The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data, such as bug reports and code diffs, which can serve as vectors for indirect prompt injection. If an attacker provides a malicious diff or bug report containing instructions designed to influence the agent's behavior during the test generation or execution phase, the agent might inadvertently execute unintended actions.
Ingestion points: The skill accepts "Bug report, code diff, API change, route, component, feature flag, or sandbox path" as inputs in SKILL.md.
Boundary markers: Absent. The instructions do not define clear delimiters or instruct the agent to ignore instructions that might be embedded within the provided code or bug reports.
Capability inventory: The skill workflow in SKILL.md (Section 5) involves executing shell commands such as bun test and bun run typecheck to verify the code.
Sanitization: Absent. There are no explicit steps to sanitize or validate the structure of the input data before it is used to generate or execute tests.
[COMMAND_EXECUTION]: The skill workflow explicitly instructs the agent to execute shell commands to perform verification tasks.
Evidence: The verification section in SKILL.md includes commands such as bun test path/to/regression.test.ts, bun test, and bun run typecheck.

ai-regression-testing