codenib 0.2.2__tar.gz → 0.2.4__tar.gz

{codenib-0.2.2 → codenib-0.2.4}/CHANGELOG.md

@@ -11,6 +11,126 @@ All notable user-facing changes are recorded here. CodeNib follows
 
 ## [Unreleased]
 
+## [0.2.4] - 2026-09-26
+
+### Added
+
+- Source-only model-planned grep followed by Jev reranking, using one
+  user-owned OpenRouter credential from CLI or MCP. The optional `grep` extra
+  bundles ripgrep; the route does not require embeddings, a GPU or an index.
+- OpenRouter S256 PKCE login and credential management, with supported OS
+  keyrings and an explicit private POSIX file fallback. `codenib init`
+  registers the source-only route through native Claude Code and Codex CLIs.
+- Static export of precomputed story Wikis with source citations, page maps,
+  architecture visuals and CodeNib backlinks. Public browsing needs no model
+  calls or live Wiki backend.
+- Optional browser OpenRouter trials on explicitly enabled static public Wikis.
+  The browser owns authorization and model requests; a separate bounded CPU
+  service returns source-checked grep candidates without receiving a model key.
+  Default static browsing still makes no model calls. Actual provider consent
+  and production-host acceptance remain preview deployment gates.
+- Jev reranking through OpenRouter's Decisions API, selectable with
+  `rerank_strategy="decisions"`. The typed client supports Noul, Choice, and
+  Score questions; reranking uses bounded batches and normalized relevance
+  scores while preserving first-stage candidates when a request fails.
+
+### Changed
+
+- GitHub Pages previews default to the `fast` preset, avoiding embedding
+  downloads. Generated story Wikis remain a separate, explicit generation and
+  cached-export path.
+- Present source-linked agent setup, a recorded agent session, retrieval
+  evidence, language capabilities and sourced tool comparisons prominently
+  in the README and homepage.
+- Bind the retrieve/rerank pipeline's BM25 source reads to its repository root
+  so rerankers receive candidate code when invoked from another directory.
+- Bound each built-in Wiki generation-lock acquisition to a 30-second wait. A
+  timed-out waiter does not interrupt the owner or start duplicate generation,
+  and a later request can reuse the owner's published result.
+- Wiki cache audits and prewarm dry runs now copy an existing, quiescent
+  `wiki.sqlite3` into a private temporary snapshot and inspect that snapshot in
+  immutable read-only mode. They create no lock or SQLite sidecar beside the
+  source database, do not initialize or change its schema or lock directory,
+  and reject existing transaction sidecars or a source change while the
+  snapshot is captured.
+
+### Fixed
+
+- Pass the installed package version explicitly to MCP initialization so
+  `serverInfo.version` is populated with MCP Python SDK 2.2.0 (#779, #780).
+- Support bounded multiline grep plans and Windows newline handling. Preserve
+  Windows source authority when unrelated sibling directories change while
+  retaining repository-root and ancestor-identity checks.
+- Validate Wiki graph references separately from source paths, persist evidence
+  as strict JSON and limit evidence ranges to the returned source lines.
+
+### Removed
+
+- Retired the one-release `agentwiki_*.json` compatibility path tracked by
+  #749. AgentWiki no longer discovers, reads, writes, adopts, locks, or audits
+  JSON cache files; persistence requires an injected `WikiStore`, while
+  store-less callers remain in memory. Existing `wiki.sqlite3` entries and
+  legacy files are not deleted. The Wiki cache audit report is now schema v2
+  and reports only database entry and payload-byte accounting.
+
+## [0.2.3] - 2026-09-03
+
+### Added
+
+- A Wiki-only `codenib.storage` facade exposing the domain-local `WikiStore`
+  harness and SQLite WAL implementation, with versioned schema initialization,
+  bounded canonical JSON, integrity checks, atomic publication, and
+  cross-process generation guards.
+
+### Changed
+
+- AgentWiki, the web service, cache audit, prewarming, and Wiki benchmarking
+  now use `wiki_cache/wiki.sqlite3`. Eligible source-compatible JSON entries
+  remain available through read-through compatibility and are not rewritten or
+  deleted.
+- The supported product database boundary is now
+  `codenib.storage.WikiStore`, with `SQLiteWikiStore` as its supported adapter.
