The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using npx and bunx to run the @agent-native/core CLI utility. These commands are used to scan git diffs, generate recap prompts, and serve local plan previews.
[EXTERNAL_DOWNLOADS]: The skill downloads the @agent-native/core package from the npm registry as a dependency for its core functionality.
[DATA_EXFILTRATION]: Summarized code changes and UI mockups are uploaded to the Agent-Native Plans service at https://plan.agent-native.com. This is the intended purpose of the skill to provide a hosted, interactive review interface.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from git diffs and PR metadata to generate summaries. \n
Ingestion points: Reads data from git diffs, branch history, and conversation context (referenced in references/agent-native-recap.md). \n
Boundary markers: None explicitly defined for input data. \n
Capability inventory: Executes shell commands (npx), writes local files (MDX), and performs network requests to the plan service. \n
Sanitization: Instructs the agent to redact secrets using patterns like sk-••• and <redacted>, and strictly grounds all inferences in changed lines to prevent invention of recap details.

visual-recap