The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill ingests untrusted data from test logs and error messages which could contain malicious instructions designed to hijack the agent's workflow.
Ingestion points: Data entering the agent via listen_to_events and helpmetest_status (SKILL.md).
Boundary markers: None present to distinguish between test logs and agent instructions.
Capability inventory: helpmetest_run_interactive_command (executes shell commands), helpmetest_upsert_test (modifies test files), and helpmetest_run_test (executes code).
Sanitization: No evidence of validation or sanitization of ingested test data before it is interpolated into prompts or used as input for command-execution tools.
[COMMAND_EXECUTION]: Dynamic Command and Code Execution. The skill uses helpmetest_run_interactive_command to interactively investigate failures and helpmetest_upsert_test to modify test files. While this is the primary purpose of a test-healing agent, it provides a high-privilege execution environment that could be exploited if the agent is influenced by malicious test output.

helpmetest-self-heal