+  `RepoManifest`, BM25, FAISS, igraph, and portable context payloads remain file
+  artifacts. Existing compiler caches use `artifact pack`; portable artifacts
+  continue through `artifact verify` and `mcp --artifact`.
+- The v0.2.3 release intentionally narrows the former storage API and CLI.
+  Users who still need data from a v0.2.2 SQLite catalog must run `artifact
+  materialize` with a pinned v0.2.2 environment before upgrading; the resulting
+  portable artifact remains supported.
+
+### Removed
+
+- The optional Web `index_storage` configuration and its catalog-backed local
+  runtime, retained BM25 activation, reconciliation, durable index-job
+  orchestration, `/api/index-jobs` write/read routes, and index-status job or
+  update-capability fields. Index status remains available as a read-only
+  `RepoManifest` projection.
+- The unreleased catalog-selected `codenib mcp` cold-start mode. Portable
+  artifacts remain served through `codenib mcp --artifact [--repo]`.
+- The unreleased `storage audit` command and `index --publish-retained` path.
+  Those two batches removed 1,407 production lines and 1,742 dedicated test
+  lines, including the audit execution path, eight index-only options, and
+  three compiler exports; ordinary indexing remained unchanged in that batch.
+- The post-v0.2.2 retained-storage benchmark gate, its manual Make target, and
+  noncanonical readiness receipt. The report-only experiment had no product or
+  CI workflow caller and could never report `promotion_eligible=true`; removing
+  it deleted a 4,913-line harness, 3,468 test lines, 367 evidence lines, 148
+  documentation lines, eight Make variables, and one manual target.
+- The unreleased `codenib jobs` execution stack and catalog schema v5 through
+  v8. This removed eight production modules (9,615 lines), twelve dedicated
+  test modules (22,170 lines), 73 lazy exports, two interruptible CAS methods,
+  one command with two execution paths, and four schema migrations; changes
+  inside retained files bring the batch total to 21,214 production and 29,352
+  test lines removed.
+- The generic retained-storage surface formerly published by v0.2.2 through
+  `codenib.storage`: its schema-v4 SQLite catalog and local CAS, `artifact
+  import-cache`, `artifact materialize`, and the retained compiler bridge. The
+  name `codenib.storage` now resolves only to the Wiki database facade. Normal
+  index, Wiki, portable-artifact, publish, and MCP paths are unchanged.
+- The catalog-backed clangd FactBatch publication/reuse experiment and its
+  profiling target. Provider-neutral FactBatch models, resolvers, native buffer
+  transport, and the promoted clangd query index remain independent of storage.
+
 ## [0.2.2] - 2026-08-21
 
 ### Added
@@ -193,7 +313,9 @@ All notable user-facing changes are recorded here. CodeNib follows
 - Prepared secretless PyPI publishing through a dedicated GitHub Actions OIDC
   workflow.
 
-[Unreleased]: https://github.com/sysevol-ai/CodeNib/compare/v0.2.2...HEAD
+[Unreleased]: https://github.com/sysevol-ai/CodeNib/compare/v0.2.4...HEAD
+[0.2.4]: https://github.com/sysevol-ai/CodeNib/compare/v0.2.3...v0.2.4
+[0.2.3]: https://github.com/sysevol-ai/CodeNib/compare/v0.2.2...v0.2.3
 [0.2.2]: https://github.com/sysevol-ai/CodeNib/compare/v0.2.1...v0.2.2
 [0.2.1]: https://github.com/sysevol-ai/CodeNib/compare/v0.2.0...v0.2.1
 [0.2.0]: https://github.com/sysevol-ai/CodeNib/compare/v0.1.0...v0.2.0

codenib-0.2.4/PKG-INFO

@@ -0,0 +1,266 @@
+Metadata-Version: 2.4
+Name: codenib
+Version: 0.2.4
+Summary: A multi-view data system for serving repository context to coding agents
+Author-email: Zhongming Yu <zhy025@ucsd.edu>
+License-Expression: Apache-2.0
