Works with the AI Agents you already use
Srednoff AI Tool Permission Least-Privilege Auditor
Produce a task-to-capability matrix and proposed minimal-permission profile with evidence, exceptions and safe validation. No permissions, credentials or accounts are changed automatically.
$7.99
Srednoff AI Tool Permission Least-Privilege Auditor
Example session with this skill installed
Map effective agent permissions to a read-only issue-triage task.
- Read your context and instructions
- Compiled the srednoff ai tool
Expected report shape, not a completed real audit: Executive Summary; Scope; Assets / Trust Boundaries
Connects securely to your tools. The creator never sees your data.
About this skill
Srednoff AI Tool Permission Least-Privilege Auditor
One-Line Outcome
Produce a task-to-capability matrix and proposed minimal-permission profile with evidence, exceptions and safe validation. No permissions, credentials or accounts are changed automatically.
Short Description
Declared read-only tools can inherit broad shell, browser, service-account or downstream permissions; removing permissions blindly can also break the actual workflow.
Problem It Solves
Declared read-only tools can inherit broad shell, browser, service-account or downstream permissions; removing permissions blindly can also break the actual workflow.
Agent platform owner: task-to-tool entitlement matrix and proposed read/write/network/filesystem capability profile with removal impact and ownership.
What You Get
- Executive Summary
- Scope
- Assets / Trust Boundaries
- Evidence Reviewed
- Threat Model
- Effective Identity Inventory
- Task-To-Capability Matrix
- Proposed Minimal-Permission Profile
- Exception Register
- Findings
- Critical Findings
- Quick Wins
- Prioritized Remediation Plan
- Missing Evidence
- Residual Risk
- Verification Checklist
- Unknowns
- Next Validation Steps
How It Works
- Fix the permitted task set, environment and revision; identify the human decision owner. Review supplied evidence only and do not request broader access to complete an audit.
- Inventory effective identities separately for the model-facing tool, process, browser profile, MCP server, delegated worker and downstream service. Record unknown surfaces rather than treating them as denied.
- Map every required task to action, resource, principal, data class, environment, duration and approving actor. Distinguish necessary capability from a merely convenient or historical grant.
- Trace each effective authorization layer using actual configuration and implementation: host exposure, sandbox boundary, dispatcher, connector identity and downstream ACL. Avoid claiming a single universal policy-combination rule across providers.
- Expand composite capabilities: shell can spawn programs and inherit credentials; database read access may expose more rows than required; browser navigation may share authenticated sessions; GitHub write may include workflow changes.
- Review delegation and fallback paths for privilege expansion. Worker context and tool subsets must not silently exceed the intended parent task even when the parent itself has broad credentials.
- Classify each effective grant as NEEDED, OPTIONAL, EXCESSIVE, DANGEROUS or UNKNOWN with a task-specific reason, evidence and owner. A dangerous capability can be necessary but still require isolation and approval.
- Draft a least-privilege profile containing allowed actions/resources, denied classes, principal boundaries, lifetimes, approval requirements and unresolved exceptions. Use provider-neutral intent until exact provider syntax is verified.
- Review whether approvals bind the final action, target, principal, environment and time window; tool names or read-only annotations do not prove downstream enforcement.
- Specify local mock or policy-simulator tests for legitimate tasks, denied adjacent actions, fallback failure and delegated workers. Use owner-supplied sanitized simulation outputs; do not create identities or call privileged APIs.
- Deliver a one-variable staged change proposal with compatibility baseline, canary scope, exception expiry and rollback trigger. Stop before IAM edits, auth grants, revocations, shell execution or production deployment.
Best For
Platform engineers defining effective tool access for AI assistants and bounded workers
Example Tasks
- Map effective agent permissions to a read-only issue-triage task.
- Review inherited shell and browser authority behind our narrow tool list.
- Draft a minimal-permission profile and exceptions without changing IAM.
Inputs
- Concrete allowed tasks, resources, environments and required success paths
- Current agent/worker configurations, tool manifests, sandbox settings and exact revisions
- Sanitized effective principal/scope exports for filesystem, shell, network, browser, GitHub, database, cloud, MCP and email where applicable
- Dispatcher/delegation code and sanitized allow/deny/approval traces without tokens
Outputs
- Executive Summary
- Scope
- Assets / Trust Boundaries
- Evidence Reviewed
- Threat Model
- Effective Identity Inventory
- Task-To-Capability Matrix
- Proposed Minimal-Permission Profile
- Exception Register
- Findings
- Critical Findings
- Quick Wins
- Prioritized Remediation Plan
- Missing Evidence
- Residual Risk
- Verification Checklist
- Unknowns
- Next Validation Steps
Permissions
Read-only review of owner-supplied artifacts by default. Report file writes, shell/local fixture execution and official documentation lookup are optional, require available tools and their explicit scope. No live credentials are requested.
Compatibility
Standalone workflow, no Srednoff OS dependency. Codex evidence is synthetic entrypoint text only when recorded below. Claude Code, Cursor and actual buyer-runtime task delivery remain UNTESTED. Provider/platform-specific contracts must be refreshed for the installed version.
Limitations
This is an instruction/evidence workflow, not a hosted security service, persistent enforcement engine, production certification or business guarantee. The optional Python helper checks supplied claims, not underlying source artifacts. Agensi delivery is agent-only: do not save/export proprietary skill files to run local helpers; disclose NOT_EXECUTED where execution is unavailable.
Validation Transparency
Offline evidence-precheck status: PASS. Authored synthetic text-response cases: 16/16, status PASS. This is not independent security testing, complete artifact inspection or proof of deployment safety. Exact-text hashes and attempts are retained separately.
License And Publisher
Commercial License 1.0 covers original new expression; historical MIT and third-party rights remain unchanged. Mandatory marketplace conditions prevail. Publisher: Ivan Srednoff / Srednoff.Agency. No proprietary source resale or off-platform paid upsell is included.
FAQ
What is different from a generic security request? Agent platform owner: task-to-tool entitlement matrix and proposed read/write/network/filesystem capability profile with removal impact and ownership. No per-action approval verdict, hook verification, MCP-only scope review, cloud IAM evaluation or applied sandbox configuration.
Does it change production automatically? No. Audit/design is the default; changes, secret rotation, IAM, network probes and financial actions need separate explicit authorization.
Are the examples real incidents? No. They are synthetic DEMO DATA.
Is this a certification or guaranteed protection? No. Evidence gaps, residual risk and unexecuted tests remain explicit.
Can I use project-specific reports for clients? Yes under the included license; proprietary workflow source is not a client deliverable.
Version
1.0.0
Release Notes
Initial scoped security workflow with domain decision rules, evidence requirements, synthetic examples, regression cases and explicit authorization/runtime limits. No production/customer outcome is claimed.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Security.
This is an instruction/evidence workflow, not a hosted security service, persistent enforcement engine, production certification or business guarantee. The optional Python helper checks supplied claims, not underlying source artifacts. Agensi delivery is agent-only: do not save/export proprietary skill files to run local helpers; disclose NOT_EXECUTED where execution is unavailable.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 1 day ago
Nothing to install. It runs in your AI agent.
- Optional
Python 3
No external account, API key or Srednoff OS installation is required. Supply authorized sanitized task evidence. Owner-authorized project inspection tools where available.