The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill accesses local configuration at ~/.opik.config and the OPIK_API_KEY environment variable to authenticate with the Opik service.
[DYNAMIC_EXECUTION]: The skill instructs the agent to dynamically generate and execute Python code using the opik SDK as a fallback when the hosted MCP is unavailable.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute diagnostic scripts and repository search commands such as Grep and Glob during root-cause analysis.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill fetches trace content and span data, which may contain arbitrary external LLM inputs and outputs, through client.get_trace_content() and client.search_spans() as described in SKILL.md.
Boundary markers: The instructions do not provide explicit delimiters or prompts to the agent to ignore instructions that might be embedded within the external trace data.
Capability inventory: The skill uses its reasoning capabilities alongside Read, Grep, Glob, and Bash tools to investigate the repository based on findings from the ingested trace data.
Sanitization: There is no mention of sanitizing, escaping, or validating the external trace content before the agent processes it for analysis.

opik-explain