Repository navigation
Open
Conversation
Ontogeny (an Apache HugeGraph subproject) models an enterprise's nouns (objects, properties, links) and verbs (actions, functions, permissions) as declarative YAML in Git, compiled into strongly-typed object tables and one unified API shared by apps, analytics and AI agents. Writes go through a five-stage Action runtime (params -> rules -> Cedar policy -> effects -> immutable audit) with side effects on a transactional outbox; an RSI loop improves the model itself inside the governance track (the LLM proposes, deterministic evals select, promotion follows T0-T3 tiers). Docs: docs/README.md (中文) · docs/en/README.md (English) Merge plan with hugegraph-ai: docs/hugegraph-ai-merge-plan.md
- bilingual docs split into docs/zh/ and docs/en/ with language-matched screenshots; etymology of the name told in both READMEs - console defaults to English (saved preference still wins) - chore: untrack .zcodeignore (local tooling config)
- four screenshots paired with the sections they illustrate: the Overview architecture board (intro), the ontology canvas (Git-native modeling), the dashboard's ontology-model section (quick start) and the audit stream plus the agent-console popup (MCP/governance) - section headings stripped of emoji; badge anchors fixed - documentation section now points at both docs/zh and docs/en
Merge upstream main (9922568) into third_party/ preserving upstream history and authorship; zero code changes, zero import coupling, zero dependency wiring -- per docs/hugegraph-ai-merge-plan.md §1. Ontogeny-owned root files are kept as-is.
- THIRD_PARTY.md registers source, merged point (main@9922568) and the sync policy (no direct edits under third_party/) - NOTICE gains the upstream attribution line - ruff extend-exclude += third_party (upstream code is not held to our lint rules; uv.lock carries zero new dependencies) - third-party-smoke workflow: presence + LICENSE + zero import coupling
Deleted as root-only machinery: .asf.yaml, .github/ (issue templates + upstream CI), .gitattributes (source-packaging rules), .licenserc.yaml, .pre-commit-config.yaml. Upstream .gitignore entries unique to the vendored tree merged into the root .gitignore, scoped to third_party/. Content files (LICENSE, NOTICE, README, workspace pyproject.toml) and all code stay byte-for-byte.
Contributor
There was a problem hiding this comment.
🟡 Changes recommended
Action-state bugs, inaccurate function contracts, altered project attribution, and removed compliance gates must be resolved.
10 open findings
What changed in this PR
Introduces the Ontogeny operational-ontology platform, its React console and manufacturing example, while vendoring HugeGraph AI components for future integration.
Changes:
- Adds Ontogeny’s governed ontology, action, policy, agent, projection, and evolution layers.
- Adds the web console, extensions, example domain, tests, and bilingual documentation.
- Vendors HugeGraph AI under
third_party/with provenance and smoke checks.
| File | Description |
|---|---|
server/ontogeny/** |
Core platform services and public package APIs. |
extensions/** |
Ollama, HugeGraph, and built-in agent adapters. |
domains/product-manufacturing/** |
Manufacturing ontology, actions, policies, functions, and agents. |
web/src/** |
React console entry points, components, and agent styling. |
web/*.json, web/*.ts, web/index.html |
Frontend build, type-check, and integration-test configuration. |
web/README.md, web/.gitignore, web/.oxlintrc.json |
Frontend guidance and tooling configuration. |
docs/en/**, docs/zh/** |
Bilingual quick-start documentation. |
third_party/** |
Vendored HugeGraph AI source, tests, deployment assets, and documentation. |
THIRD_PARTY.md |
Vendored-source provenance and synchronization policy. |
NOTICE, LICENSE |
Root licensing and attribution changes. |
.github/workflows/third-party-smoke.yml |
Vendored-tree presence and coupling checks. |
.github/workflows/check-dependencies.yml |
Removes dependency and license checks. |
.github/workflows/sync.yml, .github/workflows/labeler.yml, .github/workflows/auto-pr-comment.yml |
Removes existing repository automation. |
.github/pull.yml, .github/labeler.yml, .github/ISSUE_TEMPLATE/config.yml |
Removes repository maintenance configuration. |
.pre-commit-config.yaml, .gitattributes |
Removes local validation and release-export rules. |
.dockerignore |
Defines the Ontogeny container build context. |
🧠 Review effort: Balanced
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
- complete-operation refuses CANCELLED operations; assign-operation is QUEUED-only; record-inspection rejects negative defect counts; add-bom-line constrains scrap_rate to [0,1] (null defaults to 0) - new contract test covers all four hardened rules end to end - core/__init__ imports lint before exporting it via __all__ - tsconfig.node covers vitest.integration.config + tests-integration; the e2e helper's json typing fixed accordingly - upstream NOTICE merged into the root NOTICE (attribution verbatim), vendored copy removed; THIRD_PARTY.md curation record updated Function returns stay declared as string: the DSL type vocabulary has no union/object shape, and the three functions return success/failure unions that a struct declaration would misrepresent
'draws one directed, selectable edge per link type' touched __ooG6 after waiting only for the flow container — the fake canvas registers a few ticks later once the domain loads, so the CI runner read .cfg off undefined. Same guard as clickNode, scoped to the edge under test.
