codenib 0.2.4__tar.gz → 0.2.5__tar.gz

codenib-0.2.5/CHANGELOG.md

@@ -0,0 +1,389 @@
+<!--
+SPDX-FileCopyrightText: 2025-2026 CodeNib Contributors
+
+SPDX-License-Identifier: Apache-2.0
+-->
+
+# Changelog
+
+All notable user-facing changes are recorded here. CodeNib follows
+[Semantic Versioning](https://semver.org/).
+
+## [Unreleased]
+
+## [0.2.5] - 2026-11-04
+
+### Added
+
+- `codenib codegraph check` verifies call-path rules from
+  `.codenib/checks.yaml`: every path to a symbol must pass a guard, or a
+  function that calls a sanitizer. It reports a witness path per violation,
+  limits the report to a diff with `--base`, and exits 1 for CI. Unresolved
+  Python calls are matched by name so a missing type cannot hide a path.
+- Bounded static entry paths in `explore_context`, with source locations,
+  explicit ambiguity and omitted-path diagnostics, and readable text output.
+- Public-repository visitor Wikis with linked chapters, persistent generation
+  progress, resumable saved results, model selection, share cards and explicit
+  owner publication to Browse. Saved results begin unlisted.
+- Source-bound architecture diagrams, saved code maps, a seekable system-map
+  film and a self-contained first-visit repository example.
+- OKF v0.2 export of complete cached Wikis, with source excerpts, evidence
+  sidecars, version identities and a hashed file inventory. Explicit
+  `--agent-dir` maintains Wiki discovery pointers in agent instruction files.
+- Optional BM25-first CodeGraph initialization with `--defer-graph`, a composed
+  exploration tool surface and on-demand graph preparation. Deferred Python
+  full-tool sessions can return bounded source definition candidates.
+- ContextBench and Agent Retrieval Bench adapters with reproducible inputs,
+  source-delivery receipts and benchmark-native scoring. Their reported results
+  remain separate from CodeGraph workflow-benefit claims.
+
+### Changed
+
+- Use pinned Pyrefly 1.4.0 as the default Python graph provider, with explicit
+  scip-python selection and fallback, source-scope checks and native parity.
+- Prepare deferred Python graphs in the background and answer from source
+  retrieval while graph preparation is still running.
+- Accept dirty checkouts during CodeGraph setup, reject stale indexed queries
+  after edits and reload explicitly rebuilt indexes in the same MCP session.
+- Share the architecture reader between prepared and visitor Wikis, show
+  unavailable visual coverage and preserve source-pane layout during navigation.
+- Add camelCase parts to BM25 tokens and retain validation-test candidates
+  without crowding out implementation matches.
+
+### Fixed
+
+- Refresh compatible locked HTTP/crypto dependencies and frontend build/test
+  dependencies for the release candidate.
+- Refresh deferred retrieval after source edits, retain stale-source detection
+  across reloads, and cancel owned graph builds during MCP shutdown.
+- Preserve call-chain source lines, delivery budgets and delegation; retain
+  uncertainty for bounded or ambiguous same-named method searches.
+- Resolve Python names in enclosing function scopes, reject rebound receiver
+  fields, and prevent PEP 695 names from inheriting their owner's identity.
+- Recover Python module, field and wildcard-reference identities; bind retrieval
+  seeds to indexed spans and retain complete requested definitions within source
+  delivery limits.
+- Cover Go test files and nested modules, preserve LSP replies across reads and
+  idle drains, and bound language-server teardown.
+- Preserve source authority across ignored cache churn and publish scip-python
+  indexes only after the indexer finishes, without analyzing caller site-packages.
+- Bound SCIP test-fixture retries and failure cleanup; repair schematic map
+  routing and title/tag overlap.
+
+### Removed
+
+- Retired the source-only H1 hybrid-index experiment. Product storage remains
+  Wiki-only; manifest-bound indexes and portable artifacts remain file-based.
+
+## [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
+
+- Persistent, exact repository-relative source exclusions through
+  `--exclude-dir` and `--clear-exclude-dirs`, with one recorded policy reused
+  by CodeGraph, indexing, MCP, Wiki, Web, and public compiler entry points.