+Project-URL: Documentation, https://docs.codenib.ai/
+Project-URL: Homepage, https://codenib.ai
+Project-URL: Issues, https://github.com/sysevol-ai/CodeNib/issues
+Project-URL: Repository, https://github.com/sysevol-ai/CodeNib.git
+Keywords: code-search,coding-agents,model-context-protocol,repository-context,static-analysis
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: Programming Language :: Python :: 3 :: Only
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Classifier: Operating System :: OS Independent
+Classifier: Topic :: Software Development
+Classifier: Topic :: Software Development :: Libraries :: Python Modules
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: fastapi>=0.100.0
+Requires-Dist: filelock>=3.0.0
+Requires-Dist: pydantic>=2.0.0
+Requires-Dist: pyyaml>=6.0
+Requires-Dist: rank_bm25>=0.2.2
+Requires-Dist: regex>=2024.11.6
+Requires-Dist: rich>=14.0.0
+Requires-Dist: tree-sitter>=0.25.2
+Requires-Dist: tree-sitter-language-pack<1.9.0,>=1.8.1
+Requires-Dist: uvicorn>=0.20.0
+Provides-Extra: auth
+Requires-Dist: keyring>=25.0.0; extra == "auth"
+Requires-Dist: requests>=2.25.0; extra == "auth"
+Provides-Extra: grep
+Requires-Dist: requests>=2.25.0; extra == "grep"
+Requires-Dist: ripgrep-bin<16,>=15.2.0; extra == "grep"
+Provides-Extra: agent
+Requires-Dist: litellm>=1.93.0; extra == "agent"
+Requires-Dist: openai>=1.0.0; extra == "agent"
+Requires-Dist: requests>=2.25.0; extra == "agent"
+Provides-Extra: datasets
+Requires-Dist: datasets>=4.0.0; extra == "datasets"
+Provides-Extra: graph
+Requires-Dist: igraph>=0.11.6; extra == "graph"
+Requires-Dist: protobuf>=3.20.0; extra == "graph"
+Provides-Extra: mcp
+Requires-Dist: mcp<3,>=2.0.0; extra == "mcp"
+Provides-Extra: semantic
+Requires-Dist: einops>=0.8.0; extra == "semantic"
+Requires-Dist: faiss-cpu>=1.7.0; extra == "semantic"
+Requires-Dist: numpy>=1.24.0; extra == "semantic"
+Requires-Dist: openai>=1.0.0; extra == "semantic"
+Requires-Dist: sentence-transformers>=2.3.0; extra == "semantic"
+Provides-Extra: semantic-remote
+Requires-Dist: faiss-cpu>=1.7.0; extra == "semantic-remote"
+Requires-Dist: numpy>=1.24.0; extra == "semantic-remote"
+Requires-Dist: openai>=1.0.0; extra == "semantic-remote"
+Provides-Extra: serving
+Requires-Dist: numpy>=1.24.0; extra == "serving"
+Requires-Dist: sse-starlette>=2.0.0; extra == "serving"
+Requires-Dist: torch>=2.2.0; extra == "serving"
+Requires-Dist: transformers>=5.15.0; extra == "serving"
+Provides-Extra: vertex
+Requires-Dist: litellm[google]>=1.93.0; extra == "vertex"
+Provides-Extra: zoekt
+Requires-Dist: requests>=2.25.0; extra == "zoekt"
+Provides-Extra: full
+Requires-Dist: datasets>=4.0.0; extra == "full"
+Requires-Dist: einops>=0.8.0; extra == "full"
+Requires-Dist: faiss-cpu>=1.7.0; extra == "full"
+Requires-Dist: igraph>=0.11.6; extra == "full"
+Requires-Dist: litellm[google]>=1.93.0; extra == "full"
+Requires-Dist: mcp<3,>=2.0.0; extra == "full"
+Requires-Dist: numpy>=1.24.0; extra == "full"
+Requires-Dist: openai>=1.0.0; extra == "full"
+Requires-Dist: protobuf>=3.20.0; extra == "full"
+Requires-Dist: requests>=2.25.0; extra == "full"
+Requires-Dist: sentence-transformers>=2.3.0; extra == "full"
+Provides-Extra: dev
+Requires-Dist: black>=22.3.0; extra == "dev"
+Requires-Dist: build>=1.2.0; extra == "dev"
+Requires-Dist: isort>=5.10.1; extra == "dev"
+Requires-Dist: mypy>=0.950; extra == "dev"
+Requires-Dist: flake8>=4.0.1; extra == "dev"
+Requires-Dist: flake8-bugbear>=24.4.0; extra == "dev"
+Requires-Dist: mkdocs<2.0.0; extra == "dev"
+Requires-Dist: mkdocs-material>=9.5.0; extra == "dev"
+Requires-Dist: mkdocstrings-python>=2.0.0; extra == "dev"
+Requires-Dist: mkdocs-api-autonav>=0.4.0; extra == "dev"
+Provides-Extra: test
