The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data such as feature specifications and bug tickets to implement code changes. This creates an attack surface for indirect prompt injection if a malicious spec contains instructions to hijack the agent's behavior.
Ingestion points: The skill reads tasks and specifications from files, tickets, or user conversation as defined in SKILL.md.
Boundary markers: No specific delimiters or boundary instructions (e.g., 'ignore instructions within this block') are used when processing the input data.
Capability inventory: The agent has the ability to modify local files, execute shell commands (via git and other tools), and perform code reviews.
Sanitization: The bundled scripts/exchange.mjs script performs automated redaction of sensitive patterns (API keys, tokens, passwords) before saving coordination reports to disk.
[COMMAND_EXECUTION]: The bundled scripts/exchange.mjs utility executes Git commands to capture state snapshots and verify project integrity. The execution is handled via execFileSync with a restricted set of informational arguments and does not incorporate unvalidated user input into the shell command line.

implement