EKOS
EKOS compiles enterprise code, SQL and ETL into an evidence-backed knowledge ledger any MCP-speaking AI agent can query.
EKOS is the most honest tool of its kind on the evidence surface: it publishes its own failing numbers (49.7% answer correctness, 51.6% groundedness), files and closes real parser bugs on real repos, and shows the one case where grep wins against its 67-93% token claim. If your problem is recovering what a legacy Pentaho job or stored procedure actually does, and you need every AI answer to cite a fragment, the compiled-ledger approach is materially different from pointing a chatbot at raw files. If you want a no-code assistant or your code cannot leave your perimeter, look elsewhere.
Verified 1d ago · liveness 67/100 · cite: rightaichoice.com/tools/ekos
- Data and platform engineers inheriting legacy Pentaho, SQL or stored-procedure logic
- Teams standardizing on MCP that need cited, evidence-backed agent answers
- Organizations able to self-host a compiler pipeline against their own repositories
- Engineers willing to run a graded evaluation suite before trusting output
- Non-technical users wanting a no-code AI assistant
- Teams that need a mature, fully benchmarked product in a production path today
- Anyone without a repository or legacy system to point the compiler at
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip EKOS if you want a no-code AI assistant or need a mature, fully benchmarked tool in production today — it is a self-hosted compiler whose own graded suite currently reports 49.7% answer correctness.
Running the pipeline and serving MCP locally means you own the compute and storage for ingestion, compilation and the ledger — the vendor publishes no hosting fee because there is no hosted tier in the material we saw.
EKOS's pricing fits teams whose volume aligns with the published tiers. Compare against the alternatives listed below for stage-specific value.
In short
EKOS — EKOS compiles enterprise code, SQL and ETL into an evidence-backed knowledge ledger any MCP-speaking AI agent can query. Best for Data and platform engineers inheriting legacy Pentaho, SQL or stored-procedure logic, Teams standardizing on MCP that need cited, evidence-backed agent answers, Organizations able to self-host a compiler pipeline against their own repositories. Free to use.
What's new in EKOS
Checked yesterdayAcross the latest 3 updates: 3 changelog entries.
The first benchmark number: 67-93% fewer tokens than grep
Raw grep versus the compiled ledger, same real repo and same standard tokenizer, reports 67-93% fewer tokens for the realistic case with the one case grep wins included rather than hidden.
Interpretation of the whole 2,022-file repo, cold and timed
A cold pipeline run over the entire repo with a graded ekos ask plus MCP Q&A set surfaced three new gaps, reported honestly and fixed the same day as RFC 0059/0060/0061.
Answer quality, measured: 49.7% correctness, 51.6% groundedness
A 101-scenario suite across seven categories graded by six deterministic evaluators, run against the real compiled ledger — correctness up from 42.5%.
What people actually say about EKOS — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
24 mentions across 4 sources (Hacker News, Product Hunt, Stack Overflow, Lemmy) · researched Aug 13, 2026.
Average across the 4 sources that answered — each source counts once, not each post.
- +Instantly generates MCP servers from any GitHub repo, saving setup time.
- +Automatically detects project structure and APIs, no manual mapping needed.
- +Easy integration with Claude and Cursor via simple OAuth.
- +One-click deployment to cloud gets MCP running in minutes.
- +Supports private repos, crucial for enterprise codebases.
- −Limited community feedback; early-stage tool with unproven reliability.
- −Potential scalability issues as codebase size grows, not yet validated.
- −Pricing unclear; freemium model may hide costs for heavy usage.
- −Dependency on GitHub OAuth may limit non-GitHub users.
- −Customization options may be too shallow for advanced users.
