The Agent Skills Directory

[DATA_EXPOSURE]: The skill accesses configuration and preference files located in the user's home directory to manage cross-repository defaults and legacy settings.
Evidence: references/preferences.md instructs the agent to read from and write to ~/.config/softleader/agent-skills/scope-it/sources.json and ~/.config/softleader/agent-skills/scope-it-remake/sources.json.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from potentially untrusted external sources, such as repository planning files and existing issue tracker items, which could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: The skill reads repository guidance, existing planning artifacts, and issue tracker content as described in SKILL.md (Workflow Step 1) and references/recovery.md.
Boundary markers: While the skill uses specific delimiters () for its own generated output in the Delivery Map, it does not specify similar boundaries for the external content it ingests from other skills or repository files.
Capability inventory: The skill is capable of performing file read/write operations, invoking other agent skills, and interacting with repository management tools and issue trackers.
Sanitization: The instructions do not specify any explicit sanitization, filtering, or escaping of data retrieved from external repository files or issue tracker artifacts before it is incorporated into the agent's context.

scope-it