Hacktoberfest 2026: as issues que os mantenedores marcaram para outubro, abertas e boas para iniciantes. Ver issues do Hacktoberfest

[rig-summarizer] Daily file summary — 2026-10-13

Fechada
  • Algo errado?
#638 0 comentários 0 reações 0 responsáveis Ver no GitHub

Mantenedores costumam responder em até 1 dia

Ninguém assumiu esta issue ainda.

Avaliação

Dificuldade
3/5
Tempo estimado
1-3 horas
Facilidade para iniciantes
8/100
Tipo de issue
Documentação
Clareza
Precisa de esclarecimento
Status de atividade
Ativa
Stack de tecnologia
github-actions, typescript
Domínio
ai-infra-agents, devtools

Direção de pesquisa

This is an automated daily file summary with no requested change, so there is nothing to implement. Start by checking whether any listed file, such as .agents/skills/rig/rig.ts or .github/fixtures/assert-three-judges.ts, has a summary that is wrong or stale. Done would mean a maintainer confirms the summary needs a specific correction, which should be asked for in the thread first.

Escrita pelo modelo de indexação a partir do texto da issue.

Descrição

ai-agent automation
Global Summary

This project (Rig) is a minimal TypeScript multi-agent harness centered on skills/rig/rig.ts, which defines schemas, prompt intents, agent/workflow abstractions, and a Copilot SDK runtime. Additional pluggable engines (Anthropic, Codex, DeepSeek, Gemini, Pi) adapt third-party AI SDKs or CLIs to Rig's AgentFactory interface, sharing lifecycle/tool-schema utilities. The .github directory contains CI fixtures and drivers that exercise a 'three-judges' evaluation workflow across these engines inside sandboxed/dockerized environments, with assertion helpers validating judgment output. Sample and test files (.tmp-agent, scripts) demonstrate real usage patterns, such as a file-summarization multi-agent pipeline and integration tests spawning the launcher CLI against sample agent programs.

Individual File Summaries
File Summary
./.agents/skills/rig/engines/anthropic.ts Implements an AgentFactory wrapping the Anthropic SDK beta client, converting Rig tools to Anthropic tool schemas and managing ask/close lifecycle with a single-turn guard.
./.agents/skills/rig/engines/codex.ts Implements an AgentFactory around the OpenAI Codex SDK thread API; does not support Rig tools and configures developer instructions from the system message.
./.agents/skills/rig/engines/deepseek.ts Implements an AgentFactory using the DeepSeek harness SDK session API, without Rig tool support, handling session close and turn lifecycle.
./.agents/skills/rig/engines/gemini.ts Implements an AgentFactory that shells out to the Gemini CLI via child_process, parsing JSON stdout responses and tracking a session id across turns.
./.agents/skills/rig/engines/pi.ts Implements an AgentFactory bridging to the pi-agent-core/pi-ai SDKs, resolving models by provider and converting Rig tools to Pi tool definitions.
./.agents/skills/rig/engines/utils.ts Shared engine helpers: agentLifecycle manages single-turn concurrency/abort semantics, plus objectToolSchema and toolResultText utilities for tool schema/result handling.
./.agents/skills/rig/globals.ts Ambient workflow context helpers (call, pipeline, parallel) that delegate to the currentWorkflow() context, easing porting of Claude-style dynamic workflows.
./.agents/skills/rig/launch-tool.ts Creates a 'run_rig' Copilot SDK tool that spawns a trusted Rig TypeScript program as a subprocess, scrubbing connection credentials from its environment.
./.agents/skills/rig/rig.ts Core Rig runtime (~3500 lines): defines Json/Schema types, AgentSpec/AgentFn, prompt intents, workflow orchestration (call/pipeline/parallel), and the Copilot SDK-based engine and launcher CLI.
./.agents/skills/rig/run.ts Thin CLI entrypoint that imports and invokes runLauncherCli from rig.ts when executed directly.
./.github/drivers/codex-fixture-mcp.ts Builds a one-shot fixture launcher that spawns a sandboxed Codex process via docker/chroot and validates output against the three-judges assertion helper.
./.github/drivers/copilot-sdk-driver.ts Driver script that loads action helper modules, parses multi-provider/tool config from environment, extracts a Rig fixture source from a prompt file, and runs it against the Copilot SDK.
./.github/drivers/pi-rig-extension.ts Creates a one-shot Pi agent tool ('run_rig') that executes the checked-in Pi three-judge Rig fixture and validates its output.
./.github/fixtures/assert-three-judges.ts Test assertion helper validating the shape and content of three-judge workflow results (engine, model, judgments, verdict) per engine.
./.github/fixtures/pi-gateway.ts Validates and extracts a harness-provisioned Pi 'aw-gateway' provider configuration (baseUrl, api, model, limits) from an arbitrary JSON value.
./.github/fixtures/provider-fixture.ts Selects and configures a provider engine (codex, gemini, deepseek, etc.) from environment variables to build the three-judges workflow for integration testing.
./.github/fixtures/provider-three-judges.ts Tiny entrypoint exporting the default workflow instance produced by createProviderFixture().
./.github/fixtures/three-judges.ts Defines the three-judge dummy-request workflow: a judge agent evaluating clarity/safety/feasibility criteria orchestrated via Rig's workflow/call API.
./scripts/haiku.integration.test.ts Vitest integration test that spawns the Rig launcher against haiku/sonnet sample programs over stdio or SDK-URI transport, skipped unless Copilot credentials are present.

Generated by Daily Rig File Summarizer · copilot · auto · 70.5 AIC · ⌖ 7.07 AIC · ⊞ 7.5K · ◷

  • expires on Oct 20, 2026, 2:45 PM UTC
Linguagem predominante
TypeScript
Estrelas
14
Forks
0
Merge médio
3h 38min
PRs com merge (30d)
29

Preparar o ambiente

Este projeto não oferece contêiner de desenvolvimento, Dockerfile nem guia de contribuição, então a configuração fica por sua conta: comece pelo README e veja nosso guia da primeira contribuição para os passos gerais.

Primeiros passos

  1. Leia a issue inteira e depois o guia de contribuição do projeto.
  2. Comente na issue dizendo que vai assumir — evita que duas pessoas façam o mesmo trabalho.
  3. Faça um fork do repositório e trabalhe em uma branch.
  4. Abra um pull request que referencie o número da issue.

Mais de githubnext/rig

Todas as issues de githubnext/rig

Issues semelhantes

Mais issues de TypeScript

Receba novas issues na sua caixa de entrada

Um resumo curto de issues do GitHub para quem está começando.