The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes institutional knowledge by reading markdown files from the docs/solutions/ directory, which represents an attack surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: The agent reads content from various documentation files in docs/solutions/ and specifically targets docs/solutions/patterns/critical-patterns.md as mandatory reading.
Boundary markers: The instructions lack explicit boundary markers or instructions to treat the documentation content strictly as data, which may lead the agent to follow instructions embedded within the files it researches.
Capability inventory: The skill uses the content-search and Read tools to access and process the local file system.
Sanitization: There is no evidence of sanitization, filtering, or escaping of the content read from the documentation files before it is processed or summarized.
[COMMAND_EXECUTION]: The skill instructions direct the agent to perform search and read operations on the local file system using platform tools.
Evidence: Under the "Search Strategy" section, the skill provides specific commands for content-search to find candidate files and Read tools to extract content from the found documentation.

learnings-researcher