Is intellectiaai/intellectia-stock-screener safe?
https://github.com/openclaw/skills/tree/main/skills/intellectiaai/intellectia-stock-screener
The intellectia-stock-screener skill is a clean API documentation stub that instructs an agent to make a single unauthenticated GET request to the Intellectia stock screening API. No prompt injection, sensitive file access, hidden code, git hooks, or suspicious network behavior was observed during installation. The primary residual risk is that the API returns AI-generated financial predictions ('probability', 'profit') from an unaudited third-party model, which could mislead users if taken as authoritative investment advice.
Category Scores
Findings (4)
LOW Third-party financial predictions without quality guarantees -15 ▶
The Intellectia API returns 'probability' (model confidence) and 'profit' (expected return) fields generated by a proprietary AI model. There is no disclosure of the model's accuracy, backtesting, or conflict-of-interest policy. An agent presenting these values as authoritative could harm users who act on them.
LOW Runtime pip dependency installation -7 ▶
The skill's metadata declares a pip install of the 'requests' package at agent setup time. While 'requests' is a widely trusted package, any pip install at runtime introduces a transitive dependency chain. A supply-chain compromise of 'requests' on PyPI would affect this skill.
INFO Outbound GET to third-party API -7 ▶
The skill instructs the agent to issue an HTTP GET to api.intellectia.ai. This is unauthenticated and does not transmit local data, but it does expose the agent's public IP and usage patterns to Intellectia.
INFO Canary file reads by audit harness (not skill) -8 ▶
auditd PATH records show reads of .env, id_rsa, .aws/credentials, .npmrc, .docker/config.json, and gcloud credentials. Timestamps (1771919975 pre-clone, 1771919993 post-install) and process context indicate these reads are from the oathe monitoring infrastructure performing baseline hashing, not from any skill-installed code.