+- Manifest 1.2 source-selection identities and builder receipts, while retaining
+  strict read compatibility for manifest 1.1 artifacts.
+- Authenticated optional SCIP occurrence-sidecar receipts bound to the same
+  owned symbol-graph generation as `graph.pkl`.
+
+### Changed
+
+- Authenticated indexing now accepts absolute symbolic links only when their
+  lexical target remains inside the same pinned repository authority. Expo,
+  CocoaPods, and pnpm-style contained link layouts no longer require a broad
+  default-ignore rule.
+- CodeGraph status reports the effective source selection, and Wiki, publish,
+  doctor, and toolchain flows reuse it instead of silently widening the source
+  surface.
+- Manifest-bound C and C++ queries temporarily use the verified persisted
+  symbol graph instead of consuming project-local clangd postings directly.
+
+### Security
+
+- Source paths, graph artifacts, native query routes, and retained manifests
+  are checked against the same authenticated source selection before they can
+  be published or served. Unproven native routes fail closed to the persisted
+  graph.
+- Zoekt builds use a fixed commit-tree contract, private generation
+  publication, and an authenticated shard-tree receipt; custom exclusions are
+  rejected until the backend can prove that narrower surface end to end.
+- Authenticated MCP `search_zoekt` serving uses a retained Linux `/proc`
+  descriptor snapshot. Other platforms fail closed until they have an
+  equivalent immutable runtime handoff.
+
+## [0.2.1] - 2026-08-14
+
+### Added
+
+- A one-command `codenib codegraph init` path that installs repository-aware
+  graph providers, builds BM25 and symbol-graph views, and configures installed
+  Codex and Claude Code clients through their native MCP CLIs.
+- Machine-readable CodeGraph status, idempotent client receipts, safe uninstall,
+  and a no-write dry run.
+- Installed-wheel black-box coverage for `explore_context`,
+  `dependency_subgraph`, both client contracts, repeat initialization, and
+  checkout cleanliness.
+
+### Changed
+
+- The primary agent quickstart now leads with the local, model-free CodeGraph;
+  the Wiki and semantic routes remain available as independent product paths.
+
+### Security
+
+- Client setup refuses unmanaged name collisions and drifted managed
+  registrations. Uninstall removes only receipt-owned entries and preserves
+  repository indexes.
+
+## [0.2.0] - 2026-08-05
+
+### Added
+
+- A build-once distribution path that publishes a static Wiki and a verified,
+  commit-addressed context artifact for local or MCP reuse.
+- Capability-aware `search_context` serving through the official
+  `ai.codenib/codenib` MCP package, plus managed package-level SCIP and LSP
+  providers selected from repository languages.
+- Native repository-exploration adapters and quality-constrained token and
+  cost reports for supported agent localization policies.
+
+### Changed
+
+- Hybrid BM25+dense retrieval is the recommended installed default when the
+  semantic extra is present; the base package retains an explicit no-model
+  BM25 path.
+- Repository filtering, graph definition provenance, and TypeScript/TSX import
+  edges use updated builder contracts. Incompatible 0.1 views rebuild once;
+  compatible 0.2 views remain reusable.
+
+### Security
+
+- Downloaded context artifacts are inventory-checked, source-bound, and reject
+  unsafe archive members before any view is loaded.
+
+## [0.2.0a2] - 2026-08-05
+
+### Added
+
+- A capability-aware `search_context` MCP tool and version-matched metadata for
+  publishing the local stdio server as `ai.codenib/codenib` in the official MCP
+  Registry.
+- A repository-aware `codenib toolchain` command that detects graph and LSP
+  providers from the language registry, installs pinned package-managed tools
+  under `~/.codenib/toolchains`, and reports system or project prerequisites
+  without invoking `sudo` or mutating the target checkout.
+
+### Changed
+
+- Local `index` and `wiki` commands now select semantic hybrid retrieval when
+  the semantic extra is installed and explicitly report the no-model BM25
+  fallback otherwise. The reusable Pages workflow selects semantic retrieval
+  by default and caches its pinned local embedding model; `fast` remains an
+  explicit no-model route.
