The Agent Skills Directory

[SAFE]: The skill's instructions are focused on legitimate customer research workflows, such as extracting pain points, triggers, and vocabulary from transcripts and reviews. It provides structured synthesis steps and confidence-scoring guidelines to ensure research quality.
[SAFE]: All external sources referenced (Reddit, G2, Hacker News, SparkToro, LinkedIn, etc.) are well-known, public services. The skill uses these platforms appropriately to gather voice-of-customer data and does not engage in automated exfiltration or credential harvesting.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because its core function involves processing untrusted external content from web platforms and user-provided transcripts.
Ingestion points: Customer transcripts, survey results, support tickets, and various public online platforms mentioned in SKILL.md and references/source-guides.md.
Boundary markers: The instructions focus on quote extraction and thematic clustering but do not explicitly use delimiters or instruction-ignore warnings for the ingested data.
Capability inventory: The skill relies on the agent's file-reading and web-browsing capabilities to synthesize research.
Sanitization: No explicit sanitization or filtering of instructions within the external data is described.

customer-research