The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines patterns for building web applications that process untrusted data, which is an inherent surface for indirect prompt injection.
Ingestion points: User-controlled data is ingested through path parameters, query strings, and request bodies via c.req.param(), c.req.query(), and c.req.json() as seen in SKILL.md.
Boundary markers: While no specific text delimiters are mentioned for prompt isolation, the skill promotes strict data boundaries through schema validation.
Capability inventory: The skill allows network fetching via curl and application testing/inspection via npx hono.
Sanitization: The skill provides high-quality security guidance by demonstrating how to use zValidator and sValidator to enforce strict schemas on all incoming data.
[EXTERNAL_DOWNLOADS]: The skill fetches documentation from hono.dev and installs the @hono/cli package. These resources are official components of the framework managed by the skill's author.
[COMMAND_EXECUTION]: The skill utilizes npx hono for application context and testing. It includes a specific security warning advising developers to use environment variables for sensitive values rather than passing them as command-line arguments, which prevents credential exposure in process logs.

hono