The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external files, specifically MVP.md and BRAND.md, which creates a surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: The importar_mvp.mjs script reads content from MVP.md and BRAND.md located in the project root or superproject.
Boundary markers: Instructions in SKILL.md direct the agent to categorize themes and avoid inventing answers for gaps in the source text, providing some logical separation.
Capability inventory: The skill has the ability to write files to the local file system (milestones, backlogs, specs) and execute local utility scripts via Node.js.
Sanitization: The script implements a containsSensitiveData function that scans input for private keys, secrets, tokens, and passwords using regular expressions, aborting the process if such data is detected.
[COMMAND_EXECUTION]: The skill performs shell commands and executes local JavaScript files to facilitate project setup and data processing.
Evidence: The skill uses git -C <root> rev-parse --show-superproject-working-tree to resolve repository structures.
Evidence: The Node.js script utilizes child_process.execFile to run sibling scripts including capturar_inbox.mjs, iniciar_backlog.mjs, and iniciar_spec.mjs. The use of execFile with an argument array is a secure practice that avoids shell-level command injection.

specsfy-mvp-milestone-interviewer