The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the Bash tool to run a "safe representative path" from the project to verify that traces are being generated. While the instructions include explicit warnings to avoid production or irreversible actions, the automated execution of project code remains a security consideration.
[EXTERNAL_DOWNLOADS]: The skill uses package managers such as pip, npm, and yarn to install the opik library and its integrations. These are official packages provided by the skill author.
[INDIRECT_PROMPT_INJECTION]: The skill reads and analyzes various project files (source code, manifests, and configuration files) to detect frameworks and execution paths. This process is susceptible to instructions embedded within these files that could attempt to influence the agent's logic or tool usage.
Ingestion points: Project source files (*.py, *.ts), package.json, pyproject.toml, and .env files (SKILL.md).
Boundary markers: None identified.
Capability inventory: Read, Edit, Write, Bash (SKILL.md).
Sanitization: None identified.
[CREDENTIALS_UNSAFE]: The skill accesses sensitive information including OPIK_API_KEY, OPENAI_API_KEY, and ~/.opik.config to configure the observability backend. This is consistent with the skill's primary purpose of setting up tracing services.

opik-instrument