The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is centered around executing shell commands to build feedback loops and verify bugs.
Mentions the use of git bisect run, curl, and various test runners (unit, e2e, integration) in SKILL.md (Phase 1).
Utilizes a helper script scripts/hitl-loop.template.sh to facilitate human-in-the-loop debugging.
[EXTERNAL_DOWNLOADS]: The skill interacts with network resources and external data files.
Instructions include using curl against development servers for reproduction (Phase 1).
The skill processes external artifacts such as HAR files, log dumps, and captured network traces.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface due to its processing of untrusted external data.
Ingestion points: The skill reads external data including logs, traces, and HAR files which could contain embedded instructions (Phase 1).
Boundary markers: No explicit delimiters or warnings to ignore instructions within ingested data are defined.
Capability inventory: The agent has extensive shell execution capabilities and file system access within the project scope (Phase 1, Phase 5).
Sanitization: No sanitization steps are specified for data before it is processed or used in terminal commands.

vibe-debug