The Agent Skills Directory

[COMMAND_EXECUTION]: The skill enables the execution of the restic command-line interface and provides documentation for the --stdin-from-command flag, which triggers the execution of data-producer processes like mysqldump as child processes.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted external data from backup repositories and local filesystems.
Ingestion points: The agent is instructed to read and parse snapshot metadata and repository statistics through restic snapshots --json and restic stats --json in files such as SKILL.md and references/performance-and-observability.md.
Capability inventory: The skill leverages the restic binary for significant filesystem and network operations and can execute arbitrary child processes via specific restic flags.
Boundary markers: The instructions do not explicitly mandate the use of delimiters or 'ignore' instructions when interpolating repository data into the LLM context, although they favor structured JSON parsing.
Sanitization: The skill provides strong remediation guidance, including the use of separate targets for restoration and the systematic redaction of secrets from all logs and output.
[DYNAMIC_EXECUTION]: The test utility scripts scripts/test-restic-preflight.sh and scripts/test-restic-verify.sh dynamically generate and execute temporary shell scripts to mock the restic binary. These operations are performed within a controlled environment and do not ingest external network data.

restic