- Home
- Skills
- Agents & Orchestration
- reliability failure semantics reviewer
Works with the AI Agents you already use
reliability failure semantics reviewer
Senior principal-level review of agent reliability, failure semantics, recovery design, and operational bounds. Outputs severity-ranked findings only.
$45
reliability failure semantics reviewer
Example session with this skill installed
Review the reliability and failure semantics of the following agent system:
The agent calls an LLM in a while True loop with no maximum retry count. Tool calls (web_search, db_query, file_read) have no timeouts. Session state is written with a plain json.dump after each step. On parse failure the agent returns None and continues. Recovery retries a failed step 5 times then retries 5 more times with no human escalation.
- Read your context and instructions
- Compiled the reliability failure semantics
Critical
-
Unbounded retry on LLM API calls with no circuit breaker
Evidence: while True loop with no maximum attempt count.
Impact: Indefinite retry on provider outage consumes quota and blocks the worker.
Required fix: Bounded retry (max 3), exponential backoff with jitter, circuit breaker after 5 consecutive failures. -
State corruption on crash mid-workflow with no transactional writes
Evidence: json.dump used without atomic write.
Impact: Crash leaves partially written session state. Restart may proceed on corrupted data.
Required fix: Atomic write (temp file + replace) + schema validation on load.
High
- Timeout configured only on main LLM call, not on tool sub-calls
Evidence: web_search, db_query, and file_read have no timeout.
Impact: Hung tool call can stall the event loop indefinitely.
Required fix: Explicit timeouts on every tool call.
Verdict: FAIL
Connects securely to your tools. The creator never sees your data.
About this skill
This skill performs a principal-engineer review of agent system reliability and failure semantics.
It examines system boundaries, state handling, tool failure modes, retry behavior, timeout design, recovery paths, resource bounds, and observability of failures. Every finding is classified by severity with evidence, impact, and a required fix. No praise, no filler — only actionable engineering findings.
Use it when you need a hard review of reliability design before production, during incident analysis, or when auditing recovery behavior. The output is strictly severity-ranked and ends with a clear PASS / CONDITIONAL / FAIL verdict.
Built for senior and principal engineers who require production-grade standards, not demo-level reviews.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
Does not suggest new tools or frameworks. Requires a detailed description of the system architecture to provide meaningful results.
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 today