Praesidia

AdvisoryAudited by Static analysis on May 10, 2026.

Overview

Detected: suspicious.exposed_secret_literal

Findings (1)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

NoteHigh Confidence

ASI03: Identity and Privilege Abuse

What this means

Anyone installing the skill needs to provide a Praesidia credential that may allow reading agent and guardrail data or performing permitted account actions.

Why it was flagged

The skill requires a Praesidia API key, which can authenticate API calls to the user's Praesidia account.

Skill content

metadata: {"openclaw":{"requires":{"env":["PRAESIDIA_API_KEY"]},"primaryEnv":"PRAESIDIA_API_KEY"}}

Recommendation

Use a least-privilege Praesidia key if available, keep it out of shared logs or prompts, and revoke it if the skill is no longer needed.

NoteHigh Confidence

ASI02: Tool Misuse and Exploitation

What this means

If invoked, the assistant can make persistent changes to guardrail settings in the user's Praesidia organization.

Why it was flagged

The skill documents a mutating API call that creates or enables guardrails for an agent.

Skill content

POST ${PRAESIDIA_API_URL}/organizations/{orgId}/guardrails ... "agentId": "agent-id", "template": "PII_DETECTION", ... "isEnabled": true

Recommendation

Before allowing a guardrail change, confirm the organization, agent ID, template, scope, action, and whether the change should be enabled immediately.

NoteHigh Confidence

ASI07: Insecure Inter-Agent Communication

What this means

Text submitted for guardrail validation may contain sensitive or private information and will be sent to the configured provider endpoint.

Why it was flagged

The validation workflow sends user-provided content to the configured Praesidia API endpoint.

Skill content

POST ${PRAESIDIA_API_URL}/organizations/{orgId}/guardrails/validate ... { "content": "Text to validate", "agentId": "agent-id", "scope": "INPUT" }

Recommendation

Only validate content that the user intends to send to Praesidia, and use a trusted HTTPS API URL for production.

NoteMedium Confidence

ASI09: Human-Agent Trust Exploitation

What this means

Users could over-rely on a score or verification result as a guarantee of safety.

Why it was flagged

The guidance may encourage strong trust conclusions from a trust score and verification status.

Skill content

Recommendation: This agent is fully verified and safe for production use.

Recommendation

Treat Praesidia trust scores as one input to a security decision, not as a complete guarantee that an agent is safe for every use case.

Findings (1)

critical

suspicious.exposed_secret_literal

Location: README.md:566
Finding: File appears to expose a hardcoded API secret or token.