The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various shell commands including git log, git branch, git diff, grep, and ls to analyze the repository state, check for implementation progress, and verify acceptance criteria. It also invokes local Node.js wrappers such as scripts/resolve-feature.js to determine feature context.
[DYNAMIC_EXECUTION]: The skill programmatically invokes local scripts (e.g., scripts/resolve-feature.js, scripts/resolve-feature-cli.js) and shell wrappers to perform feature resolution and implementation analysis at runtime.
[INDIRECT_PROMPT_INJECTION]: The skill reads and processes content from request tickets, git commit logs, and source code files. This external content is interpolated into a prompt for a subagent (subagent_type: "Explore") tasked with verifying acceptance criteria (AC).
Ingestion points: Request tickets located in docs/features/*/requests/*.md, git commit history, and project source/test files identified in the "Related Files" section of requests.
Boundary markers: The prompt template for the AC verification agent (Phase 2.5) does not utilize specific delimiters or instructions to disregard potentially malicious instructions embedded within the ingested code or documentation.
Capability inventory: The skill possesses the Bash, Write, and Agent tools, which could be leveraged if the subagent is successfully manipulated by injected instructions.
Sanitization: The skill implements regex-based slug validation (/^[a-z0-9][a-z0-9._-]*$/i) for feature keys to mitigate path traversal, though it does not explicitly sanitize the text of acceptance criteria or code comments before processing.

create-request