The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to wrap and execute the tiangong-lca command-line interface. The script run-lifecyclemodel-automated-builder.mjs constructs shell commands by appending user-supplied arguments (such as manifest paths and output directories) to canonical CLI commands like tiangong-lca lifecyclemodel auto-build.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from local process-build runs and manifest files which could theoretically contain malicious instructions designed to influence the assembly or validation logic.
Ingestion points: The manifest file specified via the --input, --manifest, or --request flags in run-lifecyclemodel-automated-builder.mjs.
Boundary markers: The system relies on the JSON schema of the manifest and TIDAS model contracts, but does not explicitly include delimiters or instructions to the agent to ignore natural language instructions found within the process-build data.
Capability inventory: The skill can execute subcommands of the tiangong-lca CLI and write numerous JSON and Markdown artifacts to the local file system (e.g., run-plan.json, selection-brief.md, tidas_bundle files).
Sanitization: The wrapper performs basic argument normalization and validation (checking for required flags like --out-dir), but the semantic content of the inputs is passed through to the underlying vendor-provided CLI for processing.

lifecyclemodel-automated-builder