Metadata-Version: 2.4
Name: dolphinos
Version: 0.1.4
Summary: DolphinQuant shared agent infrastructure — runtime, memory, skills, prompts, retrieval, reflector, tools, messaging. Consumed by qorka, navika, and DigitalEmployees.
Author: DolphinQuant
Classifier: Private :: Do Not Upload
Requires-Python: <3.13,>=3.11
Requires-Dist: httpx>=0.27.0
Requires-Dist: jinja2>=3.1.0
Requires-Dist: openai-agents==0.19.4
Requires-Dist: openai==2.53.0
Requires-Dist: prompt-toolkit
Requires-Dist: pydantic>=2.0.0
Requires-Dist: rich>=13.0.0
Requires-Dist: structlog>=23.0.0
Provides-Extra: feishu
Requires-Dist: lark-oapi>=1.4.0; extra == 'feishu'
Provides-Extra: util
Requires-Dist: numpy>=1.24.0; extra == 'util'
Requires-Dist: pandas>=2.0.0; extra == 'util'
Requires-Dist: pillow>=10.0.0; extra == 'util'
Requires-Dist: scipy>=1.10.0; extra == 'util'
Description-Content-Type: text/markdown

# dolphinos

**DolphinQuant shared agent infrastructure.**

`dolphinos` is the foundation layer that DolphinQuant AI agents are built on. It provides agent SDK runtime, prompt engine, and (in later phases) memory primitives, skills system, retrieval, reflector base, tool registry, and messaging adapters. Each consumer initializes dolphinos primitives with its own role, schema, and prompts.

**Status:** Phase 1A in flight (2026-05-15). Phase 0.5 (qorka-side prep sprint) shipped 2026-05-15. See [`docs/3_roadmap/00_overview.md`](docs/3_roadmap/00_overview.md) for sequencing.

## Consumers

- **qorka** — AI R&D pipeline (per-session strategy generation, refinement, evaluation)
- **navika** — live-feedback reflection agent (periodic monitoring, briefing, qorka-trigger)
- **DigitalEmployees** — non-quant agents (growth_marketer + future digital employees)

## Subpackages

Status legend: **POPULATED** (real code shipped) · **DRAFT** (design only, may be reshaped by Phase 3.0) · **DELETED** (removed from planned API).

| Subpackage | Status | Scope |
|---|---|---|
| `dolphinos.runtime` | POPULATED (Phase 1A) | Agent SDK wrappers, generic `RuntimeContext`, sandboxed tools, MCP clients, terminal renderer, transcript logger, retry/cancellation/loop-detection primitives |
| `dolphinos.prompts` | POPULATED (Phase 1A, partial) | Jinja env + `PROMPTS_ROOT` resolver + filters. Prompt-factory + critic-gate framework still DRAFT. |
| `dolphinos._util` | POPULATED (Phase 1A, private) | Internal helpers: JSON serialization, image utilities, statistical metrics |
| `dolphinos.memory` | DRAFT | sqlite-vec + FTS5 store with pluggable Schema per consumer. Will be redesigned in Phase 3.0. |
| `dolphinos.skills` | DRAFT | Skill loader, promoter, citation tracking — workspace-pluggable via SkillRoot |
| `dolphinos.retrieval` | DRAFT | Hybrid vector + lexical retrieval; embedding model wrapper |
| `dolphinos.reflector` | DRAFT | Reflector base class (subclassed by each consumer's reflection cycle) |
| `dolphinos.tools` | DRAFT | Tool registry, capability scopes, sandbox primitives |
| `dolphinos.messaging` | DRAFT | WeChat / email / Slack adapters with templating + delivery state |
| `dolphinos.graph` | DELETED in Phase 2 | LangGraph utilities — retired alongside qorka's LangGraph orchestration |

## Documentation

- [`docs/README.md`](docs/README.md) — five-layer docs index
- [`docs/3_roadmap/00_overview.md`](docs/3_roadmap/00_overview.md) — phase sequencing (Phase 0.5 → 1 → 2 → 3)
- [`docs/2_design/00_system_overview.md`](docs/2_design/00_system_overview.md) — subpackage map + cross-cutting concerns
- [`docs/2_design/decisions_log.md`](docs/2_design/decisions_log.md) — append-only architectural decisions

## Installing as a consumer

Private git+ssh dependency. Consumers (qorka, navika, DigitalEmployees) install via:

```toml
# pyproject.toml
dependencies = [
    "dolphinos @ git+ssh://git@github.com/DolphinQuant/dolphinos.git@v0.1.0",
]
```

For qorka (which uses `requirements.txt`):

```
dolphinos @ git+ssh://git@github.com/DolphinQuant/dolphinos.git@v0.1.0
```

**SSH deploy-key setup** for CI environments: each consumer's CI needs an SSH key with read access to this repo. Add the public key as a deploy key on `github.com/DolphinQuant/dolphinos` (Settings → Deploy keys). No PyPI publication is planned until Wave 4 D3 (commercial decision).

## Distribution

Private — DolphinQuant proprietary. Not for redistribution.

## License

Private — see `LICENSE` (TBD).
