Repository navigation
Releases
OntoIndex 3.1.7
Changes
inspect contextreads method metadata fromMethodcolumns by label, andgraph_walkresolves its seed by id and name only. Before, both read whole nodes, so large file content exhausted the default 64 MiB committed read pool andinspect contextreportedmethod-metadata-query-failedincontextCompleteness.ontoindex analyze --workspace-graphproduces a usable workspace graph whentsconfig.jsonor another resolver configuration file is a symlink to a file inside the repository. In 3.1.6 the overlay check treated the link as deleted and failed with "workspace graph input paths changed". When the link target's bytes change, the workspace graph becomes stale.audit safe_edit_checkrethrows a database resource failure from the process, cluster, co-change, recent-touch and export probes. Any other failure in those probes, and any upstream or downstream query warning, adds a warning, names the failed probe in the reasoning and prevents aSAFEverdict. Stronger verdicts are unchanged.impactbatch reportsunion.risk: UNKNOWNandunion.failedTargetswhen any target fails, andstatus: errorwhen every target fails. Before, failed targets were excluded from the union risk, so the batch could reportNONEorLOW.ontoindex query,contextandimpactrecord usage errors, typed-query parse errors and a missing index in tool telemetry as failed calls. The printed message and exit code 1 are unchanged.cypheranddetect-changesalso print the missing-index message and exit 1. Whenontoindex impactfails for another reason, it sets exit code 1 and lets the process end normally, so its failed-call record is still written.- Workspace graph validation opens each recorded input without following symlinks and checks that it is the same regular file it checked first. An input replaced by a symlink makes the workspace graph unusable, and the link target is never read.
- Index CSV writing stores at most 16 Mi UTF-16 characters of each file's content, followed by
... [File content truncated]. SetONTOINDEX_MAX_FILE_CONTENT_CHARSto change the limit, or to0to store full content. Before, one very large file produced a CSV row that could exceed the batch limit and the V8 string limit. Full-text search does not match text past the limit. - The native CSV writer (
ONTOINDEX_NATIVE_CSV=1) writes rows in batches of 500 rows or 64 Mi characters. Before, it held every row of a table in memory until the table was finished. ontoindex analyze --workspace-graphexits 3 when workspace capture refuses an input, and 1 for other failures; runs interrupted bySIGINTorSIGTERMstill exit 130 or 143. In workspace-graph modeontoindex watchtreats exit 3 as a refusal and waits for the inputs to change. It retries other failures, such as a lock conflict or timeout, up to 8 times with backoff before it waits for repository changes.
OntoIndex 3.1.6
Changes
inspect context,impact,audit safe_edit_check,search repomapandsearch semanticwithinclude_content: truefind matching nodes from metadata first, then read requested content by node label and id from the tables that have a content column. Large file content no longer exhausts the default 64 MiB committed read pool on these paths, so they return correct answers instead of a buffer-pool error.impactandaudit safe_edit_checkreport a database resource failure as an error. Previouslyimpactreturnedrisk: UNKNOWNandsafe_edit_checkreported the symbol as not found withCAUTION. Inimpactbatch, a failed target carries its own error and the other targets still return results.- Symbols-only workspace builds accept any number of changed paths. When the path list is larger than 64 KiB as one argument, a full workspace build runs instead.
- Workspace capture reads a symlinked resolver configuration such as
tsconfig.jsonthrough its link when the target stays inside the repository, asontoindex analyzedoes. A dangling link counts as an absent file. A link that leaves the repository is refused with its path and the kindsymlink. Symlinked.gitignoreand.ontoindexignorefiles are still refused. ontoindex watchand the MCP watcher rebuild after a configuration-only edit, at startup or at the periodic check, even when no recursive file event fires. When configuration changed, standalone watch runsontoindex analyze --workspace-graph --skip-agents-md --no-stats. A refused build is not retried until a watched input changes.- Index CSV writing flushes a batch before adding a row would take it past 64 Mi characters, newlines included, so one batch cannot exceed the V8 string limit. A row larger than that is written by itself.
