The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes platform-specific instructions to ensure strict execution of the protocol and compatibility between different agent environments (e.g., 'Ignore Claude-specific mode-switch instructions'). These are interpreted as instructional framing rather than malicious attempts to bypass safety guardrails.\n- [COMMAND_EXECUTION]: The workflow utilizes the spawn_agent tool to handle multi-step gap-filling tasks as sub-agents. This is an authorized use of platform tools designed to maintain context boundaries and follow project-specific workflow advancement rules.\n- [EXTERNAL_DOWNLOADS]: The generated slide deck references Google Fonts (fonts.googleapis.com) for typography. This is a well-known and trusted service, and its use is documented neutrally as it does not pose a security risk.\n- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by processing external artifact data.\n
Ingestion points: Project artifacts such as ideas, PBIs, and specs are read from the team-artifacts/ and docs/specs/ directories.\n
Boundary markers: Embedded mockups are isolated within iframe elements using the srcdoc attribute.\n
Capability inventory: The skill has the ability to write to the filesystem and spawn sub-agents for expanded workflows.\n
Sanitization: A mandatory 'escape-once' protocol for HTML entities (ampersand, quotes, and brackets) is enforced for all embedded content to prevent script breakout and cross-site scripting (XSS) within the generated deck.\n- [SAFE]: No malicious patterns, obfuscation, or persistence mechanisms were detected. The skill's primary purpose is functional artifact synthesis with appropriate security mitigations for its dynamic output.

feature-presentation