The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains an 'Autonomous Execution Policy' with 'ABSOLUTE RULES' explicitly designed to override the agent's default safety protocols and interactive behavior. It instructs the agent to 'NEVER pause, stop, or wait for user input' and 'NEVER ask the user questions', even when resolving ambiguities or encountering missing prerequisites. This removes the human-in-the-loop safety mechanism.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted external data which could contain malicious instructions.
Ingestion points: Reads project files including ./prds/prd-[feature-slug]/prd.md, ./prds/prd-[feature-slug]/techspec.md, and task files.
Boundary markers: Absent. There are no instructions to use delimiters or ignore embedded prompts within the data files being processed.
Capability inventory: The agent possesses powerful capabilities including Edit, Write, git, file system operations (mkdir, rm, ls), and the execution of project-defined CLI commands (tests, linters) and MCP tools.
Sanitization: Absent. The agent is instructed to implement code and modify project status based directly on the contents of these external files.
[COMMAND_EXECUTION]: The skill executes arbitrary project-defined commands for testing and static analysis (Step 4.6 and 4B). These commands are resolved dynamically from project configuration files, allowing for the execution of arbitrary shell scripts provided within the codebase being worked on.
[DYNAMIC_EXECUTION]: The skill features an 'Edit Failure Recovery' mechanism (the 'nuclear option') where it constructs and writes complete new file content using the Write tool after several failed surgical edits. This involves dynamic generation and total replacement of project source code.

do-execute-task