The Agent Skills Directory

[SAFE]: The skill defines a professional diagnostic workflow for Android developers, focusing on lifecycle management, threading safety, and state validation.
[COMMAND_EXECUTION]: The skill recommends using codegraph, rg, and find to locate suspect code and explore call hierarchies. These are standard local search tools used within the context of the agent's environment to assist in bug investigation.
[INDIRECT_PROMPT_INJECTION]: The skill processes external, untrusted data including crash logs, stack traces, and reproduction steps.
Ingestion points: Crash logs and reproduction steps are provided as input to the workflow (SKILL.md).
Boundary markers: The skill does not explicitly define markers to separate untrusted log data from agent instructions.
Capability inventory: The available tools are restricted to code exploration and search; there are no network-write or file-modification capabilities specified.
Sanitization: No explicit sanitization or filtering is performed on the ingested logs. However, the risk is negligible given the diagnostic purpose and limited tool capabilities.

android-crash-analyzer