The Agent Skills Directory

[DATA_EXFILTRATION]: The skill accesses sensitive local session logs from ~/.claude/projects/ and ~/.codex/sessions/. The included helper script scripts/mo-debug.mjs incorporates security best practices, including verifying that logs are owned by the current user and are not symbolic links. It also uses extensive regular expressions to redact API keys, private keys, and absolute system paths before the agent processes the content.
[COMMAND_EXECUTION]: The helper script scripts/mo-debug.mjs executes the git command via child_process.spawnSync to perform version attribution against a local Git checkout if provided by the user.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from session history files which could potentially contain malicious instructions.
Ingestion points: Reads Claude and Codex session logs in JSONL format from the user's home directory.
Boundary markers: The helper script provides some structure in its output, but excerpts from logs are included without formal delimiters or isolation markers.
Capability inventory: The skill can execute git via spawnSync and write report files to the local file system.
Sanitization: The script performs comprehensive regex-based redaction of known credential formats and system paths.

mo-debug