- Home
- Skills
- Testing & Debugging
- Scheduled Job Reliability Audit
Works with the AI Agents you already use
Scheduled Job Reliability Audit
Audit recurring-job logs without inflating success rates. Separate retries, late completions, missing observations and unfinished work, then produce traceable metrics and a prioritized investigation plan.
$5
Scheduled Job Reliability Audit
Example session with this skill installed
Synthetic demonstration request
Audit hourly invoice-export runs for 2026-11-01 from 00:00 through 08:00 UTC, excluding 08:00. Each slot must finish within 30 minutes. The evidence cutoff is 08:00 UTC. This export is complete for the window, and final=true means no further retries are pending. Report delivery, timeliness, missing evidence and next investigations. Do not rerun anything.
All timestamps below are UTC on 2026-11-01. These are attempt snapshots, not a stream of status-change events.
| Row | Slot | Attempt | Start | Finish | Status | Final | Error |
|---|---|---|---|---|---|---|---|
| 1 | 00:00 | a0 | 00:00 | 00:10 | success | true | |
| 2 | 01:00 | a1 | 01:00 | 01:05 | failed | false | upstream timeout |
| 3 | 01:00 | a1r | 01:06 | 01:15 | success | true | |
| 4 | 02:00 | a2 | 02:00 | 02:10 | failed | true | export rejected |
| 5 | 04:00 | a4 | 04:00 | 04:40 | success | true | |
| 6 | 05:00 | a5 | 05:00 | running | false | ||
| 7 | 06:00 | a6 | 06:00 | 06:10 | success | true | |
| 8 | 06:00 | a6 | 06:00 | 06:10 | success | true | |
| 9 | 07:00 | a7 | 07:00 | 07:05 | cancelled | true | operator cancellation |
- Read your context and instructions
- Compiled the scheduled job reliability
Four of eight due slots have delivered successfully by cutoff; three were on time. The 03:00 slot has no observation in the complete export, and the 05:00 attempt remains running 150 minutes past its deadline. The retry at 01:00 recovered within its deadline and does not constitute a missed delivery.
Contract: job invoice-export; expected slots 00:00–07:00 UTC; 30-minute deadline; cutoff 08:00 UTC; complete export; explicit final-state flags. Row 8 duplicates row 7 and is excluded. There are eight distinct attempts across seven observed slots.
| Measure | Result |
|---|---|
| Due slots | 8 |
| Outcome reconciliation | 4 success + 1 failed + 1 cancelled + 1 unresolved + 1 missing = 8 |
| Observed slot coverage | 7/8 = 87.50% |
| Delivery by cutoff | 4/8 = 50.00% |
| On-time delivery | 3/8 = 37.50% |
| Closed-slot success | 4/6 = 66.67%; excludes the missing and unresolved slots |
| Retry recoveries | 1: 01:00 |
| Optional terminal-attempt success | 4/6 = 66.67%; excludes one cancellation and one running attempt |
| Slot | Outcome | Evidence / next check |
|---|---|---|
| 01:00 | Success, on time | Rows 2–3: timeout followed by successful retry, finished 01:15 |
| 02:00 | Failed | Row 4: terminal rejection; inspect the rejection details before proposing a fix |
| 03:00 | Missing observation | No row; compare scheduler dispatch history and export ingestion coverage |
| 04:00 | Success, late | Row 5: first success at 04:40, 10 minutes beyond its deadline |
| 05:00 | Unresolved | Row 6: still running at cutoff; inspect worker heartbeat/queue state before any rerun |
| 07:00 | Cancelled | Row 9: cancellation is recorded; confirm whether it reflects intended skipped work |
First-success latencies are 10, 15, 40 and 10 minutes; min/median/max are 10/12.5/40 minutes (n=4). These are job-delivery measurements, not availability or uptime.
Investigate the 03:00 evidence gap and overdue 05:00 execution first because their disposition is unknown. Then inspect the terminal 02:00 rejection and the 04:00 execution trace. Confirm the 07:00 cancellation policy. The sample does not establish an infrastructure outage or explain the late completion's cause. All steps above are investigation recommendations; no production changes were made.
Connects securely to your tools. The creator never sees your data.
About this skill
Find the delivery failures hidden by retry counts
A job can fail twice and still deliver on time. Another can disappear from your logs entirely. This skill guides your AI agent through a read-only audit of scheduled work, keeping expected deliveries separate from execution attempts.
What you receive
- A SKILL.md workflow plus two synthetic worked examples.
- A clear audit contract: time window, evidence cutoff, expected schedule and completion deadline.
- Reconciled counts for success, failure, cancellation, unresolved work and missing or unknown evidence.
- Observed coverage, delivery and on-time rates with explicit denominators.
- Retry recovery and latency analysis, with traceable source rows and prioritized investigation steps.
How to use it
Ask your compatible agent to audit your recurring job logs. Provide a redacted export or pasted records, expected slots or an unambiguous schedule, time zone, cutoff and deadline. The agent produces the report; you decide which investigations or changes to perform.
Boundaries
This is an instruction skill, not a hosted monitoring service or executable scheduler integration. It does not connect accounts, rerun jobs, repair systems, or guarantee uptime. It has no required paid API. Large exports need an analysis-capable agent with suitable local tools. Unknown export completeness is reported explicitly; missing log rows are not automatically called missed executions.
Validation
Frontmatter and package validation passed. Worked-example arithmetic was checked with Python. An independent agent applied the instructions to a separate synthetic case covering a queued retry, an exact duplicate, a missing slot, and an early completion. The demonstration below is synthetic, not customer production data. Results depend on the quality of supplied evidence and the agent used.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Testing & Debugging.
Read-only instruction skill, not continuous monitoring or an executable integration. Requires usable logs, correlation keys, an expected schedule, a time zone and deadlines for slot-level metrics. Incomplete exports cannot prove missed executions. Ambiguous cron/DST schedules need confirmed expected slots. Large exports require local analysis tools. Outputs depend on supplied evidence and the agent used.
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 9 days ago
Permissions
Read-only analysis of user-provided logs. File reading is needed only for supplied files; terminal/local analysis is optional for larger exports. No network, credentials, environment variables, production writes or scheduler access are required.