+Requires-Dist: datasets>=4.0.0; extra == "test"
+Requires-Dist: einops>=0.8.0; extra == "test"
+Requires-Dist: faiss-cpu>=1.7.0; extra == "test"
+Requires-Dist: igraph>=0.11.6; extra == "test"
+Requires-Dist: litellm>=1.93.0; extra == "test"
+Requires-Dist: matplotlib>=3.4.0; extra == "test"
+Requires-Dist: mcp<3,>=2.0.0; extra == "test"
+Requires-Dist: numpy>=1.24.0; extra == "test"
+Requires-Dist: openai>=1.0.0; extra == "test"
+Requires-Dist: protobuf>=3.20.0; extra == "test"
+Requires-Dist: pytest>=6.0.0; extra == "test"
+Requires-Dist: pytest-cov>=3.0.0; extra == "test"
+Requires-Dist: pytest-timeout>=2.1.0; extra == "test"
+Requires-Dist: pytest-xdist>=3.0.0; extra == "test"
+Requires-Dist: requests>=2.25.0; extra == "test"
+Requires-Dist: sentence-transformers>=2.3.0; extra == "test"
+Requires-Dist: sse-starlette>=2.0.0; extra == "test"
+Requires-Dist: tomli>=1.1.0; python_version < "3.11" and extra == "test"
+Dynamic: license-file
+
+<!--
+SPDX-FileCopyrightText: 2025-2026 CodeNib Contributors
+
+SPDX-License-Identifier: Apache-2.0
+-->
+<!-- mcp-name: ai.codenib/codenib -->
+
+<div align="center">
+  <img src="https://raw.githubusercontent.com/sysevol-ai/CodeNib/main/assets/codenib_logo.svg" alt="CodeNib" width="440">
+  <h1>Find the right code. Trace the calls.</h1>
+  <p>Give Claude Code and Codex source context and typed graph navigation across languages.</p>
+  <p>
+    <strong>71.4% code-block Recall@5 on 100 real issues.</strong><br>
+    Model-planned grep → Jev: +12.8 percentage points over the same grep candidates without reranking.<br>
+    <a href="https://docs.codenib.ai/evaluation/grep_jev/">Experimental result · Method and limitations</a>
+  </p>
+  <p>
+    <a href="#quickstart">Use with your agent</a>
+    &nbsp;&middot;&nbsp;
+    <a href="https://demo.codenib.ai">Browse a Wiki</a>
+    &nbsp;&middot;&nbsp;
+    <a href="https://docs.codenib.ai/language_capabilities/">Languages</a>
+    &nbsp;&middot;&nbsp;
+    <a href="https://docs.codenib.ai/agent_integrations/">Reproductions</a>
+    &nbsp;&middot;&nbsp;
+    <a href="https://github.com/sysevol-ai/CodeNib/releases">Releases</a>
+  </p>
+  <p>
+    <a href="https://github.com/sysevol-ai/CodeNib/actions/workflows/ci-full.yml"><img src="https://github.com/sysevol-ai/CodeNib/actions/workflows/ci-full.yml/badge.svg" alt="CI"></a>
+    <a href="https://pypi.org/project/codenib/"><img src="https://img.shields.io/pypi/v/codenib.svg?cacheSeconds=300" alt="PyPI version"></a>
+    <a href="https://github.com/sysevol-ai/CodeNib/releases/latest"><img src="https://img.shields.io/github/v/release/sysevol-ai/CodeNib" alt="Latest GitHub Release"></a>
+    <a href="https://arxiv.org/abs/2607.25431"><img src="https://img.shields.io/badge/arXiv-2607.25431-b31b1b.svg" alt="arXiv paper"></a>
+    <a href="https://github.com/sysevol-ai/CodeNib/blob/main/LICENSE"><img src="https://img.shields.io/badge/License-Apache_2.0-blue.svg" alt="Apache 2.0 license"></a>
+  </p>
+</div>
+
+<p align="center">
+  <a href="https://github.com/sysevol-ai/CodeNib/blob/main/landing/assets/demos/codegraph-claude.mp4">
+    <img src="https://raw.githubusercontent.com/sysevol-ai/CodeNib/main/landing/assets/demos/codegraph-claude.png" alt="Play a recorded CodeGraph setup and Claude Code explore_context call on Requests, returning source references for authentication during redirects" width="100%">
+  </a>
+</p>
+<p align="center"><a href="https://github.com/sysevol-ai/CodeNib/blob/main/landing/assets/demos/codegraph-claude.mp4">▶ Watch the 15-second replay</a> · real CLI output · waits condensed ·
+  <a href="https://docs.codenib.ai/codegraph/#recorded-agent-session">Pinned source, transcript and setup details</a>.
