The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, which could potentially contain malicious instructions intended to manipulate the agent's planning logic.
Ingestion points: Processes "fuzzy goal", "sparse requirements", and "codebase context" as described in SKILL.md.
Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are defined to protect the agent from instructions hidden within the source requirements or codebase.
Capability inventory: The skill possesses the ability to edit workspace files (e.g., project-graph.json, manifest.json), interact with the PlanWeave CLI, and define "validation commands" for blocks.
Sanitization: There is no evidence of sanitization or escaping of the input data before it is used to generate the plan draft.
[DYNAMIC_EXECUTION]: The skill facilitates the creation of executable content that is intended to be run in the user's environment.
The SKILL.md instructions guide the agent to generate "Block prompts" that include "exact execution instructions" and "validation commands".
These commands (e.g., shell scripts, test runners) are dynamically constructed based on the requirements and codebase context discovered during the planning phase.

plan-maker