The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple local shell scripts located in ~/.claude/skills/nanostack/bin/ to manage workflow phases, find existing solutions, and handle telemetry data. These tools are internal to the author's nanostack ecosystem.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sprint artifacts to generate solution documentation, creating an attack surface for indirect prompt injection.
Ingestion points: Reads JSON artifact files (review, security, qa, and ship phases) resolved via the resolve.sh script (SKILL.md).
Boundary markers: No explicit delimiters or instructions to ignore embedded prompts are defined for the artifact reading process.
Capability inventory: The skill has the capability to write solution files via save-solution.sh, save artifacts, and modify skill files via graduate.sh (SKILL.md).
Sanitization: No explicit sanitization, validation, or filtering of the artifact content is documented.
[DYNAMIC_EXECUTION]: The graduation process within graduate.sh enables the promotion of validated solutions directly into SKILL.md files, which constitutes a form of automated instruction generation based on previously captured data.

compound