Agent Dev Workflow
Security checks across static analysis, malware telemetry, and agentic risk
Overview
The skill is a coherent coding workflow, but it asks to run coding agents with permissions bypassed and to create persistent project context, so it deserves careful review before use.
Install only if you are comfortable letting coding agents operate on your project, ideally inside a clean branch or sandbox. Pay special attention to the `--permission-mode bypassPermissions` requirement, review any generated llmdoc/CLAUDE.md files, and inspect all final code changes before merging.
Static analysis
No static analysis findings were reported for this release.
VirusTotal
VirusTotal findings are pending for this skill version.
Risk analysis
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
A dispatched coding agent could modify project files or run development commands with fewer interactive permission prompts than users may expect.
The workflow is built around dispatching coding agents, and this prerequisite explicitly uses a mode that bypasses normal Claude Code permission checks. That is high-impact local tool authority and is only partly contained by the documented planning gates.
Claude Code (`claude` CLI) installed with `--permission-mode bypassPermissions --print`
Use this only in a clean repository, preferably on a separate branch or sandbox. Avoid permission-bypass mode when possible, and review the task plan, generated files, commands, and final diff before accepting changes.
The actual behavior will partly depend on external tools and plugins outside this skill’s reviewed files.
The skill depends on an external plugin and sub-agents that are not included in the reviewed artifact set. This is purpose-aligned, but users must separately trust and maintain those dependencies.
[cc-plugin](https://github.com/TokenRollAI/cc-plugin) installed in Claude Code (provides llmdoc read/write, investigator/scout/recorder sub-agents)
Install dependencies only from trusted sources, review their documentation and permissions, and prefer pinned or known-good versions.
Generated project documentation may persist in the repository and be reused by future agent runs.
The workflow creates and reuses persistent project documentation as agent context. That is appropriate for a development workflow, but stored context can contain sensitive project details or later influence agent behavior.
`llmdoc/` exists? Yes → read `llmdoc/index.md` + overview files ... No → dispatch Claude Code with `/tr:initDoc` to generate it first
Review generated llmdoc and CLAUDE.md files, exclude secrets or private data, and keep generated context under normal code-review/version-control practices.