paperclip/packages/paperclip-runner/docs
nickyleach fe4f55462c
refactor: remove the wake payload from the launch environment
The server sent the wake payload two times: once as fenced JSON in the
rendered prompt, and once as the PAPERCLIP_WAKE_PAYLOAD_JSON launch
environment variable. Only the environment copy passed through the
base64 launch envelope, and only that envelope has a per-string size
limit (131,072 bytes). A normal driver task's payload can use over
100% of that limit on its own.

The prompt is the single delivery path now. No production code read
the environment variable; only the agent skill told the agent to
check it. Every writer of the variable also rendered the prompt from
the same normalized payload, so no content is lost.

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-09-10 19:47:40 -07:00
..
adr feat(runner): add offline evaluation tooling (#12653) 2026-09-01 05:19:47 -05:00
design refactor: remove the wake payload from the launch environment 2026-09-10 19:47:40 -07:00
images/evalbook-chat fix(evals): make the chat viewer the default published Evalbook (#12952) 2026-09-07 07:47:17 -05:00
research
tutorials
adding-a-harness.md
architecture.md feat(runner): add offline evaluation tooling (#12653) 2026-09-01 05:19:47 -05:00
capability-authorization-and-exposure.md
capability-clean-room-chat.md
capability-contract.md feat: add experimental native chat connectors (#13038) 2026-09-10 10:06:45 -05:00
capability-disposition.md
capability-eval-conformance.md
capability-eval-slice.md feat(runner): add offline evaluation tooling (#12653) 2026-09-01 05:19:47 -05:00
capability-execution-modes.md
capability-future-binding-boundary.md
capability-issue-thread-ui.md
capability-live-runnerd-codex.md
capability-mock-control-plane-port.md
capability-scenario-explorer.md
capability-semantic-catalog.md
capability-semantic-tools.md
capability-verification-commands.md
codex-driver.md fix(runner): restore local session and task integrity (#12721) 2026-09-02 16:11:26 -05:00
durable-recovery.md fix(runner): preserve durable native session authority across recovery (#13092) 2026-09-09 11:02:59 -05:00
evals-integration.md feat(runner): add offline evaluation tooling (#12653) 2026-09-01 05:19:47 -05:00
index.md feat(runner): add offline evaluation tooling (#12653) 2026-09-01 05:19:47 -05:00
live-console-protocol-server.md
live-console.md
local-runner.md
protocol-compatibility.md
runner-protocol-live-evals.md fix(evals): make the chat viewer the default published Evalbook (#12952) 2026-09-07 07:47:17 -05:00
runner-workflow-evals.md feat(runner): restore direct live eval campaigns and reports (#12909) 2026-09-05 20:18:11 -05:00
scenario-chat.md
sdk.md
standalone-thin-paperclip-adapter.md