The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to interact with the iOS development environment.
Evidence: Use of xcodebuild, xcrun simctl, xcrun xcresulttool, and xcsym (referenced in axiom-tools) to run tests and extract diagnostic data in SKILL.md.
Context: These are standard operations for an automated test debugging agent and are constrained to the local development environment.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by ingesting and acting upon untrusted data generated during test execution.
Ingestion points: The agent reads content from /tmp/xcodebuild-debug.log, console.log, test-results.txt, and failure screenshots (multimodal analysis) as described in SKILL.md.
Boundary markers: There are no explicit boundary markers or instructions to treat log output as untrusted data within the analysis phases.
Capability inventory: The agent has the capability to execute shell commands (xcodebuild) and modify source code via the Edit tool based on the analysis of these logs.
Sanitization: The instructions do not specify any sanitization, filtering, or escaping of the content extracted from logs before it is processed by the LLM, creating a path where malicious log content could influence the agent's code modifications.

axiom-debug-tests