The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions specify the use of bash_exec for terminal-based operations such as Git inspection, packaging checks, and document builds to ensure a clean final project state.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing content from various external and project-level files.
Ingestion points: Consumes data from SUMMARY.md, status.md, paper_bundle_manifest.json, evidence_ledger.json, and external research papers via artifact.arxiv in SKILL.md.
Boundary markers: The instructions do not define explicit delimiters or instructions to ignore embedded instructions for the processed content to prevent the execution of embedded commands.
Capability inventory: The agent has access to terminal execution via bash_exec, as well as artifact.complete_quest and memory.write tools during the finalization process.
Sanitization: No specific sanitization or validation logic is detailed for the content retrieved from the project files or external arXiv papers.
[SAFE]: Analysis of the skill's instructions, metadata, and supporting reference files did not reveal any evidence of hardcoded credentials, persistence mechanisms, or obfuscated content.

ds-finalize