+</p>
+
+## Quickstart
+
+Requires Python 3.10+, Git, a clean repository, and Claude Code or Codex.
+
+```bash
+python -m pip install "codenib[graph,mcp]"
+codenib codegraph init /path/to/your/repo
+```
+
+Then ask your agent: **“Use CodeNib's `explore_context` to find where request
+retry behavior is implemented. Cite the files and lines.”**
+
+This shipping path builds local search and a typed symbol graph, then connects
+installed agent clients. It needs no model, API key, or GPU for CodeNib;
+your agent uses its own model. It is separate from the grep/Jev experiment
+above. Language toolchain and project prerequisites vary.
+
+[Setup and troubleshooting](https://docs.codenib.ai/codegraph/)
+· [Source exclusions](https://docs.codenib.ai/codegraph/#select-the-repository-source-surface)
+· [All MCP tools](https://docs.codenib.ai/mcp/)
+· [Local Wiki](https://docs.codenib.ai/quickstart/)
+
+## Why CodeNib
+
+- **Inspect the evidence.** Retrieve bounded code context with file and line
+  references, then follow callers, callees, definitions, and references.
+- **Use the same tools across languages.** The registry tracks 14 language
+  entries, including 12 with graph backends. Check capability and setup
+  differences in the [language matrix](https://docs.codenib.ai/language_capabilities/).
+- **Compare methods on the same source.** Pinned agent contracts, datasets,
+  and scorer checks make the [reproduction surface](https://docs.codenib.ai/agent_integrations/)
+  inspectable.
+- **Share what you found.** Export a Wiki with source citations to your own
+  [GitHub Pages](https://docs.codenib.ai/github_pages/), or browse the
+  [public examples](https://demo.codenib.ai).
+
+## How it compares
+
+“Local” and “key” describe the repository tool, excluding the agent's model.
+This compares capabilities, not benchmark scores.
+
+| Tool | Local processing | Tool needs a model/key | Graph / navigation | Languages | Updates |
+| --- | --- | --- | --- | --- | --- |
+| grep / read | Yes | No | Text and files | Any text | Current files |
+| **CodeNib CodeGraph** | Yes | No | Typed SCIP/LSP graph | [14 chunkers / 12 graph entries](https://docs.codenib.ai/language_capabilities/) | Reuse unchanged views; rebuild changed views |
+| [Serena](https://github.com/oraios/serena) | Yes | No retrieval model | LSP/IDE symbol navigation and editing | [Backend matrix](https://oraios.github.io/serena/01-about/020_programming-languages.html) | Backend-managed project state |
+| [CodeGraph (Lordymine)](https://github.com/Lordymine/codegraph) | Yes | No retrieval model | Typed call graph | Go, TS/JS | Re-index on launch; no-op if unchanged |
+| [DeepWiki public MCP](https://docs.devin.ai/work-with-devin/deepwiki-mcp) | Hosted | No user key | Wiki and generated answers | Public indexed repos | Service-managed |
+
+[Detailed comparison and boundaries](https://docs.codenib.ai/comparison/).
+CodeNib's **experimental grep → Jev** route uses OpenRouter for planning and
+reranking; selected code goes to remote models. Its measured result does not
+apply to the model-free CodeGraph row.
+
+## Languages
+
+Graph backends: **Python, Go, Rust, C/C++, C#, Java, Ruby, PHP, Kotlin,
+Scala, JavaScript, and TypeScript**. Swift and Lua support chunking/retrieval.
+Provider prerequisites and coverage differ by language.
