Advanced Evaluation
Design LLM-as-judge rubrics, direct scores and pairwise comparisons. Use for judge models, evals, position bias and calibration against human labels.
4,691,852 AI resources indexedWorking toward the world’s largest AI resource index.
Protocol and DeFi risk evaluation covering hack history, oracle dependencies, treasury health, TVL concentration, and yield sustainability. Use when the user asks "is X safe", "how risky is", protocol security, risk analysis, or wants to evaluate risk before investing or depositing funds.
--- name: risk-assessment-c9cb1342425d description: > Protocol and DeFi risk evaluation covering hack history, oracle dependencies, treasury health, TVL concentration, and yield sustainability. Use when the user asks "is X safe", "how risky is", protocol security, risk analysis, or wants to evaluate risk before investing or depositing funds. --- # Risk Assessment Evaluate the risk profile of a DeFi protocol by examining its security history, oracle dependencies, treasury, fundamentals, and yield levels. ## Workflow ### Step 1 - Resolve the protocol entity ``` defillama:resolve_entity entity_type: "protocol" name: "<user-provided name>" ``` ### Step 2 - Hack history Check whether the protocol has been exploited before. ``` defillama:get_events protocol: "<slug>" event_type: "hacks" ``` Any past hacks are a significant risk signal. Note the date, amount lost, and whether funds were recovered. ### Step 3 - Oracle dependencies Identify which oracle the protocol relies on and how much value it secures. ``` defillama:get_oracle_metrics ``` Filter results for the protocol's oracle. Oracle metrics automatically filter to `tvl_component = 'base'` and aggregate with SUM/GROUP BY to exclude double-counted TVL. A protocol using a small or unproven oracle has higher risk than one using Chainlink or Pyth. ### Step 4 - Treasury health Assess the protocol's financial reserves. ``` defillama:get_treasury treasury: "<slug>" ``` A healthy treasury provides a safety net for bug bounties, insurance, and continued development. ### Step 5 - Protocol fundamentals Check TVL, revenue, and trends to evaluate sustainability. ``` defillama:get_protocol_metrics protocol: "<slug>" ``` Key signals: Is TVL growing or declining? Is the protocol generating real revenue? ### Step 6 - Yield analysis Examine pool APYs for sustainability red flags. ``` defillama:get_yield_pools protocol: "<slug>" ``` ## Risk Signals Evaluate each factor and assign a risk level: | Signal | Risk Level | Explanation | |--------|-----------|-------------| | Recent hack (< 1 year) | HIGH | Protocol was recently exploited | | Past hack (> 1 year), no recurrence | MODERATE | Was exploited but has since hardened | | No hack history | LOW | No known exploits | | No oracle or small oracle TVS | MODERATE | Oracle risk, potential manipulation | | Uses Chainlink/Pyth with high TVS | LOW | Battle-tested oracle infrastructure | | Treasury < $1M | HIGH | No meaningful safety net | | Treasury > $10M | LOW | Strong financial reserves | | APY > 100% from rewards only | HIGH | Likely unsustainable, token emissions | | APY from real yield (fees) | LOW | Sustainable yield source | | TVL declining > 20% in 30d | MODERATE | Users are leaving | | Revenue declining with TVL | HIGH | Fundamentals deteriorating | ## Output Format Present the report with these sections: 1. **Risk Summary** - Overall risk rating (LOW / MODERATE / HIGH) with a one-sentence justification. 2. **Security History** - Hack incidents, amounts, and outcomes. 3. **Oracle Risk** - Which oracle is used, TVS secured, reliability. 4. **Treasury Health** - Total treasury value, composition, runway. 5. **Fundamental Health** - TVL trend, revenue, user activity direction. 6. **Yield Sustainability** - Are yields from real fees or emissions? 7. **Risk Factors** - Bullet list of all identified risk signals with their severity level. ## Tips - No data from `get_events` for hacks is a positive signal, not missing data. - A protocol with high TVL but zero revenue may be subsidizing usage with token emissions - flag this. - Compare treasury value to TVL: treasury < 1% of TVL means limited ability to cover losses. - Multiple past hacks on the same protocol is a stronger negative signal than a single incident.