The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use standard package managers (npm, pip, go get, etc.) to install OpenTelemetry and PostHog SDKs for the detected platform.
[EXTERNAL_DOWNLOADS]: It references and downloads well-known OpenTelemetry packages from official public registries (NPM, PyPI, Maven, etc.) and official PostHog SDKs.
[DATA_EXFILTRATION]: The skill is designed to configure the application to send logs to PostHog's official ingestion endpoints (e.g., us.i.posthog.com or eu.i.posthog.com). This is the intended and transparent primary function of the skill.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes codebase files like package.json, go.mod, and .env to determine the project's framework and current logging setup. While this establishes an ingestion surface for data from the user's project, the instructions are highly specific to PostHog instrumentation and do not present a high risk of bypassing agent constraints.
Ingestion points: Project configuration files (package.json, requirements.txt, .env, etc.) and existing source code files scanned during Step 1.
Boundary markers: The skill does not explicitly define delimiters for processed data, but it uses targeted extraction of dependency names and environment variable keys.
Capability inventory: File read/write, package installation, and network configuration for OTLP exporters.
Sanitization: The skill emphasizes using environment variables instead of hardcoding, which is a standard security practice.

instrument-logs