+
+The [generated language matrix](https://docs.codenib.ai/language_capabilities/)
+separates chunking, graph backends, incremental-backend support, and decoder
+parity. The product currently reuses or rebuilds views; file-level delta
+repair is not enabled.
+
+## Results and reproductions
+
+| Start here | What you can inspect |
+| --- | --- |
+| [grep → Jev result](https://docs.codenib.ai/evaluation/grep_jev/) | 100-issue retrieval comparison, candidate controls, model use and limitations |
+| [Agent integration matrix](https://docs.codenib.ai/agent_integrations/) | Revision-pinned LocAgent, Agentless, CoSIL, OrcaLoca and RepoNavigator contracts; compatibility does not imply reproduced paper scores |
+| [Dataset and benchmark matrix](https://docs.codenib.ai/evaluation/) | CodeNib Base/Synthesis, SWE-bench variants, Loc-Bench and SWE-Explore support |
+| [SWE-Explore validation](https://docs.codenib.ai/evaluation/swe_explore/) | 1,020/1,020 real-output metric cells match the pinned official evaluator on a fixed 20-case run |
+| [DGX Spark deployment](https://docs.codenib.ai/guides/reference-deployments/dgx-spark/) | Local Wiki, CodeGraph and model serving on GB10; a deployment guide, not a token-saving benchmark |
+
+## Documentation and community
+
+[Documentation](https://docs.codenib.ai/)
+· [Architecture](https://docs.codenib.ai/concepts/architecture/)
+· [Contributing](https://github.com/sysevol-ai/CodeNib/blob/main/CONTRIBUTING.md)
+· [CI and testing](https://docs.codenib.ai/ci_cd/)
+· [Changelog](https://github.com/sysevol-ai/CodeNib/blob/main/CHANGELOG.md)
+· [Discord](https://discord.gg/ySer6CGn4)
+
+CodeNib is in beta; public interfaces may change before a stable release.
+Historical research artifacts retain their published dataset identifiers.
+
+## Citation
+
+If you use CodeNib in your research, please cite our
+[arXiv paper](https://arxiv.org/abs/2607.25431):
+
+```bibtex
+@misc{yu2026codenibmultiviewdataserving,
+      title={CodeNib: A Multi-View Data System for Serving Repository Context to Coding Agents},
+      author={Zhongming Yu and Hengjia Yu and Boqin Yuan and Shuting Zhao and Yizhao Chen and Aryan Dokania and Mihir Jagtap and Jiayu Chang and Yitong Ma and Yash Jayswal and Wentao Ni and Hejia Zhang and Zhaoling Chen and Gangda Deng and Jishen Zhao},
+      year={2026},
+      eprint={2607.25431},
+      archivePrefix={arXiv},
+      primaryClass={cs.SE},
+      url={https://arxiv.org/abs/2607.25431},
+}
+```
+
+CodeNib is licensed under [Apache 2.0](https://github.com/sysevol-ai/CodeNib/blob/main/LICENSE).

codenib-0.2.4/README.md

@@ -0,0 +1,150 @@
+<!--
+SPDX-FileCopyrightText: 2025-2026 CodeNib Contributors
+
+SPDX-License-Identifier: Apache-2.0
+-->
+<!-- mcp-name: ai.codenib/codenib -->
+
+<div align="center">
+  <img src="https://raw.githubusercontent.com/sysevol-ai/CodeNib/main/assets/codenib_logo.svg" alt="CodeNib" width="440">
+  <h1>Find the right code. Trace the calls.</h1>
+  <p>Give Claude Code and Codex source context and typed graph navigation across languages.</p>
+  <p>
+    <strong>71.4% code-block Recall@5 on 100 real issues.</strong><br>
+    Model-planned grep → Jev: +12.8 percentage points over the same grep candidates without reranking.<br>
+    <a href="https://docs.codenib.ai/evaluation/grep_jev/">Experimental result · Method and limitations</a>
+  </p>
+  <p>
+    <a href="#quickstart">Use with your agent</a>
+    &nbsp;&middot;&nbsp;
+    <a href="https://demo.codenib.ai">Browse a Wiki</a>
+    &nbsp;&middot;&nbsp;
+    <a href="https://docs.codenib.ai/language_capabilities/">Languages</a>
