The Agent Skills Directory

[PROMPT_INJECTION]: The skill analyzes terminal output and logs from previous executions. This creates an indirect prompt injection surface where malicious content in the logs could attempt to influence the agent's diagnostics.
Ingestion points: Terminal output, dynobox-transcript.log, and dynobox-chat-history.jsonl (referenced in SKILL.md).
Boundary markers: None present.
Capability inventory: Shell execution (npx dynobox), file modification (write_file, edit_file), and directory listing (found in SKILL.md and dyno/dyno-run-debug.dyno.mjs).
Sanitization: None detected.
Mitigation: The skill's workflow strictly requires user approval for all changes and mandatory validation using npx dynobox validate and node --check.
[COMMAND_EXECUTION]: The skill utilizes several standard command-line tools including npx dynobox, node, jq, cat, and ls. These are used for debugging, parsing logs, and verifying the syntax and schema of generated files. Access to these tools is explicitly scoped in the configuration files (.claude/settings.json and .codex/default.rules).
[EXTERNAL_DOWNLOADS]: The skill uses npx to invoke the dynobox CLI tool. This is a standard mechanism for executing Node.js-based developer tools and is consistent with the skill's purpose and the author's identity.

dyno-run-debug