+- CodeNib processes automatically prefer the managed tool directory, removing
+  the shell `PATH` export previously required after source-checkout installs.
+- The public quickstart recommends BM25+dense serving and uses a normal,
+  version-pinned PyPI command instead of exposing a TestPyPI wheel URL.
+
+## [0.2.0a1] - 2026-08-05
+
+### Added
+
+- A reusable GitHub Pages workflow and composite Action that build a no-model
+  static Wiki plus a commit-addressed BM25 context artifact. Semantic builds
+  can use a pinned local CodeRankEmbed model or an explicit OpenAI-compatible
+  embedding endpoint.
+- Portable context artifacts with complete file inventories, source identity,
+  capability metadata, and an artifact-backed MCP path that reuses BM25 and
+  vector views without rebuilding the repository.
+- Secret-free inference route identities for local and remote embeddings,
+  compatibility fingerprints, provider-aware diagnostics, and fail-closed
+  handling of the retired GitHub Models service.
+- A native repository explorer with selective view loading and revision-pinned
+  compatibility layers for SWE-Explore, Agentless, CoSIL, LocAgent, and
+  OrcaLoca SearchAgent contracts.
+- Quality-constrained cost reports that pair localization success with token,
+  USD, and amortized build cost rather than reporting cost without a quality
+  denominator.
+- Graph schema 5 records semantic symbol kinds, explicit definition
+  provenance, and anchored TypeScript/TSX import and re-export edges.
+
+### Changed
+
+- Static Wiki exports now preserve mount paths, source citations, generated
+  pages, and dependency data while keeping query engines in the local or MCP
+  runtime.
+- Pull requests build and inspect distributions once; the complete
+  cross-version and installed-service release matrix runs on `main`, release
+  tags, and explicit TestPyPI dispatches.
+- Static navigation and source adapters no longer treat reference-only symbol
+  provenance as a definition location. Existing schema 4 graph artifacts need
+  a one-time `codenib index <repo> --preset graph --rebuild`.
+
+### Security
+
+- Downloaded context artifacts reject pickle payloads, unsafe archive paths,
+  symbolic and special files, unbounded expansion, inventory mismatches, and
+  source checkouts that do not match the recorded commit and fingerprint.
+- GitHub artifact downloads verify the API digest and never forward the GitHub
+  bearer token to redirects. Pages publication scans outputs for configured
+  credentials and rejects secret-bearing fork execution paths.
+
+## [0.1.0] - 2026-07-26
+
+### Added
+
+- A unified `codenib` CLI with `index`, `wiki`, `mcp`, and `doctor` commands.
+- A two-command local Wiki path with automatic language detection and reusable
+  repository manifests.
+- Fast, semantic, and full indexing presets with explicit optional
+  dependencies.
+- A deterministic static Wiki that requires no LLM credentials.
+- A source-linked MCP server for retrieval, graph, and static navigation tools.
+- Multi-view incremental maintenance and evaluation infrastructure.
+
+### Changed
+
+- Renamed the maintained package, command, state, and protocol namespace to
+  CodeNib.
+- Reduced the default installation to the BM25 and Wiki runtime; model, graph,
+  dataset, MCP, and agent dependencies now use named extras.
+- Made public package imports lazy so CLI startup does not initialize model or
+  graph runtimes.
+
+### Security
+
+- Prepared secretless PyPI publishing through a dedicated GitHub Actions OIDC
+  workflow.
+
+[Unreleased]: https://github.com/sysevol-ai/CodeNib/compare/v0.2.5...HEAD
+[0.2.5]: https://github.com/sysevol-ai/CodeNib/compare/v0.2.4...v0.2.5
+[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
+[0.2.0a2]: https://github.com/sysevol-ai/CodeNib/compare/620a82d...78e12ac
+[0.2.0a1]: https://github.com/sysevol-ai/CodeNib/tree/620a82d
+[0.1.0]: https://github.com/sysevol-ai/CodeNib/releases/tag/v0.1.0