Finance Search Agent Strategy

v0.1.0

AI agent for finance search agent strategy tasks

0· 556·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
Purpose & Capability
The name and description promise a finance research/search strategy agent. The skill is instruction-only, requests no binaries, env vars, or installs, and the SKILL.md contains step-by-step research rules and tool-call priorities — all coherent with a search-agent strategy role.
Instruction Scope
The instructions define a strict research workflow (data collection then synthesis), limits (word-count caps, deadlock handling), and explicit tool-call ordering (prefer discover_tools and execute_search_tool). They do not ask the agent to read system files or unrelated credentials. Note: the skill instructs the agent to treat external tool results as the authoritative source even when they conflict with internal knowledge — this is expected for a real-time search strategy but increases risk of blindly accepting adversarial/malicious external content.
Install Mechanism
No install spec is present. This is instruction-only, which minimizes filesystem footprint and matches the declared manifest.
Credentials
The skill requires no environment variables, credentials, or config paths. There is no disproportionate request for secrets or unrelated access.
Persistence & Privilege
always is false and model invocation is allowed (the platform default). The skill does not request persistent or elevated privileges or attempt to modify other skills or system configurations.
Assessment
This skill is primarily a set of agent instructions for conducting finance research and appears coherent. Before installing, consider: 1) the agent is instructed to treat external tool outputs as the single source of truth — review which 'available_tools' the agent will have (ensure they are trustworthy and sandboxed) because malicious or compromised data sources could mislead results; 2) the strict automatic behaviors (word limits, automatic blocking after two failed attempts, tool ordering) can be useful but may also prematurely stop useful investigation — test with representative queries; 3) because the skill will call runtime tools, ensure the runtime environment restricts network and credential access appropriately (the skill itself requests none, but tools it calls might); and 4) if you need safer defaults (e.g., require corroboration before overriding internal knowledge), modify the SKILL.md accordingly. If you want higher assurance, request visibility on the agent's available_tools and perform initial tests in a sandboxed environment.

Like a lobster shell, security has layers — review code before you run it.

latestvk97810nh00cbzrpd3j0rtsz6yn819wcf

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Comments