{"id":49720210,"url":"https://github.com/selfagency/agentsy","last_synced_at":"2026-05-09T00:09:06.506Z","repository":{"id":343769351,"uuid":"1179086438","full_name":"selfagency/agentsy","owner":"selfagency","description":"🤖 WIP agentic development framework","archived":false,"fork":false,"pushed_at":"2026-05-03T23:47:14.000Z","size":1181,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-05-04T00:23:32.887Z","etag":null,"topics":["agentic","agentic-ai","agentic-framework","agentic-workflow","agents","ai","anthropic","chatbot","chatgpt","json","llm","mistral","models","ollama","openai","parser","xml"],"latest_commit_sha":null,"homepage":"http://llmstreamparser.self.agency/","language":"TypeScript","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/selfagency.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":null,"funding":null,"license":"LICENSE.md","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-03-11T17:13:58.000Z","updated_at":"2026-05-03T23:47:17.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/selfagency/agentsy","commit_stats":null,"previous_names":["selfagency/llm-stream-parser","selfagency/agentsy"],"tags_count":9,"template":false,"template_full_name":null,"purl":"pkg:github/selfagency/agentsy","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/selfagency%2Fagentsy","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/selfagency%2Fagentsy/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/selfagency%2Fagentsy/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/selfagency%2Fagentsy/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/selfagency","download_url":"https://codeload.github.com/selfagency/agentsy/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/selfagency%2Fagentsy/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":32661193,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-05T11:29:49.557Z","status":"ssl_error","status_checked_at":"2026-05-05T11:29:48.587Z","response_time":54,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.5:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agentic","agentic-ai","agentic-framework","agentic-workflow","agents","ai","anthropic","chatbot","chatgpt","json","llm","mistral","models","ollama","openai","parser","xml"],"created_at":"2026-05-09T00:09:05.785Z","updated_at":"2026-05-09T00:09:06.494Z","avatar_url":"https://github.com/selfagency.png","language":"TypeScript","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Agentsy\n\nComposable infrastructure for LLM systems.\n\nBuild headless agentic workflows in Node.js-compatible runtimes, plug in only the layers you need, and stay aligned with open standards like MCP and AG-UI instead of a captive vendor stack.\n\n[![Tests](https://github.com/selfagency/agentsy/actions/workflows/tests.yml/badge.svg)](https://github.com/selfagency/agentsy/actions/workflows/tests.yml) [![codecov](https://codecov.io/gh/selfagency/agentsy/graph/badge.svg?token=4U6b4yU5Ln)](https://codecov.io/gh/selfagency/agentsy) [![Codacy Badge](https://app.codacy.com/project/badge/Grade/be00077d20c54f9097c7f38bf575603f)](https://app.codacy.com/gh/selfagency/agentsy/dashboard?utm_source=gh\u0026utm_medium=referral\u0026utm_content=\u0026utm_campaign=Badge_grade) [![Quality Gate Status](https://sonarcloud.io/api/project_badges/measure?project=selfagency_llm-stream-parser\u0026metric=alert_status)](https://sonarcloud.io/summary/new_code?id=selfagency_llm-stream-parser) [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE.md)\n\n## Why choose Agentsy\n\n- **Streaming-first by default**: handle chunk boundaries, partial outputs, tool-call deltas, and finish states as first-class concerns.\n- **Headless-first architecture**: ship robust automation, CLI tooling, and agent workflows before committing to a UI surface.\n- **Composable package ecosystem**: adopt one package or a full pipeline — no all-or-nothing framework tax.\n- **Type-safe and defensive**: strict TypeScript contracts and recoverable parsing behavior for untrusted model output.\n- **Open ecosystem alignment**: practical interoperability with standards and protocols such as MCP and AG-UI.\n\n## Already used in production\n\nAgentsy already powers third-party model support in VS Code extensions for GitHub Copilot Chat:\n\n- [Opilot](https://marketplace.visualstudio.com/items?itemName=selfagency.opilot) — Ollama models with local workflows, tool calling, vision, and streaming\n- [Z.ai for Copilot](https://marketplace.visualstudio.com/items?itemName=selfagency.z-models-vscode) — Z.ai coding models with streaming, tool calling, and MCP-assisted capabilities\n- [Mistral for Copilot](https://marketplace.visualstudio.com/items?itemName=selfagency.mistral-models-vscode) — Mistral AI models with streaming, tool calling, and vision support\n\n## Example workflows\n\n- [CLI log summarizer (easy)](./docs/examples/cli-log-summarizer.md) — stream model output into terminal-ready summaries with minimal wiring.\n- [DNS blocklist workflow](./docs/examples/dns-blocklist.md) — ingest multiple logs, stream model output, validate schema, and update remote DNS when a source IP should be blocked.\n- [Multi-provider policy gate](./docs/examples/multi-provider-policy-gate.md) — switch providers while keeping one schema-gated automation contract.\n- [Agent tool loop with retries + continuation](./docs/examples/tool-loop-retries-continuation.md) — run a multi-step tool loop, retry transient failures, and resume interrupted streams.\n- [Stateful ops copilot backend](./docs/examples/stateful-ops-copilot.md) — project stream events into state and AG-UI clients with continuation recovery.\n- [All-tooling end-to-end workflow](./docs/examples/all-tooling-end-to-end.md) — combine stream ingestion, normalization, processing, tool loops, recovery, state projection, AG-UI conversion, rendering, and action gating in one flow.