I · Three questions a working engineering team asks
From "where is it" to "why is it slow" to "who still owns this."
Demo 1 · Structural where-is-it search0.84 ms · local
where do we dispatch background jobs to the workers, not the dispatcher?
01WorkerDispatcher.EnqueueAsync — fans out via channel, called from BackgroundService root. src/.../WorkerDispatcher.cs:172
0.94
02JobRouter.Resolve chooses worker class by tag; fallback to "dispatcher" when tag missing. src/.../JobRouter.cs:84
0.88
03Tests covering the fan-out path · BackgroundDispatchTests.WorkersReceive_AllJobs. tests/.../BackgroundDispatchTests.cs:51
0.80
Demo 2 · "Expensive computation per chunk"0.91 ms · local
expensive computation per chunk — the word "expensive" isn't in the code
01HeaderRegexFieldExtractor.HasFolderMarker — complexity 404, called per chunk in inner loop. src/.../FieldExtractors/HeaderRegexFieldExtractor.cs:216
0.92
02Same function recompiled regex each call · cache miss × N chunks. scan · code.perf.regex-recompile · same file:289
0.88
Demo 3 · Dead code, with confidencescan · per-commit
dead code · public types unused outside test scaffolding · 90-day window
01An unused encoder class · 0 callers, last edit 2024-08, on an inactive branch. docs/reference/.../UnusedEncoder.cs
0.96
02OllamaConfig · class referenced only from a removed test fixture; live in DI graph. src/.../DaemonConfig.cs
0.89
Demos shown against Indago's own codebase (~22 420 passages). Your workspace is indexed identically, locally.
II · What's included · free
16 languages. Real grammars. Free.
The Code bundle is free forever, capped at 3 workspaces · 25 k passages/workspace. Upgrade to Plus for cross-domain bridges, contradictions, novelty and full dead-code; Pro for autonomous discovery loops and decision ledgers.
- C# · C++ · Python · TypeScript · Swift · Rust · Go · Java · Kotlin · Ruby · PHP · JavaScripttree-sitter grammars · every result lands on a real syntactic unit
- Ranked search by meaningfinds the 300× bug grep would miss
- Dead-code & complexity hotspotsscan · code.dead-code.{lang} · code.quality.complexity
- Cross-domain bridgesconnect code to papers / docs / regulations (Plus tier)
- Conversation indexingyour Claude / Codex sessions are searchable — nothing lost in compaction
Free sourcesGitHubStackOverflowPyPInpmcrates.ioNuGetarXiv (broad)Wikipedia
II.5 · Sample workflows
Real workflows. Real numbers.
Three workflows shaped for this bundle. Where the box reads Verified, the numbers come from a real run logged in this codebase. Where it reads Pending validation, the workflow shape is implemented and we are still gathering the headline numbers before we publish them.
Workflow 01
Verified
RepoQA on this repo · R@1 = 0.498
Indago dogfood
- InputIndago indexed its own source tree — 22 420 passages across 3 581 files (C#, Python, TypeScript, Swift, Shell, Ruby).
- MethodRun the public RepoQA benchmark in code mode end-to-end with the shipping ranker.
- OutputPer-language R@1 numbers and a recall ceiling report.
R@1 = 0.498 overall
python 0.58 · ts 0.56 · rust 0.55 · go 0.55 · cpp 0.39 · java 0.36 — measured on the public benchmark against this very codebase, 2026-04-26.
Workflow 02
Verified
Complexity hotspots · indago_brief
Senior engineers
- InputPoint Indago at any workspace.
- Method
indago_brief walks the index and ranks complexity / dead-code / untested-types in one pass.
- OutputTop-5 hotspot list with cyclomatic, nesting and line-count metrics.
Real hotspot, this repo
HeaderRegexFieldExtractor.HasFolderMarker — complexity 404, nesting 1, 332 lines — surfaced on the current Indago repo, 2026-05-15.
Workflow 03
Pending validation
Past-conversation recovery · after /compact
Agent-using engineers
- InputA workspace where Claude / Codex / Cursor sessions have been auto-indexed.
- Method
indago_find mode=conversation:SESSION_ID over the local conversation cache.
- OutputA ranked thread reconstruction — the full transcript is recoverable after compaction.
Compaction-safe
Conversation indexing is shipped; pending a published demo workflow to show the exact recall.
Verified workflows are reproducible against the index hash listed in each run. Pending workflows are a written commitment, not a customer quote — we publish numbers only when we have measured them.
Code bundleFree· 3 workspaces · 25 k passages· Plus & Pro for full breadth, autonomous loops, decision ledgers
III · Also useful for engineering teams