The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources such as issue trackers, milestone descriptions, and design documents via commands like rota item show, rota milestone show, and rota design show. This data is used to generate implementation plans. While the skill includes instructions to treat specific data markers as boundaries and mandates a user confirmation step before writing the plan, there remains a risk that adversarial instructions embedded in the external content could influence the agent's planning output.
Ingestion points: External project data is loaded in SKILL.md (Step 2) using rota CLI tools to query items, milestones, and design history.
Boundary markers: The skill explicitly identifies "DECISIONS" markers as boundaries in SKILL.md (Step 2) and requires the agent to surface conflicts.
Capability inventory: The skill has the ability to write to the backend using rota plan add, rota plan put, rota item comment add, and rota item create.
Sanitization: The skill relies on a manual user confirmation step ("Confirm this plan? (yes / changes)") in SKILL.md (Step 3) before committing any plan to the backend.

rota-plan