\n\n## Package ecosystem\n\nThis repository is a **pnpm workspace monorepo** orchestrated with **Turborepo** and organized as focused `@agentsy/*` packages.\n\n### Integration\n\n| Package                                        | Role                                                                      | Status    |\n| ---------------------------------------------- | ------------------------------------------------------------------------- | --------- |\n| [`@agentsy/vscode`](./docs/packages/vscode.md) | VS Code chat-provider, rendering, settings, and MCP integration utilities | Published |\n\n### Runtime and orchestration\n\n| Package                                                  | Role                                                             | Status    |\n| -------------------------------------------------------- | ---------------------------------------------------------------- | --------- |\n| [`@agentsy/processor`](./docs/packages/processor.md)     | Event-driven stream orchestrator and transform pipeline          | Published |\n| [`@agentsy/normalizers`](./docs/packages/normalizers.md) | Provider-specific normalization into a common stream shape       | Published |\n| [`@agentsy/agent`](./docs/packages/agent.md)             | Multi-step agent loop and stop-condition helpers                 | Published |\n| [`@agentsy/adapters`](./docs/packages/adapters.md)       | Integration-oriented wrappers around stream-processing pipelines | Published |\n| [`@agentsy/renderers`](./docs/packages/renderers.md)     | Plain-text rendering plus shared renderer primitives             | Published |\n\n### Core parsing and shaping utilities\n\n| Package                                                | Role                                                       | Status    |\n| ------------------------------------------------------ | ---------------------------------------------------------- | --------- |\n| [`@agentsy/thinking`](./docs/packages/thinking.md)     | Incremental `\u003cthink\u003e` content extraction                   | Published |\n| [`@agentsy/tool-calls`](./docs/packages/tool-calls.md) | XML and native tool-call helpers                           | Published |\n| [`@agentsy/structured`](./docs/packages/structured.md) | JSON parsing, repair, validation, and streaming helpers    | Published |\n| [`@agentsy/context`](./docs/packages/context.md)       | Context extraction, dedupe, and normalization helpers      | Published |\n| [`@agentsy/formatting`](./docs/packages/formatting.md) | Display-focused text sanitization and formatting utilities | Published |\n| [`@agentsy/recovery`](./docs/packages/recovery.md)     | Recovery snapshots and continuation prompt helpers         | Published |\n| [`@agentsy/xml-filter`](./docs/packages/xml-filter.md) | XML tag filtering and privacy-oriented scrubbing utilities | Published |\n| [`@agentsy/sse`](./docs/packages/sse.md)               | Server-sent-event parsing utilities                        | Published |\n| [`@agentsy/types`](./docs/packages/types.md)           | Shared type contracts across the ecosystem                 | Published |\n\n### State and protocol\n\n| Package                                      | Role                                                   | Status    |\n| -------------------------------------------- | ------------------------------------------------------ | --------- |\n| [`@agentsy/ui`](./docs/packages/ui.md)       | Conversation-state store and processor binding helpers | Published |\n| [`@agentsy/ag-ui`](./docs/packages/ag-ui.md) | AG-UI protocol bridge utilities                        | Published |\n\n### Private package\n\n| Package                                                  | Role                                   | Status  |\n| -------------------------------------------------------- | -------------------------------------- | ------- |\n| [`@agentsy/integration`](./docs/packages/integration.md) | Cross-package integration test harness | Private |\n\n## Pick your adoption path\n\n- **I need robust Node streaming pipelines** → start with [`@agentsy/normalizers`](./docs/packages/normalizers.md) + [`@agentsy/processor`](./docs/packages/processor.md)\n- **I need JSON schema gates before automation** → add [`@agentsy/structured`](./docs/packages/structured.md)\n- **I need multi-step agent behavior** → add [`@agentsy/agent`](./docs/packages/agent.md)\n- **I need VS Code integration utilities** → add [`@agentsy/vscode`](./docs/packages/vscode.md)\n- **I need protocol/state helpers** → add [`@agentsy/ag-ui`](./docs/packages/ag-ui.md), [`@agentsy/ui`](./docs/packages/ui.md), and focused utility packages\n\nIf you prefer a thinner integration layer, start with `@agentsy/normalizers` + `@agentsy/processor`, then add `@agentsy/structured` once you want strict schema gates before automated actions.\n\n## Documentation map\n\n- [Documentation home](./docs/index.md)\n- [Getting started](./docs/getting-started.md)\n- [Why Agentsy](./docs/why-agentsy.md)\n- [Examples](./docs/examples/index.md)\n- [API index](./docs/api.md)\n- [Package catalog](./docs/packages.md)\n- [Roadmap](./docs/roadmap.md)\n- [Developer guide](./docs/developers/index.md)\n\nDeep architecture material lives under [`docs/architecture/`](./docs/architecture/index.md), while contributor and automation specifics remain in the developer docs and repository instructions.\n\n### Monorepo commands\n\n```bash\npnpm install\npnpm build\npnpm check-types\npnpm test\n```\n\n## License\n\n[MIT](LICENSE.md)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fselfagency%2Fagentsy","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fselfagency%2Fagentsy","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fselfagency%2Fagentsy/lists"}