--- name: risk-assessment-c9cb1342425d description: > Protocol and DeFi risk evaluation covering hack history, oracle dependencies, treasury health, TVL concentration, and yield sustainability. Use when the user asks "is X safe", "how risky is", protocol security, risk analysis, or wants to evaluate risk before investing or depositing funds. --- # Risk Assessment Evaluate the risk profile of a DeFi protocol by examining its security history, oracle dependencies, treasury, fundamentals, and yield levels. ## Workflow ### Step 1 - Resolve the protocol entity ``` defillama:resolve_entity entity_type: "protocol" name: "<user-provided name>" ``` ### Step 2 - Hack history Check whether the protocol has been exploited before. ``` defillama:get_events protocol: "<slug>" event_type: "hacks" ``` Any past hacks are a significant risk signal. Note the date, amount lost, and whether funds were recovered. ### Step 3 - Oracle dependencies Identify which oracle the protocol relies on and how much value it secures. ``` defillama:get_oracle_metrics ``` Filter results for the protocol's oracle. Oracle metrics automatically filter to `tvl_component = 'base'` and aggregate with SUM/GROUP BY to exclude double-counted TVL. A protocol using a small or unproven oracle has higher risk than one using Chainlink or Pyth. ### Step 4 - Treasury health Assess the protocol's financial reserves. ``` defillama:get_treasury treasury: "<slug>" ``` A healthy treasury provides a safety net for bug bounties, insurance, and continued development. ### Step 5 - Protocol fundamentals Check TVL, revenue, and trends to evaluate sustainability. ``` defillama:get_protocol_metrics protocol: "<slug>" ``` Key signals: Is TVL growing or declining? Is the protocol generating real revenue? ### Step 6 - Yield analysis Examine pool APYs for sustainability red flags. ``` defillama:get_yield_pools protocol: "<slug>" ``` ## Risk Signals Evaluate each factor and assign a risk level: | Signal | Risk Level | Explanation | |--------|-----------|-------------| | Recent hack (< 1 year) | HIGH | Protocol was recently exploited | | Past hack (> 1 year), no recurrence | MODERATE | Was exploited but has since hardened | | No hack history | LOW | No known exploits | | No oracle or small oracle TVS | MODERATE | Oracle risk, potential manipulation | | Uses Chainlink/Pyth with high TVS | LOW | Battle-tested oracle infrastructure | | Treasury < $1M | HIGH | No meaningful safety net | | Treasury > $10M | LOW | Strong financial reserves | | APY > 100% from rewards only | HIGH | Likely unsustainable, token emissions | | APY from real yield (fees) | LOW | Sustainable yield source | | TVL declining > 20% in 30d | MODERATE | Users are leaving | | Revenue declining with TVL | HIGH | Fundamentals deteriorating | ## Output Format Present the report with these sections: 1. **Risk Summary** - Overall risk rating (LOW / MODERATE / HIGH) with a one-sentence justification. 2. **Security History** - Hack incidents, amounts, and outcomes. 3. **Oracle Risk** - Which oracle is used, TVS secured, reliability. 4. **Treasury Health** - Total treasury value, composition, runway. 5. **Fundamental Health** - TVL trend, revenue, user activity direction. 6. **Yield Sustainability** - Are yields from real fees or emissions? 7. **Risk Factors** - Bullet list of all identified risk signals with their severity level. ## Tips - No data from `get_events` for hacks is a positive signal, not missing data. - A protocol with high TVL but zero revenue may be subsidizing usage with token emissions - flag this. - Compare treasury value to TVL: treasury < 1% of TVL means limited ability to cover losses. - Multiple past hacks on the same protocol is a stronger negative signal than a single incident.
Imported from the listed AgentSkill.sh category. Complete supplied package files and notices are retained. Required external tools and credentials are not installed. No bundled code was executed during import. Individual agent task behavior has not been tested.