The Agent Skills Directory

[DYNAMIC_EXECUTION]: The scripts/diagnose utility employs the eval command to execute arbitrary shell strings provided as arguments for its layers and test-isolation commands. While intended to support pipes and redirects, this pattern allows for unrestricted shell execution.
Evidence: eval "$cmd" and eval "$test_cmd" in scripts/diagnose.
[COMMAND_EXECUTION]: The find-polluter.sh script executes npm test on files discovered via a user-defined search pattern. This could lead to the execution of malicious code if the search path or filenames are manipulated.
Evidence: npm test "$TEST_FILE" in find-polluter.sh.
[INDIRECT_PROMPT_INJECTION]: The systematic debugging process requires the agent to ingest and analyze external data such as error messages, stack traces, and logs. A malicious instruction embedded in these sources could influence agent behavior.
Ingestion points: Error messages and logs analyzed in Phase 1 as described in SKILL.md and references/four-phases-detailed.md.
Boundary markers: None identified in the instructions for handling external data.
Capability inventory: Arbitrary shell execution via the diagnose script and test runner execution via find-polluter.sh.
Sanitization: No instructions provided for sanitizing or escaping content from logs or stack traces before processing.
[DATA_EXFILTRATION]: The CREATION-LOG.md file exposes a local file system path that reveals a specific local username.
Evidence: /Users/jesse/.claude/CLAUDE.md in CREATION-LOG.md.

systematic-debugging