Files
PaperClipAI/server
DottaandPaperclip ae2d2956b3 Reject relative local Runner paths before native admission (#19776)
## Thinking Path

> - Paperclip manages AI agents and their work.
> - Heartbeat selects the runtime that will execute a task.
> - A new local Native run uses the controller's Runner binary setting.
> - A relative binary path is invalid. The resolver already rejects it,
but that check occurs after Native admission.
> - This pull request checks the same setting before it creates Native
execution authority.
> - Remote providers and admitted Native recovery keep their existing
behavior.

## Linked Issues or Issue Description

**What happened?**

A relative `PAPERCLIP_RUNNER_BINARY` setting can reach local Native
admission before the binary resolver rejects it. The run has transferred
preparation ownership even though this setting cannot start the local
Runner.

**Expected behavior**

Reject this invalid setting before creating a completion contract or
Native coordinator. Keep the existing preparation owner and the run row
unchanged.

**Steps to reproduce**

Set `PAPERCLIP_RUNNER_BINARY=relative-runner`. Select a new local
`paperclip_runner` Native run. The selection must report the existing
absolute-path error before Native admission.

Refs: #19466 and #19340. Those merged changes repair Retry authority and
saved-workspace recovery. This change only moves local path validation
before new admission.

## What Changed

- Add a pure absolute-path check beside the existing binary resolver.
Both use the actual controller environment setting.
- Run that check before a new local Runner admission. Skip remote
execution, API-only providers, and an already admitted Native run.
- Test the rejection, unchanged preparation ownership, controller
setting selection, remote behavior, and immutable recovery. Keep the
existing resolver checks.

## Verification

- The full binary test file passed: 32 tests.
- The full runtime-selection file passed against a dedicated PostgreSQL
database: 62 tests. It exercises real admission transactions, contracts,
coordinator rows, and instruction-copy controls. It replaces context
materialization and remote workspace synchronization at their existing
test boundaries.
- After normal integration with current master, the normal Shared/SDK,
database, eval-kernel and Runner builds passed, together with all seven
normal application TypeScript programs. The original
recovery-preparation case passed again on this integrated revision
against a fresh dedicated database (1 passed; 61 filtered). The earlier
full 32-test binary and 62-test selection results carry only where their
source and executed boundaries are unchanged.
- The first private selection wrapper failed during module resolution,
before any case or database use. A TypeScript wrapper with the same body
recovered that setup failure. This was not a product regression.
- Reproduce with `pnpm --filter @paperclipai/server exec vitest run
src/services/native-runtime/native-codex-runner.test.ts --maxWorkers=1
--no-file-parallelism`.
- Set `PAPERCLIP_RUNTIME_SELECTION_TEST_DATABASE_URL` to a dedicated
disposable test database before running `pnpm --filter
@paperclipai/server exec vitest run
src/services/heartbeat/runtime-selection.test.ts --maxWorkers=1
--no-file-parallelism`.

## Risks

- This check validates path syntax only. An absolute path can still fail
the existing file-access or artifact checks later.
- A wrong branch condition could reject remote or existing Native work.
The tests cover these exclusions.
- This does not prove that a real provider can start or that a deployed
service is healthy. It changes no database schema or retry authority.

## Model Used

OpenAI `gpt-6.1-sol` through Codex, with source inspection and
code-execution tools. The host does not expose the context-window size.

## 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>
2026-11-07 01:59:09 -06:00
..