Praesidia
v1.0.1Verify AI agents, check trust scores (0-100), fetch A2A agent cards, discover marketplace agents, apply guardrails for security and compliance. Use when user mentions agent verification, trust scores, agent discovery, A2A protocol, agent identity, agent marketplace, guardrails, security policies, content moderation, or asks "is this agent safe?" or "find agents that can [task]" or "apply guardrails to protect my agent".
⭐ 1· 1.5k·0 current·0 all-time
by@msoica
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
medium confidencePurpose & Capability
The name and description (verify agents, trust scores, guardrails, A2A) align with the SKILL.md actions: GET agent-cards, list/apply guardrails, fetch templates and marketplace agents. The required credential (PRAESIDIA_API_KEY) is appropriate for an external verification/guardrails API.
Instruction Scope
Instructions correctly focus on calling the Praesidia API endpoints (agent-card, guardrails, templates, POST guardrails). Two points need clarification: (1) SKILL.md says to obtain the user's organization ID from 'their profile or context' but doesn't specify how — this may require reading user profile data or calling another API; (2) the instructions reference PRAESIDIA_API_URL but that environment/config value is only shown in an example config and is not declared as a required env var in metadata. Both are scope/clarity issues rather than functional mismatches.
Install Mechanism
Instruction-only skill with no install spec and no code files. Nothing will be written or executed at install time by the skill itself.
Credentials
The skill requests a single API key (PRAESIDIA_API_KEY), which is proportional for a service that reads and writes guardrail configurations. However: (a) the SKILL.md expects PRAESIDIA_API_URL in config but the metadata does not declare it as required, and (b) the example encourages storing the API key in ~/.openclaw/openclaw.json, which means a plaintext key stored on-disk—users should confirm acceptable storage posture. Also consider whether the provided API key is read-only or has guardrail-modifying (write) permissions; the skill's POST examples show it can modify guardrails.
Persistence & Privilege
always is false (good). The skill can be invoked autonomously (the platform default) and its documented operations include POSTing guardrails (making configuration changes). That capability is expected for this purpose, but it increases blast radius if an API key with write privileges is supplied—consider limiting key scope or requiring explicit user confirmation before write actions.
Assessment
This skill appears to do what it says: call Praesidia to verify agents and manage guardrails. Before installing: (1) Confirm PRAESIDIA_API_KEY scope — prefer a read-only key if you only want verification, and only give write permissions if you trust the skill to modify guardrails. (2) Ensure PRAESIDIA_API_URL is configured or declared — the SKILL.md expects it but metadata doesn't list it as required. (3) Be aware the example stores the key in ~/.openclaw/openclaw.json (plaintext); decide if that storage is acceptable and rotate keys regularly. (4) Clarify how the agent will obtain orgId (will it read profile data or call another API?) and whether the skill will prompt before making POST requests that change guardrails. If you need stronger guarantees, request an explicit list of API endpoints the key can access and ask for a read-only mode for verification-only usage.Like a lobster shell, security has layers — review code before you run it.
a2avk978jfmm32bkqwj8btw0hngexs80kc0yagent-communicationvk978jfmm32bkqwj8btw0hngexs80kc0yauthenticationvk978jfmm32bkqwj8btw0hngexs80kc0ylatestvk9705308me2st52xjecqv3cs0980kqmymessagingvk978jfmm32bkqwj8btw0hngexs80kc0yoauthvk978jfmm32bkqwj8btw0hngexs80kc0ypraesidiavk978jfmm32bkqwj8btw0hngexs80kc0yprotocolvk978jfmm32bkqwj8btw0hngexs80kc0ysecurityvk978jfmm32bkqwj8btw0hngexs80kc0ytrustvk978jfmm32bkqwj8btw0hngexs80kc0y
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🛡️ Clawdis
EnvPRAESIDIA_API_KEY
Primary envPRAESIDIA_API_KEY