The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands to run test suites and CI gates to establish baselines and verify repairs.
Evidence: SKILL.md (Baseline, Cut over, Preservation review) specifies recording the "result and duration of each suite", running "affected project gates", and executing "commands run".
[DYNAMIC_EXECUTION]: The skill performs mutation testing, which involves programmatically modifying production code to ensure tests fail as expected.
Evidence: SKILL.md (Preservation review) describes mutating production behavior and confirming that specific "keepers" go red, utilizing a disposable checkout and time limits.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from repository history and external tracking systems which could contain embedded instructions.
Ingestion points: SKILL.md (Baseline, Ledger) mentions reading scoped agent instructions, coding standards, gate configurations, CI routing, and tracing rationale through "commits, pull requests, and issues".
Boundary markers: Absent. No specific delimiters or instructions to ignore embedded prompts are defined for the data retrieved from Git history or issue trackers.
Capability inventory: SKILL.md (Cut over, Preservation review) involves file system writes (deleting/consolidating code), shell command execution (running test suites), and Git operations (commits and PR creation).
Sanitization: Absent. There is no evidence of validation or sanitization of content retrieved from external commit or issue metadata before it is processed by the agent.

prune