- Scope retention accepts traversal observations as the configuration reader emits them,
kind,path: '.'and a non-negative safe-integerdirsScanned, including scans of more than 100 directories. Earlier it rejected them asconfiguration-invalid. Records carrying the old limit fields or other extra keys are rejected. - Tool telemetry writes the start record for an MCP tool call when the call arrives, and for a
query,contextorimpactCLI call before argument validation, with one completion record when the call returns or throws. In 3.1.5, failed calls included error codes only if they reached the backend dispatcher. Calls that fail earlier, in argument validation, repository resolution, initialization or workspace preparation, are now recorded as failed, with the error code and diagnostic reason when the error carries valid ones. Super tools and the facade actions backed by them are recorded under theirgn_*names. Method names are unchanged. A call that returns a result, including a result marked as an error, records the size of that raw result before projection or paging, as in 3.1.5. When the MCP handler turns a thrown error into an error response, it records size 0. - MCP errors and CLI hints suggest committed reads only after confirming that the index has a trusted commit. When that check times out or fails, MCP errors leave out the suggestion,
query,contextandimpactworkspace-preparation errors print "Committed-read availability could not be confirmed.", and the rawimpactsize error keeps its fallback suggestion. - Documentation correction: since 3.1.2,
ensure_freshhas not emittedruntimeHealth.dirtyCoverage.cap,runtimeHealth.workspaceReadiness.maxIndexedFilesorruntimeHealth.workspaceReadiness.remediation. The TypeScript fields remain optional.
OntoIndex 3.1.5
Changes
- The MCP overlay watcher no longer watches ignored directories, so a large ignored build tree such as
target/cannot exhaust the server heap on a workspace read. On Linux with Node.js 22.x, 24.0–24.13 or 25.0–25.4, which cannot skip them, recursive watching is skipped and logged; reads and the 30-second check still pick up changes. - The MCP server exits on SIGTERM or end of input even when the client has already closed its stderr. Previously it could spin at full CPU until killed.
- Refused workspace inputs report a safe repository-relative path and whether the input is a symlink or non-regular file, when those details are available.
- Workspace capture and
ontoindex analyzeskip symlinks discovered during scanning. Explicit analysis-pack manifest inputs still reject symlinks. ensure_freshwarns when workspace readiness has not been checked, so index freshness does not imply that default workspace reads can be prepared.- Committed reads support
impactbatch operations andsearchCypher queries. - Tool telemetry records calls when they start, including the selected read mode, and links completion records by call ID. Failed calls include validated error codes and diagnostic reasons.
- MCP errors,
ensure_freshand CLI hints no longer recommend committed reads when the index has no trusted commit. Explicit committed-read failures explain how to enable them by runningontoindex analyzein a Git checkout. auditacceptstargetas an alias forsymbolinsafe_edit_check,can_deleteandtest_gap, and rejects conflicting values.
OntoIndex 3.1.4
Changes
ontoindex analyzeno longer marks its own index stale: the generated AGENTS.md and CLAUDE.md block no longer contains symbol, relationship, and flow counts, and AGENTS.md, CLAUDE.md and the bundled skill files are rewritten only when their content changes.--no-statsis a deprecated no-op.- The throttled stale warning reports the last refresh check and the next eligible time. In repositories without commits, a cached stale verdict is dropped once a new index generation is published.
OntoIndex 3.1.3
Changes
- Attempt full AST extraction without default node-count or nesting-depth ceilings. Positive environment overrides remain available.
- Report the actual parser fallback cause, including configured AST ceilings and parser or query failures.
- Explain how concurrent agents can use captured workspace graphs when source edits interrupt base-index analysis.
Existing analysis consistency checks, cancellation and runtime timeouts remain in force. Practical capacity depends on available memory and runtime limits.
OntoIndex 3.1.2
Changes
- Remove application-imposed limits on indexed file counts, source-file sizes, workspace capture size, and overlay storage.
- Include eligible large files and all eligible dirty files in workspace analysis.
- Preserve filesystem safety checks, input integrity checks, cancellation, and build timeouts.
Practical capacity depends on available memory, disk space, and runtime limits.
OntoIndex 3.1.1
OntoIndex 3.1.1
- Stop optional search enrichment after resource or timeout failures while retaining completed results.
- Bound overlay fixture cleanup and verify close-before-delete behavior across success, rejection and timeout cases.
Validation: Linux build, 13,429 unit tests and 3,052 integration tests passed. Installed CLI and MCP search smoke passed.
Source commit: d535fe8.
Install using the repository installer or the attached npm tarball. This is a GitHub release; no npm registry publication is included.
LadybugDB extensions v0.17.0 linux_amd64
LadybugDB extensions v0.17.0 linux_amd64
Offline cache assets for OntoIndex clients that cannot reliably download LadybugDB extensions during INSTALL fts or INSTALL VECTOR.
Assets:
libfts.lbug_extensionfromhttps://extension.ladybugdb.com/v0.17.0/linux_amd64/fts/libfts.lbug_extensionlibvector.lbug_extensionfromhttps://extension.ladybugdb.com/v0.17.0/linux_amd64/vector/libvector.lbug_extensionSHA256SUMS.txt
Use these as a temporary mirror/cache source; keep normal LadybugDB extension loading when client networking is healthy.