Most external-command alerts are false positives caused by Markdown code delimiters and TypeScript syntax. The skill genuinely performs network access, handles private keys, can run a deployment command, and trusts remotely hosted configuration and source-policy data. Remote factory metadata can redirect a signed deployment unless the returned address and transaction are independently verified.
// Node: createAccount(process.env.PRIVATE_KEY). If PRIVATE_KEY is unset,
The Node example accesses PRIVATE_KEY to sign transactions. This is a legitimate use, but it is sensitive credential handling and requires explicit user control.
The Node example accesses PRIVATE_KEY to sign transactions. This is a legitimate use, but it is sensitive credential handling and requires explicit user control.
This deploys a **fresh factory contract** (writing the new address to `.env.local`), not a new oracl
The fork deployment flow writes a new factory address to .env.local. This alters a configuration file that can influence later deployments and should be reviewed by the user.
This deploys a **fresh factory contract** (writing the new address to `.env.local`), not a new oracl
The fork deployment flow writes a new factory address to .env.local. This alters a configuration file that can influence later deployments and should be reviewed by the user.
Remote Factory Pointer Can Redirect Signed Deployment
The workflow fetches a factory address from a remote metadata endpoint before deployment. If that endpoint is compromised, an agent could prepare a signed transaction for an attacker-selected contract unless the user reviews the address and call details.
The instructions explicitly trust a remotely fetched factoryAddress and then direct deployment to that factory. The source provides no integrity verification or required user confirmation of the returned address.
The skill treats a live external source catalog as authoritative for blocked and rerouted hosts. Changes to that catalog can alter source selection and market settlement behavior without local review.
The text explicitly requires use of a mutable remote catalog for source blocking and rerouting. This is a business-logic trust dependency rather than merely a hardcoded URL.
Capability review items (21)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
// Node: createAccount(process.env.PRIVATE_KEY). If PRIVATE_KEY is unset,
The documented deployment path reads PRIVATE_KEY from the environment to create a signing account or runs a command with credential variables. A compromised execution path could expose or misuse the key.
The documented deployment path reads PRIVATE_KEY from the environment to create a signing account or runs a command with credential variables. A compromised execution path could expose or misuse the key.
// Node: createAccount(process.env.PRIVATE_KEY). If PRIVATE_KEY is unset,
The documented deployment path reads PRIVATE_KEY from the environment to create a signing account or runs a command with credential variables. A compromised execution path could expose or misuse the key.
The documented deployment path reads PRIVATE_KEY from the environment to create a signing account or runs a command with credential variables. A compromised execution path could expose or misuse the key.
// Node: createAccount(process.env.PRIVATE_KEY). If PRIVATE_KEY is unset,
The documented deployment path reads PRIVATE_KEY from the environment to create a signing account or runs a command with credential variables. A compromised execution path could expose or misuse the key.
The documented deployment path reads PRIVATE_KEY from the environment to create a signing account or runs a command with credential variables. A compromised execution path could expose or misuse the key.
cd scripts && PRIVATE_KEY=<key> RPC_URL=<rpc> npm run deploy
The documented deployment path reads PRIVATE_KEY from the environment to create a signing account or runs a command with credential variables. A compromised execution path could expose or misuse the key.
- Use the `genlayer-js` SDK from another runtime via a Node subprocess, or
The skill explicitly directs agents to run a Node subprocess or a shell deployment command. Executing it can install or run project code and, for deployment, use supplied credentials.
The skill explicitly directs agents to run a Node subprocess or a shell deployment command. Executing it can install or run project code and, for deployment, use supplied credentials.
The skill explicitly directs agents to run a Node subprocess or a shell deployment command. Executing it can install or run project code and, for deployment, use supplied credentials.
This deploys a **fresh factory contract** (writing the new address to `.env.local`), not a new oracl
The skill explicitly directs agents to run a Node subprocess or a shell deployment command. Executing it can install or run project code and, for deployment, use supplied credentials.
GET https://intelligentoracle.com/oracle-meta.json
The skill instructs network access to this endpoint as part of configuration, RPC use, source selection, or freshness checks. These requests are functional but expand the trust boundary.
The skill instructs network access to this endpoint as part of configuration, RPC use, source selection, or freshness checks. These requests are functional but expand the trust boundary.
If unreachable, default to `rpcUrl: "https://studio.genlayer.com/api"` and ask the user to paste the
The skill instructs network access to this endpoint as part of configuration, RPC use, source selection, or freshness checks. These requests are functional but expand the trust boundary.
| `dataSourceDomains` | `string[]` | Either this OR `resolutionURLs` is populated, never both, never
The skill instructs network access to this endpoint as part of configuration, RPC use, source selection, or freshness checks. These requests are functional but expand the trust boundary.
- **Never invent a source.** If the user supplies a specific source domain or URL, use it. Otherwise
The skill instructs network access to this endpoint as part of configuration, RPC use, source selection, or freshness checks. These requests are functional but expand the trust boundary.
The live verified catalog lives at `https://gym.genlayer.foundation/api/benchmarks/sources-bench/sou
The skill instructs network access to this endpoint as part of configuration, RPC use, source selection, or freshness checks. These requests are functional but expand the trust boundary.
