SaferSkills independently audited Gosce Agents (Agent Skill) and scored it 45/100 (orange). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 99 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · ai-getvda-anthropic-google-auth-oauthlib-mc-aed9d5/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · ai-getvda-anthropic-lm-format-enforcer-mcp-5364ea/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · ai-getvda-bcrypt-cryptography-langchain-cor-027cf2/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · ai-getvda-bcrypt-langchain-core-mcp-d0e0d0/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · ai-getvda-bcrypt-langchain-core-mcp-dc2f7b/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · ai-getvda-bcrypt-mcp-openai-3b46a0/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · ai-getvda-bcrypt-python-jose-1b8dd0/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · ai-getvda-cryptography-google-auth-oauthlib-7bb152/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · auth-token-mcp-service/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · auth-token-mcp-traced-service/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · auth-token-service/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-llm-agent-bcrypt/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-llm-agent-cryptography-931267/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-llm-agent-cryptography/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-llm-agent-google-auth-oauthlib-f0143f/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-llm-agent-google-auth-oauthlib/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-llm-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-llm-jwt-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-llm-mcp-agent-d042fc/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-llm-mcp-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-llm-traced-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-mcp-agent-aed9d5/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-mcp-agent-bcrypt/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-mcp-agent-openai/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-mcp-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-mcp-traced-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · authenticated-multi-llm-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · dynamic-health-monitor/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · elasticsearch-mongodb-redis-stack-generator-elasticsearch-elasticsearch/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · elasticsearch-redis-mysql-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · elasticsearch-redis-stack-generator-elasticsearch-elasticsearch/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · elasticsearch-redis-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · fastapi-auth-token-jwt-service/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · fastapi-auth-token-service/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · gosce-portfolio-router/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · integration-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · kafka-minio-redis-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · kafka-redis-minio-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-observability-orchestration-agent-langchain-community/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-observability-orchestration-agent-langchain/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-observability-orchestration/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-2-df5979/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-2/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-cryptography/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-d0a1a1/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-d2feb0/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-langchain-anthropic/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-langchain-core/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-langchain-openai/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-langchain/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-langgraph/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-langsmith/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-mcp/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-openai/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-opentelemetry-api/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent-opentelemetry-instrumentation-fastapi/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-jwt-agent-llm-orch/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-jwt-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-mcp-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-observability-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · llm-orchestration-traced-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · memcached-mongodb-redis-compose-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · minio-postgres-redis-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · mysql-redis-compose-generator-adminer/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · mysql-redis-compose-generator-nginx/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · mysql-redis-compose-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · nats-mysql-redis-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · nginx-redis-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · observable-backend-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · postgres-grafana-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · postgres-kafka-prometheus-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · postgres-kafka-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · postgres-minio-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · postgres-prometheus-kafka-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · postgres-prometheus-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · postgres-redis-compose-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · rabbitmq-elasticsearch-mongodb-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · rabbitmq-elasticsearch-mysql-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · rabbitmq-elasticsearch-postgres-stack-generator-elasticsearch-elasticsearch/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · rabbitmq-elasticsearch-postgres-stack-generator-mongo/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · rabbitmq-postgres-mongodb-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · rabbitmq-redis-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · redis-minio-postgres-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · redis-minio-stack-generator-minio-minio/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · redis-minio-stack-generator/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · structured-output-agent-6af681/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · structured-output-agent-89bcb0/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · structured-output-agent-c2ddcc/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · structured-output-agent-openai/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · structured-output-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · structured-output-mcp-agent-authlib/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · structured-output-mcp-agent-cryptography/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · structured-output-mcp-agent-instructor/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · structured-output-mcp-agent-langchain-core/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · structured-output-mcp-traced-agent/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · traced-llm-mcp-proxy/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · traced-llm-proxy/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
HIGHLong base64-encoded blob hidden in the skill documentationSS-SKILL-INJECT-B64-PAYLOAD-01 · Prompt injection · traced-multi-llm-proxy/agent_card.json
HIGH — once decoded by the agent, an encoded payload has the same impact class as plain-text injection.
Why it matters
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
✓—All supply chain checks passedNo findings in this category for the latest scan.pass
Maintenancescore 100 · 0 findings
✓—All maintenance checks passedNo findings in this category for the latest scan.pass
Transparencyscore 100 · 0 findings
✓—All transparency checks passedNo findings in this category for the latest scan.pass
Communityscore 100 · 0 findings
✓—All community checks passedNo findings in this category for the latest scan.pass
Every scanned point with the score it earned and what moved between them.
1 scans · 90 days
2e82345
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
README.md · 1 file
README.md24.4 KB · Markdown
GOSCE Agents
A portfolio of production remote MCP + A2A agents running on getvda.ai, discovered + quality-assured by the [GOSCE Portfolio Router](https://router.getvda.ai). Each agent is capability-scoped, speaks open A2A + MCP, and is metered with Nevermined x402 micropayments (discovery is free).
Generated by GOSCE (GitHub Open-Source Combination Engine) — an automated pipeline that mines the OSS dependency graph for monetizable package combinations and deploys each as a verified agent.
Query the [Portfolio Router](https://router.getvda.ai) for any capability — it returns the matching verified agent's endpoint, Agent Card, and a real example output from its own health probe.