We’ve spent 5+ years doing deep Risk Analysis and Due Diligence for $1B+ in DeFi assets. Today, we’re open-sourcing one piece of our toolkit, delivering structured & source-tagged analysis + risk assessment.
Date
Topic
Tools
Why we built this
For more than five years, Telos Consilium has done deep risk analysis and due diligence on over $1B in DeFi assets, for funds, treasuries, and individuals deploying serious capital on-chain. In that time, one pattern has repeated itself more than any other: most "due diligence" in crypto is a screenshot, a TVL number, and a vibe.
It's not that people don't care. It's that real due diligence is slow, it requires touching a dozen different sources, and it demands the discipline to say "I don't know" instead of filling gaps with the project's own marketing. Most people evaluating an opportunity don't have the time, the tooling, or the trained reflex to do all of that, so they don't, and they end up making decisions on vibes dressed up as research.
We built our internal DD process specifically to fight that pattern. Today, we're publishing one piece of it, TeloSkills, as a free, open skill for Claude. Point it at any token, protocol, or chain, and it runs a structured due diligence process across product, team, technology, economics, and on-chain reality. Not a rehash of the project's own marketing, an independent first pass, built on how we actually evaluate risk.
It's free. It's public. And it's built to raise the floor on what "doing your own research" actually means.
The core rule: no hype, no guessing
The single rule that shapes everything else in TeloSkills is simple to state and hard to enforce: it doesn't hype, and it doesn't guess.
Every claim in a TeloSkills report is tagged by where it came from:
[on-chain], genuinely read from chain state this session: a contract call, an RPC query, a transaction hash. Not a dashboard summarizing on-chain data, the chain itself.[third-party], sourced from an explorer, an aggregator, an audit report, or another independent party, cited by name and link.[team-disclosed], stated by the project itself, clearly labeled as such rather than presented as independently verified fact.[unknown], could not be established. This is not a failure state to hide; it's an honest, load-bearing part of the report.
If something can't be verified, TeloSkills says so, plainly, rather than filling the gap with something plausible-sounding. We'd rather a report say "unknown" than sound confident and be wrong, because confident-and-wrong is exactly what gets people hurt in this industry.
What it actually looks for
Generic checklists produce generic reports. TeloSkills is built around the failure modes that actually blow up protocols, the patterns we've seen recur across years of post-mortems. It screens for eleven categorical red flags, including:
Liquidity concentrated in a small number of wallets
A token used as its own collateral (reflexive design)
Manipulable or centralized oracles
Off-chain single points of failure behind an on-chain façade
Sanctions or regulatory exposure
Any one of these can be enough to set a project aside, they're not scored and averaged away by unrelated positives.
Reasoning, not box-ticking
The part that actually differentiates a serious DD from a checklist is what happens between the questions, the reasoning. TeloSkills is built to:
Run a Claims vs Reality pass, what the team says, tested directly against what the chain and independent sources show, in a structured table.
Check the math. A lot of "how could this go wrong" is answerable by simply asking: is this yield even mathematically possible given where it claims to come from?
Stress-test with real breaking points. Not "this seems risky," but the actual threshold: the collateral ratio at which a position becomes undercollateralized, the withdrawal share that empties a liquidity pool.
Sweep for recent changes. A protocol that looked solid six months ago can have broken three days ago, a depeg, an exploit, a departed auditor. Findings are dated, not just stated.
Connect the findings into one coherent thesis, rather than reporting a list of disconnected observations under separate headers.
It's also built specifically for the things that actually cause losses, with dedicated modules for:
Yield & solvency, where the yield actually comes from, and what happens under a depeg, a bank run, or a counterparty failure.
Team & track record, identity, pseudonymity handled correctly (verified without unmasking), and prior projects under the same identity.
Economic model, token design, incentive structure, and whether the mechanism is self-reinforcing or self-destructive under stress.
Three depths, your call
Not every question needs an institutional memo, and not every allocation decision should be made on a TL;DR. TeloSkills offers three depths, and you choose:
Quick Scan, a short go / set-aside verdict with the top reasons. A filter, not a memo.
Standard Brief, a structured report across the core areas. Enough for a real decision.
Full DD Report, an institutional-grade memo: every track, an adversarial pass, a visual risk matrix, and (on request) a branded PDF. A serious place to start, not a final verdict.
Depth changes the breadth of the analysis. It never changes the rigor of verification: a Quick Scan is shorter, not less honest about its sources.
It checks itself, too
One rule we built in deliberately: TeloSkills actively checks whether Telos Consilium itself has any disclosed or discoverable relationship, investment, incubation, advisory, or otherwise, with the project under review, and surfaces it near the top of the report if so.
Neutrality isn't a claim we make about the tool. It's a rule the tool applies to us, too.
The honest part, and it's the whole point
TeloSkills is a rigorous first pass. It is not a substitute for hands-on, experienced on-chain analysis, and it doesn't pretend to be. Its findings reflect only the data it could actually reach in a given session, the more you give it (API keys, an RPC endpoint, internal docs, audits under NDA where you're authorized to share them), the closer the output gets to reality. Without that, it works from public sources and is upfront about what that leaves unresolved.
That's by design. A tool that quietly papers over its own blind spots is more dangerous than one that names them.
Try it
TeloSkills is free and open. To use it:
Download the skill from the repository below.
In Claude, enable Code execution and file creation (Settings → Capabilities).
Go to Customize → Skills → Create skill, and upload it.
Ask Claude to run a due diligence on any token, protocol, or chain.
Add it to Claude: github.com/Telos-Consilium/TeloSkills
On-chain yield and risk is what we do at Telos Consilium, every day, for real capital. We built TeloSkills to raise the floor on how due diligence gets done across the industry, and to give anyone a genuinely rigorous place to start.
Need the full, hands-on version? Schedule a consultation with Telos Consilium →
Related articles
