01Safety
30.0 / 30
What changed in the harness
Selection accuracy 100→100, token cost up 4%, unconfirmed writes 100%→100%.
Category breakdown
Where the score comes from.
Earned points across the four signals Gradable measures. Safety and Legibility are scored out of 30; Economics and Discoverability are scored out of 20.
02Legibility
25.7 / 30
03Economics
19.4 / 20
04Discoverability
10.3 / 20
Highest-impact fix
Estimated gain +10 pointsMake target tools discoverable on the first call
Clarify tool names, decision boundaries, and required argument schemas so an agent can choose and construct the target call without exploratory steps.
Description evidence
Defects and rewrites.
7 defects found across the exposed tool descriptions. Suggested rewrites make purpose, inputs, boundaries, and returns easier for an agent to understand.
| Tool | Defect types | Suggested rewrite |
|---|---|---|
health_check |
name_restates_behavior no_return_description |
Verify the ForgeMesh Anomaly Tracker API is operational and measure its uptime; poll regularly to catch outages before users are affected. Returns the API's current health status and uptime. |
whale_alerts |
no_return_description |
Get recent whale movements, CEX inflows/outflows, bridge activity, and stablecoin mints/burns from monitored addresses (Binance, Coinbase, Kraken, OKX, Bybit, major bridges, Tether, Circle). Returns the detected alert events with involved addresses, direction, and timing. Costs $0.02 USDC on Base mainnet. |
address_scan |
no_return_description |
Scan any wallet address for anomalous transaction patterns. Classifies each transaction by checking counterparties against known CEX wallets, bridges, and stablecoin issuers. Returns the classification of each transaction and flags which are anomalous with the reason. Costs $0.03 USDC on Base mainnet. |
model_status |
no_return_description |
Get SequenceMiner model health and training stats per chain to confirm the detection models are trained and current. Returns the model health status, training sequence count, and last retrain time for each chain. Costs $0.01 USDC on Base mainnet. |
flight_scan |
no_return_description |
Sequence anomaly analysis for a named airspace region — detects unusual clustering of emergency squawks, rapid descents, or speed anomalies using the SequenceMiner engine. Returns a story label and anomaly score for the region along with the detected anomaly patterns. Costs $0.03 USDC on Base mainnet. |
trending_signal |
no_return_description |
GitHub repos with anomalous star velocity right now — early signals before mainstream discovery. Detects repos going viral (200+ stars/day), overnight explosions, fork surges, and issue floods. Useful for trend hunting, VC scouting, developer tool discovery. Returns a list of currently trending repos with their velocity metrics and detected anomaly signals. Costs $0.02 USDC on Base mainnet. |
claude_feature_watch |
no_return_description |
Watch Anthropic/Claude repos for recently merged features — SDK changes, Claude Code updates, new capabilities. Scans anthropic-sdk-python, anthropic-sdk-typescript, claude-code, courses, and anthropic-cookbook. Returns the recently merged feature items found across the monitored repos with details of what changed. Costs $0.02 USDC on Base mainnet. |
Selection evidence
Confusable tool pairs.
8 pairs where similar names or overlapping descriptions may send an agent toward the wrong tool.
| Tool A | Tool B | Confidence | Why they collide |
|---|---|---|---|
token_scan |
nft_scan |
high | Both are 'best-effort anomaly scan' tools that return 'source status plus coverage notes' and take a required 'token'/contract-address input; only 'ERC-20 token' vs 'NFT collection' differ, so a task like 'scan this token for wash-trading/mint patterns: 0xbc4ca0...' (an NFT contract) can route to the wrong member. |
address_scan |
defi_scan |
medium | defi_scan also accepts 'a 0x contract address' while address_scan accepts any address, so 'scan this contract address for unusual flow anomalies' is ambiguous between a wallet-pattern scan and a DeFi-protocol scan. |
address_scan |
nft_scan |
medium | Both take a single 0x address/contract as input and both detect pattern-based anomalies, so a task like 'scan 0xbc4ca0... for anomalous/wash-trading patterns' does not clearly disambiguate a wallet scan from an NFT-collection scan. |
anomaly_scan |
defi_scan |
medium | Both detect 'unusual flows', return a story label and anomaly score, and share chain/window inputs; 'look for unusual flow anomalies' is ambiguous between a chain-wide scan and a specific lending/DEX protocol scan. |
anomaly_scan |
flight_scan |
medium | Both describe 'sequence anomaly' analysis that returns anomaly scores, so a task phrased as 'run a sequence-anomaly scan for unusual patterns' is ambiguous between a blockchain chain and an airspace region without an explicit chain/region. |
anomaly_scan |
mempool_anomaly |
medium | Both detect whale-related anomalies and return a story label plus anomaly score; a vague task like 'check for whale anomaly signals' could pick the chain-wide scan instead of the real-time mempool tool (or vice versa), since 'whale' appears in both. |
anomaly_scan |
nft_scan |
medium | Anomaly_scan covers stablecoin mints/burns and nft_scan covers mint surges, both under 'pattern'/'mint' language; 'look for unusual mint patterns' with no qualifier is ambiguous between chain-wide stablecoin mints and NFT-collection mint surges. |
github_watch |
claude_feature_watch |
medium | claude_feature_watch targets the anthropic-sdk/claude-code repos that github_watch could also analyze, and both return story labels; 'watch the claude-code/anthropic-sdk repo for recent changes or activity' could be routed to either tool. |
Compare the field