The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to download and execute the Dagger CLI installation script from dl.dagger.io, which is the official distribution domain for the project.\n- [COMMAND_EXECUTION]: The maintenance script scripts/check-updates.py uses subprocess.run to execute the pip command to verify the currently installed version of the dagger-io package against the latest version on PyPI.\n- [INDIRECT_PROMPT_INJECTION]: The references/09-llm-integration.md file documents how to build AI agents that ingest source code and use Dagger functions as tools, creating a documented attack surface for indirect prompt injection.\n
Ingestion points: The LLM environment can ingest directories containing untrusted project source code (references/09-llm-integration.md).\n
Boundary markers: Examples do not explicitly demonstrate prompt delimiters or injection-prevention instructions.\n
Capability inventory: The LLM can invoke Dagger functions that include file operations, network requests, and container execution.\n
Sanitization: No specific sanitization of ingested code is described in the integration examples.\n- [CREDENTIALS_UNSAFE]: The documentation in references/12-common-patterns.md contains hardcoded database connection strings used as illustrative placeholders for local testing services (e.g., postgresql://postgres:test@db:5432/testdb).

dagger