The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains instructions that attempt to control the AI's output style and persona, specifically requiring the response to start with a "🧪" emoji and for the AI to "Respond terse like smart caveman" except for security warnings.
[COMMAND_EXECUTION]: The skill executes various testing and linting tools via shell commands, including npm test, pytest, go test, and make lint. These are used to empirically detect flakiness and verify fixes.
[DATA_EXFILTRATION]: The skill accesses the user's home directory to read configuration from ~/.absolute/config.json. While used for codebase conventions, accessing the home directory is a sensitive file operation.
[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection by ingesting and acting upon untrusted data from external sources.
Ingestion points: Test execution outputs, CI history logs, and repository configuration files (.absolute.config.json) in references/health-engine.md (Step 1 & 2).
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from obeying instructions that might be embedded within test failures or log files.
Capability inventory: The skill can modify project source code (FIX step) and execute arbitrary project scripts (VERIFY step) through subprocesses.
Sanitization: No evidence of sanitization or filtering is present for the data read from external tools or history logs before it is used to generate code fixes.

absolute-deflake