The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent to execute multiple local scripts and system-level CLI tools to manage the release process.
Evidence: Instructions in SKILL.md include running node scripts/release/check-gsi-update-limit.mjs, node scripts/release/check-pending-backfills.mjs, and bash scripts/common/sync-version.sh.
Evidence: Use of aws dynamodb describe-table to verify infrastructure state and git for branch operations and history analysis.
Evidence: Instructions to execute Python scripts using the backend virtual environment, e.g., backend/.venv/bin/python backend/scripts/backfill_tool_catalog_index.py.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it requires the agent to parse and act upon data from external repository sources like git logs and diffs.
Ingestion points: The agent is instructed to read and process git log output, git show diffs, and the contents of gsi-inventory.json, RELEASE_NOTES.md, and CHANGELOG.md to identify changes (§5 and §6 of SKILL.md).
Boundary markers: The instructions lack explicit boundary markers or directives for the agent to treat content within the git history or diffs as untrusted data.
Capability inventory: The skill has access to shell execution (bash, node, python), version control (git), and cloud infrastructure tools (aws CLI).
Sanitization: No sanitization or validation steps are defined for the information extracted from the repository's history before it is used to generate documentation or perform checks.

cutting-a-release