The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's primary logic in SKILL.md utilizes the shell tool to execute fdk commands such as fdk validate, fdk pack, and fdk version. It also uses standard Unix utilities including find, ls, cat, du, and grep to manage files and parse metadata within the app directory.
[PROMPT_INJECTION]: The SKILL.md file defines an 'AUTONOMOUS' execution mode, instructing the agent to run commands immediately without user confirmation. While this bypasses standard human-in-the-loop safeguards for shell operations, it is a declared functional characteristic for task efficiency.
[PROMPT_INJECTION]: The skill presents a surface for potential indirect prompt injection by processing external data.
Ingestion points: Reads local app files including manifest.json and coverage/coverage-summary.json (documented in SKILL.md).
Boundary markers: None identified; the skill processes command output directly.
Capability inventory: Access to shell, read, and write tools.
Sanitization: Contents of parsed JSON files are used to generate checklists without explicit sanitization steps.
[EXTERNAL_DOWNLOADS]: The README.md and reference documents provide instructions for installing the skill and the FDK CLI tool. These references point to official Freshworks developer repositories and standard package registries, which are trusted vendor sources.
[DATA_EXFILTRATION]: The skill accesses local files like manifest.json to extract metadata (app name, version, products) and coverage-summary.json to verify test metrics. This information is used solely to generate submission guidance and packages locally, with no observed unauthorized network transmission.

freshworks-publish-skill