The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest data from external session logs to generate realistic transcripts, creating a surface for indirect injection from previous interaction history.
Ingestion points: ~/.claude/projects/<project-slug>/*.jsonl (SKILL.md)
Boundary markers: Absent; the skill does not specify delimiters or warnings for the log content.
Capability inventory: The skill primarily uses PIL (Pillow) for image rendering and does not perform network operations or subprocess execution on the ingested data.
Sanitization: Absent; there is no instruction to validate or sanitize the transcript data before rendering.
[DATA_EXFILTRATION]: The instructions direct the agent to access specific local application paths to extract theme data and transcript history.
Evidence: Mentions file paths ~/.claude/projects/<project-slug>/*.jsonl and ~/.codex/config.toml in SKILL.md.
[EXTERNAL_DOWNLOADS]: The skill suggests installing the official @google/gemini-cli package via the NPM registry to verify the UI layout.
Evidence: npm i -g @google/gemini-cli in SKILL.md.

terminal-replica