The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill is designed to migrate MCP server configurations, which typically include sensitive environment variables like API keys and access tokens. These values are processed in plaintext and copied directly into the new configuration.
[COMMAND_EXECUTION]: The bundled script scripts/convert_claude_to_codex.py executes shell commands via subprocess.run to interact with the codex CLI for listing, removing, and adding configuration entries.
[DATA_EXFILTRATION]: The skill reads from ~/.claude.json, a sensitive local file containing configuration data and potential credentials. While it does not send this data to a remote URL, the content is read into the agent's context and can be displayed to the user or exported as TOML.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted JSON data from the file system, which serves as an attack surface if a malicious configuration file is present.
Ingestion points: scripts/convert_claude_to_codex.py reads ~/.claude.json using pathlib.Path.read_text().
Boundary markers: No explicit delimiters or safety warnings are applied to the ingested JSON data.
Capability inventory: The skill possesses file read access and command execution capabilities (subprocess.run).
Sanitization: The script uses list-based arguments for subprocess.run rather than shell strings and utilizes shlex.quote for display purposes, mitigating basic command injection.

mcp-convert