The skill instructs network access to this endpoint as part of configuration, RPC use, source selection, or freshness checks. These requests are functional but expand the trust boundary.
The RPC endpoint (`https://studio.genlayer.com/api`) speaks an Ethereum-style JSON-RPC superset. Rea
The skill instructs network access to this endpoint as part of configuration, RPC use, source selection, or freshness checks. These requests are functional but expand the trust boundary.
- `Factory address is not configured` β fetch `https://intelligentoracle.com/oracle-meta.json` or as
The skill instructs network access to this endpoint as part of configuration, RPC use, source selection, or freshness checks. These requests are functional but expand the trust boundary.
This skill is published at `https://intelligentoracle.com/skill.md`. If you suspect it's stale, refe
The skill instructs network access to this endpoint as part of configuration, RPC use, source selection, or freshness checks. These requests are functional but expand the trust boundary.
This deploys a **fresh factory contract** (writing the new address to `.env.local`), not a new oracl
The deploy shortcut is documented as writing a new factory address to .env.local. Modifying environment configuration files is a real sensitive filesystem side effect.
This deploys a **fresh factory contract** (writing the new address to `.env.local`), not a new oracl
The deploy shortcut is documented as writing a new factory address to .env.local. Modifying environment configuration files is a real sensitive filesystem side effect.
The skill directs agents to call create_new_prediction_market and resolve on a public RPC endpoint. These writes can deploy or resolve contracts and should require explicit approval before execution.
The workflow includes concrete writeContract calls for deployment and resolution. These are on-chain side effects, not passive documentation only.
The skill tells agents to refetch its canonical version from intelligentoracle.com if it seems stale. Loading mutable remote instructions after installation can bypass marketplace review.
The instruction directly points to a remote skill.md and invites refetching. That creates a clear post-review instruction update path.
The raw RPC section advises using the SDK from another runtime through a Node subprocess and describes shelling out to a Node script. This expands execution beyond reviewed text.
The text explicitly recommends a Node subprocess for agents outside JavaScript. The risk is contextual because no bundled script is present.
Capability review items (17)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
// Node: createAccount(process.env.PRIVATE_KEY). If PRIVATE_KEY is unset,
The deployment guidance references PRIVATE_KEY or RPC_URL values for account creation and deployment. These values can contain sensitive credentials and must not be read or logged without explicit user consent.
The deployment guidance references PRIVATE_KEY or RPC_URL values for account creation and deployment. These values can contain sensitive credentials and must not be read or logged without explicit user consent.
cd scripts && PRIVATE_KEY=<key> RPC_URL=<rpc> npm run deploy
The deployment guidance references PRIVATE_KEY or RPC_URL values for account creation and deployment. These values can contain sensitive credentials and must not be read or logged without explicit user consent.
- Use the `genlayer-js` SDK from another runtime via a Node subprocess, or
Although the scanner matched Markdown backticks, this line also advises using a Node subprocess from another runtime. That is real external execution guidance and should require explicit user approval.
This skill is published at `https://intelligentoracle.com/skill.md`. If you suspect it's stale, refe
The skill tells agents to refetch a mutable remote skill document if it seems stale. That is a real network dependency that can change behavior outside the audited file.
The hardcoded GenLayer Studio RPC endpoint is used for contract reads and writes. This is intentional outbound network access and can perform external side effects.
If unreachable, default to `rpcUrl: "https://studio.genlayer.com/api"` and ask the user to paste the
The hardcoded GenLayer Studio RPC endpoint is used for contract reads and writes. This is intentional outbound network access and can perform external side effects.
- **Never invent a source.** If the user supplies a specific source domain or URL, use it. Otherwise
The skill instructs agents to consult a live external catalog before choosing source domains. This is intentional outbound network access to mutable remote data.
The live verified catalog lives at `https://gym.genlayer.foundation/api/benchmarks/sources-bench/sou
The skill instructs agents to consult a live external catalog before choosing source domains. This is intentional outbound network access to mutable remote data.
The hardcoded GenLayer Studio RPC endpoint is used for contract reads and writes. This is intentional outbound network access and can perform external side effects.
The RPC endpoint (`https://studio.genlayer.com/api`) speaks an Ethereum-style JSON-RPC superset. Rea
The hardcoded GenLayer Studio RPC endpoint is used for contract reads and writes. This is intentional outbound network access and can perform external side effects.
// Node: createAccount(process.env.PRIVATE_KEY). If PRIVATE_KEY is unset,
The example accesses process.env.PRIVATE_KEY for account creation. Environment-variable access is real here, and the higher-risk private-key exposure is covered by the secret-key findings.
The example accesses process.env.PRIVATE_KEY for account creation. Environment-variable access is real here, and the higher-risk private-key exposure is covered by the secret-key findings.
// Node: createAccount(process.env.PRIVATE_KEY). If PRIVATE_KEY is unset,
The example accesses process.env.PRIVATE_KEY for account creation. Environment-variable access is real here, and the higher-risk private-key exposure is covered by the secret-key findings.
The example accesses process.env.PRIVATE_KEY for account creation. Environment-variable access is real here, and the higher-risk private-key exposure is covered by the secret-key findings.