The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands for project discovery and research.
Evidence: Uses cat, find, grep, git log, and ls to inspect the file system and project history.
Evidence: Utilizes python3 -c one-liners to parse and extract specific fields from package.json and tsconfig.json files.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface because it reads external project files and interpolates their content into an internal prompt used to generate plans.
Ingestion points: Reads content from README.md, package.json, CLAUDE.md, and arbitrary source files discovered during the research phase.
Boundary markers: While the skill uses structured prompt templates and optional XML tags for context, it lacks specific instructions for the AI to ignore instructions found within the ingested data.
Capability inventory: The skill can execute shell commands for discovery, read/write files (such as plan.md), and interact with the session workspace's SQL todos table.
Sanitization: No sanitization or validation is performed on the data read from files before it is processed by the AI to create the implementation plan.

deep-plan