The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands through the Bash tool to perform repository analysis, specifically using git log and git blame to research prior art. It also references the use of a test runner via Bash for mapping existing coverage.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from the local codebase and git history to inform its planning synthesis.
Ingestion points: Research data is gathered from codebase files, integration points, comments, and git history using Grep, Read, and Glob tools.
Boundary markers: The skill does not define specific delimiters or warnings to ignore embedded instructions when the research agents report findings back for synthesis.
Capability inventory: The skill can perform shell execution (Bash), file system discovery (Glob, Grep), and file reading (Read).
Sanitization: There is no explicit sanitization or filtering of the content read from the codebase before it is used to generate the execution plan.
[SAFE]: The skill implements a robust safety control by requiring a mandatory 'Approval Gate' (Phase 4), which prevents the agent from executing any proposed code changes without explicit user confirmation.

ultraplan