+    &nbsp;&middot;&nbsp;
+    <a href="https://docs.codenib.ai/agent_integrations/">Reproductions</a>
+    &nbsp;&middot;&nbsp;
+    <a href="https://github.com/sysevol-ai/CodeNib/releases">Releases</a>
+  </p>
+  <p>
+    <a href="https://github.com/sysevol-ai/CodeNib/actions/workflows/ci-full.yml"><img src="https://github.com/sysevol-ai/CodeNib/actions/workflows/ci-full.yml/badge.svg" alt="CI"></a>
+    <a href="https://pypi.org/project/codenib/"><img src="https://img.shields.io/pypi/v/codenib.svg?cacheSeconds=300" alt="PyPI version"></a>
+    <a href="https://github.com/sysevol-ai/CodeNib/releases/latest"><img src="https://img.shields.io/github/v/release/sysevol-ai/CodeNib" alt="Latest GitHub Release"></a>
+    <a href="https://arxiv.org/abs/2607.25431"><img src="https://img.shields.io/badge/arXiv-2607.25431-b31b1b.svg" alt="arXiv paper"></a>
+    <a href="https://github.com/sysevol-ai/CodeNib/blob/main/LICENSE"><img src="https://img.shields.io/badge/License-Apache_2.0-blue.svg" alt="Apache 2.0 license"></a>
+  </p>
+</div>
+
+<p align="center">
+  <a href="https://github.com/sysevol-ai/CodeNib/blob/main/landing/assets/demos/codegraph-claude.mp4">
+    <img src="https://raw.githubusercontent.com/sysevol-ai/CodeNib/main/landing/assets/demos/codegraph-claude.png" alt="Play a recorded CodeGraph setup and Claude Code explore_context call on Requests, returning source references for authentication during redirects" width="100%">
+  </a>
+</p>
+<p align="center"><a href="https://github.com/sysevol-ai/CodeNib/blob/main/landing/assets/demos/codegraph-claude.mp4">▶ Watch the 15-second replay</a> · real CLI output · waits condensed ·
+  <a href="https://docs.codenib.ai/codegraph/#recorded-agent-session">Pinned source, transcript and setup details</a>.
+</p>
+
+## Quickstart
+
+Requires Python 3.10+, Git, a clean repository, and Claude Code or Codex.
+
+```bash
+python -m pip install "codenib[graph,mcp]"
+codenib codegraph init /path/to/your/repo
+```
+
+Then ask your agent: **“Use CodeNib's `explore_context` to find where request
+retry behavior is implemented. Cite the files and lines.”**
+
+This shipping path builds local search and a typed symbol graph, then connects
+installed agent clients. It needs no model, API key, or GPU for CodeNib;
+your agent uses its own model. It is separate from the grep/Jev experiment
+above. Language toolchain and project prerequisites vary.
+
+[Setup and troubleshooting](https://docs.codenib.ai/codegraph/)
+· [Source exclusions](https://docs.codenib.ai/codegraph/#select-the-repository-source-surface)
+· [All MCP tools](https://docs.codenib.ai/mcp/)
+· [Local Wiki](https://docs.codenib.ai/quickstart/)
+
+## Why CodeNib
+
+- **Inspect the evidence.** Retrieve bounded code context with file and line
+  references, then follow callers, callees, definitions, and references.
+- **Use the same tools across languages.** The registry tracks 14 language
+  entries, including 12 with graph backends. Check capability and setup
+  differences in the [language matrix](https://docs.codenib.ai/language_capabilities/).
+- **Compare methods on the same source.** Pinned agent contracts, datasets,
+  and scorer checks make the [reproduction surface](https://docs.codenib.ai/agent_integrations/)
+  inspectable.
+- **Share what you found.** Export a Wiki with source citations to your own
+  [GitHub Pages](https://docs.codenib.ai/github_pages/), or browse the
+  [public examples](https://demo.codenib.ai).
+
+## How it compares
+
+“Local” and “key” describe the repository tool, excluding the agent's model.
+This compares capabilities, not benchmark scores.
