Pangolinfo Amazon Scraper
Analysis
This appears to be a legitimate Pangolin Amazon scraper, but it asks the agent to handle and permanently cache Pangolin credentials in ways users should review carefully.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.
Write it directly to the cache file — avoids shell history entirely: ```bash echo "<api_key>" > ~/.pangolin_api_key ``` ... This avoids passwords appearing in shell history
The documentation presents literal-secret shell commands as fully history-safe; depending on how the commands are run, secrets may still be captured in shell history, terminal logs, agent transcripts, or tool logs.
Checks whether tool use, credentials, dependencies, identity, account access, or inter-agent boundaries are broader than the stated purpose.
When the user provides their credentials, **you (the AI agent) should configure them securely**. The script will automatically cache the API key at `~/.pangolin_api_key` for all future calls.
The skill asks the agent to handle Pangolin account credentials and create a persistent reusable API-key cache; this is sensitive account access, especially because the registry metadata declares no primary credential, env vars, or config paths.