The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to automate the onboarding and verification process:
Executes open, xdg-open, or start to launch the Superlog signup flow in the default browser.
Runs native project commands (e.g., npm run dev, pnpm build, uv run python) to verify that the instrumentation does not break the application.
Uses curl to trigger application routes and verify the flow of telemetry signals.
Executes claude mcp add to configure the Superlog MCP server for the agent.
[EXTERNAL_DOWNLOADS]: The skill installs necessary telemetry packages using system package managers:
Downloads official OpenTelemetry SDKs and instrumentations (e.g., @opentelemetry/sdk-node, opentelemetry-sdk).
Installs the vendor-provided @superlog/otel-helpers package.
[DATA_EXFILTRATION]: The skill performs network operations to transfer setup and telemetry data to the vendor's domain:
Sends signup intent metadata (token hash and prefix) to api.superlog.sh.
Transmits project telemetry (traces, logs, metrics) to intake.superlog.sh for verification.
Extracts and hardcodes repository metadata, including the VCS URL and commit SHAs, as resource attributes in the telemetry payload.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface due to its analysis of the user's codebase:
Ingestion points: The agent reads and maps all applications and services within the repository (orchestrated in SKILL.md).
Boundary markers: None. The skill processes code blocks directly to identify business operations for instrumentation.
Capability inventory: File system writes, shell command execution (build/deploy), and network access (defined in SKILL.md).
Sanitization: None. The skill assumes project source code is safe to process for identifying instrumentation targets.

superlog-onboard