Member
Author
Resolution per the standing strategy:
- .github/workflows/{hugegraph-llm,hugegraph-python-client,ruff}.yml:
take main's 1.8.0 versions — upstream release CI stays functional for
the root modules after the eventual ontogeny -> main merge
- pyproject.toml: keep ontogeny's manifest (M0 zero-wiring rule); the
upstream 1.8.0 workspace changes stay recorded in the subtree copy
- rename detection carried main's deltas on the nine touched files into
their third_party/ copies, so the vendored snapshot picks up the
release fixes for exactly the files upstream touched
Verified: zero import coupling unchanged; merging this branch into main
is a fast-forward (7729a1e is an ancestor) — conflict-free by construction.
hugegraph-llm.yml, hugegraph-python-client.yml and ruff.yml came back at the root via the 1.8.0 conflict resolution, but they build/test upstream modules that live under third_party/ on this branch — they can never pass here and stay red on every push. Removed per the THIRD_PARTY.md curation principle; restoring them adapted to the merged layout belongs to merge-plan steps 2-3.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Purpose of the PR
Introduce Ontogeny — a self-improving operational ontology platform — as proposed and discussed in #376. This PR lands Step 1 (M0, structure merge) of the three-step merge plan; Steps 2–3 (GraphRAG functional merge, ML/graph-compute deep merge) will follow as normal feature PRs.
What Ontogeny is, in one line: it models an enterprise's business nouns (objects, properties, links) and verbs (actions, functions, policies) as declarative, Git-native YAML, compiles them into strongly-typed tables and a HugeGraph graph projection, and serves them to applications, analytics and AI agents through one governed API (five-stage Action runtime, Cedar-subset policy with default-deny + markings, immutable audit, transactional outbox, and a governed RSI self-evolution loop).
Main Changes
server/ontogeny/— the Python kernel: 11-kind DSL + validators, registry & compiled snapshots, sync engine, five-stage action runtime, Cedar-subset policy engine, query engine, sandboxed functions, RSI loop (telemetry + evolve), graph projection, agent session governance (MCP surface at/mcp), extension hostextensions/— pluggable implementations: llm-ollama (LLM gateway), graph-hugegraph (projection store), agent-builtin-llm (reference agent engine)web/— React 19 + TypeScript console (bilingual EN/zh-CN)domains/product-manufacturing/— showcase domain package (also the test golden package)docs/zh/+docs/en/— bilingual architecture & usage docs with UI screenshots; merge plan atdocs/hugegraph-ai-merge-plan.mdmain(9922568) merged via subtree intothird_party/, preserving upstream history and authorship — zero code changes, zero import coupling, zero dependency wiring; repo-root machinery files that only function at a repository root (.asf.yaml,.github/,.gitattributes,.licenserc.yaml,.pre-commit-config.yaml) were pruned from the vendored copy and its unique.gitignoreentries were merged (scoped) into the root.gitignore; the full curation record is inTHIRD_PARTY.mdTHIRD_PARTY.md(source, merged point, sync policy), upstream attribution line appended toNOTICE, vendoredLICENSEkept underthird_party/third_partyadded to ruffextend-exclude; newthird-party-smokeCI workflow (presence + LICENSE + zero import coupling); no changes topyproject.tomldependencies oruv.lockMerge-plan status & what is NOT here yet
[graphai]extra, GraphRAG Q&A (ask_graph), text-to-ontology proposalsGround rules held across all steps: no second governance channel (LLM calls go through the platform gateway; writes go through the Action runtime), anti-corruption layer (
server/never importsthird_party/packages), graph queries return IDs assembled back through the engine so markings always apply.Verifies this pull request
pytestgreen (456 tests); frontendvitestgreen (241) + build; E2E 8/8git grep -n "hugegraph_llm|hugegraph_ml" -- server extensions tests→ zero hitsuv.lockcarries zero new dependencies (diff empty vsmain)git merge-treedry-run: mergingontogenyintomainis conflict-freeTHIRD_PARTY.md/NOTICE/ vendoredLICENSEin place;third-party-smokeworkflow greenRelated to #376