The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill is designed to ingest and act upon data from external sources ('Lumen data' and 'Echo signal') to establish roadmap priorities. The lack of defined security boundaries or sanitization instructions for this potentially untrusted data, coupled with the skill's access to powerful tools like Bash and Write, creates a vulnerability where malicious instructions embedded in the data could influence the agent's actions.
Ingestion points: External data signals referred to as 'Lumen' and 'Echo' used in Step 1 for establishing current signal.
Boundary markers: Absent. There are no instructions for the agent to use delimiters or to disregard instructions found within these data sources.
Capability inventory: The skill has access to Bash, Write, Edit, WebFetch, and WebSearch tools as defined in the frontmatter configuration.
Sanitization: Absent. No logic is provided to validate or filter the content of the external signals before processing.
[SAFE]: Metadata Discrepancy. A version mismatch exists between the frontmatter in SKILL.md (0.6.4) and the .claude-plugin/plugin.json file (1.9.1). This is a minor administrative error and does not impact the security of the skill.

crest-roadmap