6 +360,21 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
const maxTurns = asNumber(config.maxTurnsPerRun, 0);
const dangerouslySkipPermissions = asBoolean(config.dangerouslySkipPermissions, true);
const configEnv = parseObject(config.env);
const workspaceContext = parseObject(context.paperclipWorkspace);
const workspaceCwd = asString(workspaceContext.cwd, "");
const workspaceSource = asString(workspaceContext.source, "");
const workspaceStrategy = asString(workspaceContext.strategy, "");
const workspaceBranch = asString(workspaceContext.branchName, "") || null;
const workspaceWorktreePath = asString(workspaceContext.worktreePath, "") || null;
const agentHome = asString(workspaceContext.agentHome, "") || null;
const workspaceHints = Array.isArray(context.paperclipWorkspaces)
? context.paperclipWorkspaces.filter(
(value): value is Record<string, unknown> => typeof value === "object" && value !== null,
)
: [];
const configuredCwd = asString(config.cwd, "");
const useConfiguredInsteadOfAgentHome = workspaceSource === "agent_home" && configuredCwd.length > 0;
const effectiveWorkspaceCwd = useConfiguredInsteadOfAgentHome ? "" : workspaceCwd;
const hasExplicitClaudeConfigDir =
typeof configEnv.CLAUDE_CONFIG_DIR === "string" && configEnv.CLAUDE_CONFIG_DIR.trim().length > 0;
const instructionsFilePath = asString(config.instructionsFilePath, "").trim();
@ -379,7 +403,7 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
extraArgs,
} = runtimeConfig;
let loggedEnv = initialLoggedEnv;
const effectiveExecutionCwd = adapterExecutionTargetRemoteCwd(executionTarget, cwd);
let effectiveExecutionCwd = adapterExecutionTargetRemoteCwd(executionTarget, cwd);
const terminalResultCleanupGraceMs = Math.max(
0,
asNumber(config.terminalResultCleanupGraceMs, 5_000),
@ -433,6 +457,7 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
`[paperclip] Syncing workspace and Claude runtime assets to ${describeAdapterExecutionTarget(executionTarget)}.\n`,
);
return await prepareAdapterExecutionTargetRuntime({
runId,
target: executionTarget,
adapterKey: "claude",
workspaceLocalDir: cwd,
@ -455,6 +480,26 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
});
})()
: null;
if (preparedExecutionTargetRuntime?.workspaceRemoteDir) {
effectiveExecutionCwd = preparedExecutionTargetRuntime.workspaceRemoteDir;
}
const runtimeExecutionTarget = overrideAdapterExecutionTargetRemoteCwd(executionTarget, effectiveExecutionCwd);
refreshPaperclipWorkspaceEnvForExecution({
env,
envConfig: configEnv,
workspaceCwd: effectiveWorkspaceCwd,
workspaceSource,
workspaceStrategy,
workspaceId,
workspaceRepoUrl,
workspaceRepoRef,
workspaceBranch,
workspaceWorktreePath,
workspaceHints,
agentHome,
executionTargetIsRemote,
executionCwd: effectiveExecutionCwd,
});
const restoreRemoteWorkspace = preparedExecutionTargetRuntime
? () => preparedExecutionTargetRuntime.restoreWorkspace()
: null;
@ -502,10 +547,10 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
);
}
let paperclipBridge: Awaited<ReturnType<typeof startAdapterExecutionTargetPaperclipBridge>> = null;
if (executionTargetIsRemote && adapterExecutionTargetUsesPaperclipBridge(executionTarget)) {
if (executionTargetIsRemote && adapterExecutionTargetUsesPaperclipBridge(runtimeExecutionTarget)) {
paperclipBridge = await startAdapterExecutionTargetPaperclipBridge({
runId,
target: executionTarget,
target: runtimeExecutionTarget,
runtimeRootDir: preparedExecutionTargetRuntime?.runtimeRootDir,
adapterKey: "claude",
hostApiToken: env.PAPERCLIP_API_KEY,
@ -536,7 +581,7 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
runtimeSessionId.length > 0 &&
hasMatchingPromptBundle &&
(runtimeSessionCwd.length === 0 || path.resolve(runtimeSessionCwd) === path.resolve(effectiveExecutionCwd)) &&
adapterExecutionTargetSessionMatches(runtimeRemoteExecution, executionTarget);
adapterExecutionTargetSessionMatches(runtimeRemoteExecution, runtimeExecutionTarget);
const sessionId = canResumeSession ? runtimeSessionId : null;
if (
executionTargetIsRemote &&
@ -672,7 +717,7 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
});
}
const proc = await runAdapterExecutionTargetProcess(runId, executionTarget, command, args, {
const proc = await runAdapterExecutionTargetProcess(runId, runtimeExecutionTarget, command, args, {
cwd,
env,
stdin: prompt,
@ -793,7 +838,7 @@ export async function execute(ctx: AdapterExecutionContext): Promise<AdapterExec
promptBundleKey: promptBundle.bundleKey,
...(executionTargetIsRemote
? {
remoteExecution: adapterExecutionTargetSessionIdentity(executionTarget),
remoteExecution: adapterExecutionTargetSessionIdentity(runtimeExecutionTarget),
}
: {}),
...(workspaceId ? { workspaceId } : {}),