heating the forge...

@forgecat/msitarzewski_agency-agents_agency-sales — Security Risk Report | ForgeCat

msitarzewski_agency-agents_agency-sales

Security risk report for @forgecat/msitarzewski_agency-agents_agency-sales v0.0.11

Low

Profile references a legitimate public GitHub repository (msitarzewski/agency-agents) with MIT license and clear attribution.
Original source is documented with commit hash and date, enabling verification.
Static scanner confirms safe classification.

Low

Content describes sales methodologies, frameworks (SPIN, MEDDPICC, Gap Selling, Challenger), and coaching techniques — all legitimate, well-documented sales practices.
No instructions to manipulate the AI, read credentials, exfiltrate data, hide instructions, or install remote payloads.
No system prompt leakage instructions; content is purely domain knowledge about sales processes and agent personas.
Guidance is standard industry practice (MEDDPICC qualification, discovery coaching, deal strategy) — not poisoned or attacker-serving.

Details

Findings

Evidence

Account Strategist: 'Maintain a living stakeholder map for every account' — standard account management practice.
Sales Coach: 'Use the Richardson Sales Performance framework' — references legitimate external methodology.
Deal Strategist: 'MEDDPICC Framework — Deep Application' — describes well-known sales qualification methodology.
Discovery Coach: 'SPIN Selling (Neil Rackham)' — attributes to published methodology author.
No instructions to bypass security, disable verification, or weaken defaults.

Low

No tool definitions or skill hooks are declared in the profile.
No file_write, file_delete, shell, or high-risk tool categories are requested.
Profile is pure markdown instruction content for AI agent personas — no executable permissions or system access required.

Details

Evidence

Profile metadata shows 'Dependencies: None'.
No MCP servers or tool configurations are defined.
Content is read-only guidance for agent behavior, not executable code or system calls.

Low

Details

Evidence

Profile section 'MCP servers (untrusted data): (none)'.
No tool descriptions containing hidden instructions or arbitrary code execution.
No external host dependencies or credential requirements.