The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes several local shell scripts for core functionality, including resolve.sh, summarize.sh, and find-artifact.sh. These invocations are limited to the skill's own directory and the infrastructure defined in the local NANOSTACK_ROOT.
[INDIRECT_PROMPT_INJECTION]: The skill processes artifacts from external sources, making it a potential target for indirect injection through data poisoning.
Ingestion points: The bin/summarize.sh script reads JSON artifacts from the .nanostack/ directory via the find-artifact.sh utility.
Boundary markers: The skill relies on structured JSON parsing via jq to extract only the necessary status and integrity fields, preventing the ingested data from being interpreted as instructions.
Capability inventory: The skill's operations are confined to file reading and JSON output generation. It does not possess network capabilities or the ability to execute code derived from the artifacts.
Sanitization: The skill enforces a strict verification model. It uses SHA-256 integrity hashes to validate each artifact and will automatically block the release process if any artifact is missing, tampered with, or lacks verification metadata.

release-readiness