0 Fork 0

mirror of https://github.com/paperclipai/paperclip.git synced 2026-10-27 19:35:13 +01:00

Files

T

DottaandPaperclip db349f7670 feat(runner): publish eval integration contract

Co-Authored-By: Paperclip <noreply@paperclip.ing>

2026-08-11 20:49:39 +00:00

6.1 KiB

Raw Blame History

Native Runner Tutorials

The tutorials are cumulative and are always run from the repository root. Each tutorial explains the capability it exercises and the contract it proves.

Implemented capabilities

Conformance: standalone tracer — install the workspace, validate its Rust/TypeScript boundary and knowledge bundle, then run the deterministic Rust mock-core path and cross-language parity check.
Replay: static PRP replay — validate the shared schema/fixture corpus, reduce a fixture in the CLI, and inspect the same final snapshot in the standalone browser page.
Local runner: local runner and fake harness — run a supervised Rust process, exercise scripted live scenarios, resolve requests, interrupt a turn, and confirm live/replay parity.
Durable recovery: break recovery on purpose — lose an ACK, drop the socket, restart the runner side, replay durable events, and inspect recovery diagnostics.
Codex: run the skillless Codex driver — run a safe real-model task, inspect the exact context boundary, steer or interrupt it, and confirm one replay-safe result.
Live console: run the protocol demo server — exercise the server-only Codex boundary, canonical replay, typed controls, and reconnect with curl.
Live console: run the live Codex protocol console — chat with a live session in the browser, steer it, stop it three ways, answer its requests, change its goal, break its connection, and replay the record.
SDK: run the SDK console and mini consumer — exercise the versioned public browser/React surface in two independent consumers against fake and real drivers.
Standalone: run the thin Paperclip adapter — prove mock/real conformance, run one feature-flagged local task, inspect replay/finalization, disable the kill switch, and verify legacy fallback.
Capability: the Paperclip-style issue thread over a mock control plane — the canonical clean-start tutorial for the final build: prove the 106-case conformance suite, open the issue thread in deterministic fake mode, record the byte-stable screenshot matrix, and optionally drive a real Codex turn — all from a clean checkout with no Paperclip service.
Capability: explore the capability contract and scenario explorer — companion tour of the read-only browser explorer over the mock control plane and the focused verification set.
Scenario chat: chat with the mock control plane — the scenario-explorer chat surface: send prompts to a Capability scenario and watch the mock Paperclip activity for every turn: exposure, typed calls, denials, control-plane-owned actions, state diffs, wakes, and per-turn parity.
Capability: chat with real Codex in a clean room — the second primary path: open a blank chat on a freshly minted mock tenant, send a free-form message to real Codex through real runnerd, and inspect the tool, policy, and state evidence on demand.
Cumulative end-to-end tutorial — the shortest complete workflow for the current package.

Reference

Standalone is implemented as a default-off server integration at the public package boundary. Production Paperclip UI integration remains deferred.

6.1 KiB Raw Blame History

Native Runner Tutorials

Implemented capabilities

Reference

6.1 KiB

Raw Blame History