live-mastracode-instrumentation
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to perform several shell-based operations to prepare the debugging environment. This includes installing dependencies with
pnpm install --frozen-lockfile, building the project usingpnpm build:mastracode, and searching files withrg(ripgrep). It also utilizestmuxto manage a headless terminal session for the TUI execution.- [DYNAMIC_EXECUTION]: The instructions guide the agent to generate and execute apython3script at runtime. The purpose of this script is to process and group timestamped JSONL logs generated during the TUI instrumentation phase for performance and race condition analysis.- [INDIRECT_PROMPT_INJECTION]: The skill establishes an ingestion surface for untrusted data by instructing the agent to read TUI output viatmux capture-pane. - Ingestion points: tmux capture buffer (reading TUI/CLI output in SKILL.md).
- Boundary markers: None explicitly defined for the capture-pane output reading.
- Capability inventory: Command execution (
pnpm,node,tmux), file system writes (appendFileSync), and dynamic script execution (python3). - Sanitization: The skill includes a security-conscious rule to log only metadata (timestamps, IDs, lengths) and explicitly forbids logging prompts, responses, or tool content in the debug logs.
Audit Metadata