Files
PaperClipAI/packages/paperclip-runner/docs/design/sdk-component-decisions.md
T
Dotta 560e7e48b5 feat(runner): add SDK and developer tooling (#12608)
## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work.
> - The runner package already provides the production protocol and
execution spine.
> - Contributors still need stable SDK surfaces, deterministic test
tools, and local inspection tools.
> - Those surfaces share generated contracts and must change as one
package boundary.
> - This pull request adds the package-local SDK, labs, examples, and
drift checks.
> - The benefit is a reviewable developer platform that does not change
application execution selection.

## Linked Issues or Issue Description

**Subsystem affected**

`packages/paperclip-runner` — runner SDK, conformance tools, and
developer tooling.

**Problem or motivation**

The production runner spine is present, but package consumers cannot
build deterministic integrations, inspect sessions, or verify
provider-neutral behavior through supported surfaces.

**Proposed solution**

Add browser, React, standalone, live-session, scenario, conformance, and
evaluation surfaces. Add generated contract inventories and
package-local verification scripts. Keep production application routing
unchanged.

**Alternatives considered**

We considered splitting each generated catalog, SDK surface, and demo
into separate pull requests. Those changes share exports, fixtures, and
drift gates. Splitting them would create intermediate package states
that do not build.

**Roadmap alignment**

No overlapping item appears in `ROADMAP.md`. This work extends the
runner package that is already on `master`.

## What Changed

- Add browser, React, standalone, live-session, and issue-thread SDK
surfaces.
- Add deterministic mock control-plane, scenario, conformance, replay,
and evaluation tools.
- Add bounded Codex, OpenCode, and ACPX development transports and
fixtures.
- Keep deferred managed-provider execution fail-closed. Persisted
compatibility data remains readable.
- Add generated capability inventories with their source files and drift
checks.
- Add examples, package documentation, browser checks, and
clean-consumer checks.
- Preserve the reviewed protocol bounds, replay compatibility aliases,
process environment isolation, and semantic redaction limits.
- Update the ACPX package patch that the existing workspace patch
registry already tracks.
- Do not change `pnpm-lock.yaml`, repository workflows, server runtime
selection, or the application UI.

## Verification

GitHub Actions is the verification authority for this pull request. The
repository CI, package TypeScript and Rust checks, package tests,
generated-output drift checks, browser checks, security scans, and
Greptile review must pass on the exact head.

Local test suites were not run because this series uses parallel GitHub
Actions for verification.

## Risks

This is a large greenfield package change. The main risks are public
export drift, generated-output drift, and optional React consumer
compatibility. Package boundary checks, clean-consumer checks, and
browser tests cover those risks. Production adapter selection and server
execution are outside this pull request.

## Stack

1. **This PR:** runner SDK and developer tooling.
2. [Codex production server
integration](https://github.com/paperclipai/paperclip/pull/12616).
3. [Provider-neutral task-thread
UI](https://github.com/paperclipai/paperclip/pull/12617).

## Model Used

OpenAI Codex, GPT-5, high-reasoning mode, with tool use and code
execution.

## 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 described the issue in-PR following the feature request
template
- [x] I have not referenced internal/instance-local Paperclip issues or
links
- [x] My branch name describes the change and contains no internal
Paperclip ticket id
- [ ] 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 risks above
- [ ] All Paperclip CI gates are green
- [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge
2026-08-31 21:33:11 -05:00

5.2 KiB
Raw Blame History

SDK Extraction Decision Record

Date: 2026-08-08 Status: implemented

Outcome

The accepted Live console transport and reducer contracts were copied into a versioned 0.1.2 public surface. The reference console and a deliberately small second consumer import only public package subpaths. Live console and the Replay–3 surfaces remain the comparison baseline.

Promoted contracts

The implementation promotes the framework-free client, useRunnerConsole, all approved primitives and protocol views, and RunnerConsoleApp. Reducer state and canonical events remain the only rendering authority. File, tool, plan, terminal, failure, request, goal, lineage, connection, and replay views do not maintain private protocol state.

The package exports ., ./browser, ./react, and ./styles.css. React and React DOM are peers; the surface adds zero runtime dependencies.

Extension surface

The implementation has exactly the five approved extension points:

  1. item-body renderer;
  2. request-detail renderer;
  3. Composer leading and trailing slots;
  4. scoped --pcr-* token overrides;
  5. Fetch/EventSource/base-URL transport injection.

The mini consumer demonstrates all five. No new extension registry or headless contract was added during implementation.

Kept out

Markdown Response, highlighted CodeBlock, a persistent Context meter, a command palette, headless distribution, dark mode, Tailwind/Radix, toasts, and transcript virtualization remain rejected for the reasons in the approved plan. Default text and <pre> output keeps protocol evidence inspectable; consumers can opt into rich item bodies through the one renderer contract.

Implementation findings

  • Event IDs must not be deduplicated by the hook. Exact duplicate delivery is reducer input, so the shared reducer remains the sole duplicate authority.
  • Node types are required only for browser-test and Vite configuration. They are not part of the browser runtime or public dependency surface.
  • Token enforcement must inspect CSS custom-property definitions as well as component usage. The pcr- namespace is enforced across SDK sources.
  • The reference console renders one React tree selected at the 900px breakpoint. Rendering two trees would duplicate landmarks, test IDs, and live regions.
  • The compact layout keeps the transcript and composer as the primary view. A single accessible Menu opens session controls or the protocol inspector in a modal panel; secondary controls never replace or duplicate the chat tree.
  • A real provider may explicitly disable goals. The consumer gates buttons on that capability and preserves the upstream diagnostic instead of emulating a goal.
  • Real-provider steering timing is not a stable visual fixture. Fake-driver tests own race coverage; a safe real completion owns transport, identity, redaction, reconnect, and replay evidence.
  • Minimal hosts may lack Playwright shared libraries. The package rootless helper runs the same browser commands from a run-owned cache.
  • Assistant text uses the canonical reducer text but reveals queued additions progressively at the character level. Reduced-motion clients receive the complete current text immediately; protocol events and replay stay unchanged.
  • A completed assistant item that contains a semantic result JSON object shows its summary and disposition as the primary agent response. Completion checks, remaining work, and the exact protocol JSON stay behind a Completion details disclosure, while the protocol inspector continues to expose the full event record.
  • Transcript follow mode anchors the latest settled assistant response instead of terminal or diagnostic events appended after it. The complete ordered transcript remains available below the response, and manual scrolling still disengages follow mode.

Compatibility rule

0.1.2 is the current frozen SDK surface. Any later removal or semantic change to an export, hook field, component prop, data-slot, extension point, or token needs a versioned compatibility decision. Additive protocol fields stay forward-compatible; the schema and shared reducer remain authoritative.

Direct chat presentation

The reference console opens in a normal Codex chat mode. This mode sends plain user text and permits follow-up turns in the same provider thread. It does not add the Codex task envelope, semantic completion tools, or output schema.

The workspace sandbox and server-only provider authentication stay active. The deterministic Codex fixtures remain selectable for protocol tests. The chat layout removes decorative panel outlines and keeps the protocol inspector available on demand. The header includes a package version and an iteration emoji so a reviewer can identify the deployed build.

Terminal diagnostics and responsive wrapping

TranscriptItemEntry.debugEvents is an additive, optional projection of the canonical item events already retained by the client. ToolItem shows this data in a nested folded disclosure rather than inventing a second diagnostic model. Long commands, payloads, and assistant text wrap within the center column; mobile layouts keep every transcript descendant intrinsically shrinkable so the page does not gain a horizontal scrollbar.