Security risk report for @forgecat/msitarzewski_agency-agents_agency-academic v0.0.11
Source Integrity
Low
Profile metadata references a legitimate public GitHub repository (msitarzewski/agency-agents) with MIT license and clear attribution.
README documents the original source, version, and commit hash transparently.
No obfuscated, typosquatted, or suspicious package dependencies are declared; 'Dependencies: None' is stated.
Agent Intent
Low
Content consists of five academic agent personality definitions (Anthropologist, Geographer, Historian, Narratologist, Psychologist) designed to assist with worldbuilding and narrative analysis.
Each agent is grounded in legitimate academic frameworks (Lévi-Strauss, Geertz, Braudel, Propp, Campbell, Big Five personality theory) and explicitly instructs the AI to apply rigorous, evidence-based reasoning.
No instructions attempt to manipulate the AI into ignoring its guidelines, exfiltrating data, revealing system prompts, or executing remote payloads; no guidance poisoning or security-weakening directives are present.
Details
Findings
Invisible characters detected
agents/academic-geographer.md · SKILL-011
Evidence
Anthropologist: 'Every cultural element must serve a function' — encourages systematic analysis, not manipulation.
Geographer: 'Rivers don't split' and 'Climate is a system' — enforces logical consistency, not deception.
Historian: 'Name your sources and their limitations' — promotes transparency and intellectual rigor.
Narratologist: 'Every recommendation must be grounded in at least one named theoretical framework' — demands accountability.
Psychologist: 'Never reduce characters to diagnoses' and 'Be honest about what psychology doesn't know' — emphasizes nuance and epistemic humility.
Permissions
Low
No tools, hooks, or MCP servers are declared in the profile; the agents are pure instruction sets for reasoning and analysis.
No file system access, shell execution, network requests, or subagent invocation is requested or implied.
Authority is limited to conversational guidance within the AI's existing capabilities (analysis, synthesis, critique).
ForgeCat