+
+| Tool | Local processing | Tool needs a model/key | Graph / navigation | Languages | Updates |
+| --- | --- | --- | --- | --- | --- |
+| grep / read | Yes | No | Text and files | Any text | Current files |
+| **CodeNib CodeGraph** | Yes | No | Typed SCIP/LSP graph | [14 chunkers / 12 graph entries](https://docs.codenib.ai/language_capabilities/) | Reuse unchanged views; rebuild changed views |
+| [Serena](https://github.com/oraios/serena) | Yes | No retrieval model | LSP/IDE symbol navigation and editing | [Backend matrix](https://oraios.github.io/serena/01-about/020_programming-languages.html) | Backend-managed project state |
+| [CodeGraph (Lordymine)](https://github.com/Lordymine/codegraph) | Yes | No retrieval model | Typed call graph | Go, TS/JS | Re-index on launch; no-op if unchanged |
+| [DeepWiki public MCP](https://docs.devin.ai/work-with-devin/deepwiki-mcp) | Hosted | No user key | Wiki and generated answers | Public indexed repos | Service-managed |
+
+[Detailed comparison and boundaries](https://docs.codenib.ai/comparison/).
+CodeNib's **experimental grep → Jev** route uses OpenRouter for planning and
+reranking; selected code goes to remote models. Its measured result does not
+apply to the model-free CodeGraph row.
+
+## Languages
+
+Graph backends: **Python, Go, Rust, C/C++, C#, Java, Ruby, PHP, Kotlin,
+Scala, JavaScript, and TypeScript**. Swift and Lua support chunking/retrieval.
+Provider prerequisites and coverage differ by language.
+
+The [generated language matrix](https://docs.codenib.ai/language_capabilities/)
+separates chunking, graph backends, incremental-backend support, and decoder
+parity. The product currently reuses or rebuilds views; file-level delta
+repair is not enabled.
+
+## Results and reproductions
+
+| Start here | What you can inspect |
+| --- | --- |
+| [grep → Jev result](https://docs.codenib.ai/evaluation/grep_jev/) | 100-issue retrieval comparison, candidate controls, model use and limitations |
+| [Agent integration matrix](https://docs.codenib.ai/agent_integrations/) | Revision-pinned LocAgent, Agentless, CoSIL, OrcaLoca and RepoNavigator contracts; compatibility does not imply reproduced paper scores |
+| [Dataset and benchmark matrix](https://docs.codenib.ai/evaluation/) | CodeNib Base/Synthesis, SWE-bench variants, Loc-Bench and SWE-Explore support |
+| [SWE-Explore validation](https://docs.codenib.ai/evaluation/swe_explore/) | 1,020/1,020 real-output metric cells match the pinned official evaluator on a fixed 20-case run |
+| [DGX Spark deployment](https://docs.codenib.ai/guides/reference-deployments/dgx-spark/) | Local Wiki, CodeGraph and model serving on GB10; a deployment guide, not a token-saving benchmark |
+
+## Documentation and community
+
+[Documentation](https://docs.codenib.ai/)
+· [Architecture](https://docs.codenib.ai/concepts/architecture/)
+· [Contributing](https://github.com/sysevol-ai/CodeNib/blob/main/CONTRIBUTING.md)
+· [CI and testing](https://docs.codenib.ai/ci_cd/)
+· [Changelog](https://github.com/sysevol-ai/CodeNib/blob/main/CHANGELOG.md)
+· [Discord](https://discord.gg/ySer6CGn4)
+
+CodeNib is in beta; public interfaces may change before a stable release.
+Historical research artifacts retain their published dataset identifiers.
+
+## Citation
+
+If you use CodeNib in your research, please cite our
+[arXiv paper](https://arxiv.org/abs/2607.25431):
+
+```bibtex
+@misc{yu2026codenibmultiviewdataserving,
+      title={CodeNib: A Multi-View Data System for Serving Repository Context to Coding Agents},
+      author={Zhongming Yu and Hengjia Yu and Boqin Yuan and Shuting Zhao and Yizhao Chen and Aryan Dokania and Mihir Jagtap and Jiayu Chang and Yitong Ma and Yash Jayswal and Wentao Ni and Hejia Zhang and Zhaoling Chen and Gangda Deng and Jishen Zhao},
+      year={2026},
+      eprint={2607.25431},
+      archivePrefix={arXiv},
+      primaryClass={cs.SE},
+      url={https://arxiv.org/abs/2607.25431},
+}
+```
+
+CodeNib is licensed under [Apache 2.0](https://github.com/sysevol-ai/CodeNib/blob/main/LICENSE).