The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements a persistent learning mechanism by reading from and appending to references/learnings.md.
Ingestion points: references/learnings.md is read at the start of each invocation to provide context.
Boundary markers: No explicit markers are used to separate learned content from system instructions.
Capability inventory: The skill primarily performs text-based routing and guidance; it does not possess file-writing or network-execution capabilities within its own logic (writing to the learning file is described as an agent behavior).
Sanitization: None described. This is a standard functional pattern for this agent type and does not present an immediate risk.
[DATA_EXPOSURE_AND_EXFILTRATION]: Accesses local reference files (references/platform-guide.md and references/learnings.md) to retrieve platform information. There is no evidence of access to sensitive system directories or environment variables.
[COMMAND_EXECUTION]: The documentation includes a setup command for users (npx skills add sales-skills/sales --skill sales-do) to install related utilities from the same author ('sales-skills'). This is a documented user action and not an automated execution by the skill itself.

sales-backtrack