The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the ctxora command-line interface to perform repository indexing, querying, and instruction generation.
[REMOTE_CODE_EXECUTION]: The skill uses npx to download and execute the ctxora package directly from the npm registry.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted repository data to generate guidance and rules for AI agents (such as Cursor rules or Copilot instructions), which could allow malicious content in a repository to influence agent behavior.
Ingestion points: The workspace indexing and query functions in SKILL.md ingest arbitrary repository content.
Boundary markers: The skill provides a warning to treat returned content as "untrusted evidence" and to inspect cited files before editing.
Capability inventory: The skill has the capability to write new configuration and instruction files (.md, .cursorrules) to the local workspace.
Sanitization: No specific sanitization or filtering of the repository data is described before it is used to generate agent instructions.

ctxora-repository-context