The Agent Skills Directory

[SAFE]: The skill is authored by Amplitude and exclusively utilizes official Amplitude tools for data retrieval and analysis. No unauthorized network access, credential exposure, or malicious code execution patterns were detected.- [INDIRECT_PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection by processing untrusted application data such as error messages, stack traces, and UI element text. This is consistent with its primary purpose of error diagnosis. The risk is minimized as the available tools are restricted to data querying and analysis within the vendor's ecosystem, lacking high-privilege capabilities like shell access or file writing.
Ingestion points: Processes [Amplitude] Error Logged messages/traces, [Amplitude] Error Click text, and AI-generated insights via Amplitude:use_amplitude_ai_feedback as described in Step 2 and Step 5 of SKILL.md.
Boundary markers: Absent; the skill relies on the agent's internal handling of structured query results.
Capability inventory: Tools are limited to Amplitude platform functions (query_amplitude_data, get_amp_session_replay_info, use_amp_flags, use_amp_experiments) which are used for analytical reporting.
Sanitization: The instructions do not specify explicit sanitization for ingested telemetry strings before they are included in the generated triage report.

diagnose-errors