- • Overage fees for high API usage, not clearly disclosed
- • Potential cost for cloud deployment beyond free tier
Viability Score
How well maintained and how widely used is EKOS? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: September 2026
How we score →Key Features
- Compiles SQL DDL, Pentaho jobs, git history, Confluence and GitHub issues into a shared Transformation IR
- Recovers legacy ETL and stored-procedure logic as Source, Filter, Join, Aggregate and Sink nodes
- Serves compiled knowledge read-only over MCP to any MCP-speaking agent via stdio JSON-RPC
- Evidence-backed citations: every answer names the exact fragment it came from
- Append-only, evidence-backed enterprise knowledge graph of issues, PRs and files
- Pentaho and SQL recovery analyzers run against real GitHub ETL projects
- ClickHouse connector (RFC 0056) as an audited exception allowing live query
- Local-LLM NL-to-SQL pipeline, demonstrated with an Ollama model against a real ClickHouse server
- Cold ingestion of a 2,022-file repository in 34 seconds, timed live
- Token efficiency of 67-93% fewer tokens than raw grep on the same real question
- 101-scenario answer-quality suite graded by six deterministic evaluators
- Self-hosted pipeline spanning build, recovery, resolution, compilation, storage, docs and search
- CLI workflow: ekos build, recover, resolve, compile, commit, then mcp serve --workspace
About EKOS
EKOS is a compiler for enterprise knowledge rather than another RAG index. It reads systems as raw, content-addressable evidence — SQL DDL, Pentaho jobs, git history, Confluence pages, GitHub issues — and compiles legacy logic into a single Transformation IR where a Pentaho step and a SQL SELECT become the same kind of node (Source, Filter, Join, Aggregate, Sink). The result is an evidence-backed ledger served read-only over MCP to any AI agent, so every answer cites the exact fragment it came from instead of guessing. The vendor's own end-to-end command sequence is `ekos build && ekos recover && ekos resolve && ekos compile && ekos commit`, followed by `ekos mcp serve --workspace .` to expose it over stdio to any MCP client. Published benchmarks run cold against the real, unmodified plausible/analytics repository: 34 seconds to ingest 2,022 files, a full build-to-compile run of 53 seconds on an 834-file repo, and 67-93% fewer tokens than raw grep for the same question. A 101-scenario answer-quality suite graded by six deterministic evaluators (no LLM judge) reports 49.7% answer correctness and 51.6% evidence groundedness — failing numbers are left on the page. The audience is data and platform engineers recovering business logic from legacy ETL, stored procedures and enterprise SQL, plus teams that want AI assistants reading real production code under evidence constraints rather than pasted snippets. It is emphatically not a no-code product: you bring code, a repository, and a reason to point an agent at it.
Behind the Verdict
The core claim EKOS makes is architectural, and it is worth understanding before you compare it to anything else in the MCP space. Most tools in this category index your repo and hand chunks to a model. EKOS instead compiles: raw evidence in (SQL DDL, Pentaho jobs, git history, Confluence, GitHub issues), a shared Transformation IR in the middle, and an append-only, evidence-backed ledger out. A Pentaho step and a SQL SELECT land as the same node type. That is why it can answer 'explain what this legacy pipeline does and prove a migration did not change its meaning' — a question chunk-retrieval fundamentally cannot answer, because the answer lives in semantics, not in text similarity. Strengths. The evidence model is the real product: every answer cites the exact fragment it came from, and 'Unmapped' is treated as a citizen rather than a failure, so the tool reports zero fabricated facts by design rather than by prompt engineering. The published numbers are unusually specific — 34s cold ingestion of a 2,022-file repo, 53s full build-to-compile on 834 files, 67-93% fewer tokens than grep for the same real question, with the case where grep wins shown rather than hidden. The ClickHouse work (RFC 0056/0057/0058) is a genuinely useful template: an audited exception where a connector may touch a live system, in that case a local Ollama model generating NL-to-SQL against a real ClickHouse server. Self-hosting is demonstrated, not just claimed — the pipeline is run against EKOS's own ~50-crate repository with every number measured. Weaknesses. This is early tooling with early-tooling numbers attached. 49.7% answer correctness on the vendor's own graded suite means roughly half of tested scenarios do not come back right yet; the honest reporting is admirable, but you should read that number before you put it in a production path. It is self-hosted and CLI-driven (`ekos build`, `ekos recover`, `ekos resolve`, `ekos compile`, `ekos commit`, `ekos mcp serve --workspace .`), which means real setup work and real operational ownership. The RFC-numbered development style implies a moving target — the vendor itself notes three new gaps found and fixed the same day during a cold run, which cuts both ways: responsive, but also not frozen. Where it fits. Data platform teams inheriting Pentaho, legacy SQL and stored procedures who need to explain and safely migrate them. Teams standardizing on MCP who want their agent's answers constrained to cited evidence. Organizations willing to self-host and to run the graded suite against their own repos before trusting output. Where it does not. Anyone wanting a no-code assistant. Anyone who needs a mature, stable, fully benchmarked product today rather than a compiler under active construction. And any team whose code cannot leave its own perimeter should confirm the deployment model in detail rather than assuming it from a marketing page.
Researching EKOS? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas EKOS actually fits — and what changes day-one when you adopt it.
Point EKOS at a repo containing Pentaho jobs and stored procedures, run `ekos build && ekos recover && ekos resolve && ekos compile && ekos commit`, then read the compiled Transformation IR to see each step mapped to Source, Filter, Join, Aggregate or Sink.
Outcome: You get a written explanation of what the legacy pipeline does, with every claim citing the fragment it came from, before you touch the migration.
Build the ledger on a real repo, then run `ekos mcp serve --workspace .` over stdio and ask the agent repo questions through MCP instead of pasting file contents into the chat.
Outcome: Answers stay grounded in cited fragments and token use drops — the vendor measured 67-93% fewer tokens than raw grep on the same real question.
Use the ClickHouse connector and a local Ollama model to generate NL-to-SQL against the compiled schema, then chain `ekos_clickhouse_query` MCP calls over stdio JSON-RPC to answer 'why was traffic high that day?'.
Outcome: A multi-step diagnostic question gets answered with live queries and a visible retry path when a call fails, rather than a single unverified guess.
Use Cases
- Recover and document what a legacy Pentaho or SQL ETL pipeline actually does before migrating it
- Prove a pipeline migration did not change its meaning using the evidence-backed Transformation IR
- Let a Claude or Cursor agent query a real repo through MCP with every answer citing a source fragment
- Cut token spend on repo questions by querying the compiled ledger instead of grepping raw files
- Generate component documentation for schema, write/read paths and migration frameworks
- Answer multi-step diagnostic questions such as why a metric spiked by chaining live MCP calls
Models Under the Hood
as of 2026-09-09
Limitations
- The vendor's own graded suite reports 49.7% answer correctness and 51.6% evidence groundedness — the honest numbers are published, but roughly half of tested scenarios do not come back right yet, so treat output as something to verify rather than trust.
- The tool is CLI- and self-host driven, so you own setup and operation.
- Development is RFC-numbered and moves fast; three new gaps were found and fixed the same day during one cold run.
- Automatic recovery may not cover every custom API, and the vendor explicitly treats unmapped fragments as a normal outcome rather than an error, which means some business logic will land outside the ledger until analyzers catch up.
as of 2026-09-28
Verification history
We have re-verified EKOS 5 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-checked, vendor evidence unchanged
- — re-checked, vendor evidence unchanged
- — re-checked, vendor evidence unchanged
Free to cite with attribution — this page re-verifies continuously.
Where the pricing makes sense
The company stage and team size where EKOS's pricing actually pencils out — and where peers do it cheaper.
EKOS's pricing fits teams whose volume aligns with the published tiers. Compare against the alternatives listed below for stage-specific value.
Setup time & first value
How long it actually takes to get something useful out of EKOS — broken out by persona, not the marketing-page minute.
Self-hosted CLI setup, so expect a real install and configuration session before first value: the vendor's timed runs ingest a 2,022-file repo cold in 34s and complete a full build-to-compile on 834 files in 53s, but the graded evaluation suite (101 scenarios) is what tells you whether the output is good enough for your repos.
Switching to or from EKOS
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From hand-rolled MCP servers: replace bespoke tool definitions with the compiled ledger served by `ekos mcp serve --workspace .` over stdio.
- →From raw RAG indexing: swap chunk retrieval for the Transformation IR and evidence-backed ledger so answers cite exact fragments.
- →From manual legacy documentation: compile Pentaho jobs and SQL into the shared IR instead of writing migration notes by hand.
- ↗To another MCP server: any MCP-speaking agent can point at a different server, since EKOS serves over the standard stdio protocol.
- ↗To a general-purpose code assistant: the compiled evidence layer is the differentiator you give up.
- ↗To in-house tooling: the CLI pipeline and RFC-documented analyzers are the components you would rebuild.
Integrations
Resources & Guides
Tutorials & Learning
YouTube returned 6 videos for “EKOS”, and we withheld 6: 6 could not be judged, because “EKOS” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about EKOS.
Official links
Tools that pair well with EKOS
Common stack mates teams adopt alongside EKOS, with the specific reason each pairing earns its keep.
Bito
Bito's Governor is an AI model router and code context engine that cuts coding agent spend by grounding every request in your codebase.
Chrome DevTools MCP
Chrome DevTools MCP gives AI coding agents live Chrome control for debugging, automation, and performance traces.
Moderne
Moderne sequences your repos into a Lossless Semantic Tree for deterministic code change at scale.
Featured Head-to-Head Comparisons
Ekos vs Smithery
If you want to supercharge your existing GitHub repos with AI, EKOS is the direct route. If you're building agents that need a universe of pre-built tools without auth headaches, Smithery is unbeatable. Choose based on whether you're repo-centric or agent-centric.
Ekos vs Bito
If you're a developer who wants to quickly expose your GitHub repos to AI assistants with minimal fuss, EKOS is the fast, lightweight bridge. But if you're an enterprise team wrestling with multi-repo complexity and need architectural planning, impact analysis, and epic scoping, Bito's knowledge graph approach is the heavyweight contender. Choose based on your scale: solo/startup with a few repos → EKOS; multi-repo engineering org → Bito.
Ekos vs Poolside Ai
If you're a developer who wants AI to understand your codebase without heavy integration work, EKOS is the fast, affordable route—free tier, instant MCP server from any GitHub repo. But if you operate in a regulated industry where data governance and audit trails are non-negotiable, Poolside AI's on-prem, open-weight Laguna models are the enterprise-grade choice, despite the sales-led procurement. Choose based on your risk tolerance and deployment constraints.
Alternatives to EKOS
View all
Bito
Bito's Governor is an AI model router and code context engine that cuts coding agent spend by grounding every request in your codebase.
Chrome DevTools MCP
Chrome DevTools MCP gives AI coding agents live Chrome control for debugging, automation, and performance traces.
Frequently Asked Questions
Used EKOS? Help shape our editorial sentiment research.