协作头脑风暴
brainstorming
在实现前澄清需求、探索设计选择,并形成清晰的开发方向。
techsmith-performance-tuning【开发与工程】Optimize Snagit capture and Camtasia project/export throughput from measured workstation bottlenecks and safe concurrency limits. Use when queues are slow, unstable, or resource-bound. Trigger with "TechSmith performance", "Camtasia export tuning", or "Snagit capture throughput".
独立 GitHub 快照 · 自动分类 · 以原作者文档为准
name: techsmith-performance-tuning description: >- Optimize Snagit capture and Camtasia project/export throughput from measured workstation bottlenecks and safe concurrency limits. Use when queues are slow, unstable, or resource-bound. Trigger with "TechSmith performance", "Camtasia export tuning", or "Snagit capture throughput". argument-hint: "[workload-manifest] [workstation-alias]" allowed-tools: Read, Glob, Grep, WebFetch, Write, Edit version: 1.6.0 author: Jeremy Longshore jeremy@intentsolutions.io license: MIT tags:
This skill improves throughput without assuming server-style horizontal scale. It measures one representative workload, separates capture from export bottlenecks, keeps active Camtasia data local, and changes one resource or concurrency control at a time.
Use Read, Glob, and Grep to inspect local scripts, manifests, logs, and tests. Use WebFetch only for current primary TechSmith documentation. Use Write or Edit only after confirming the target repository file and approval boundary.
TechSmith desktop activation is not API authentication. Resolve individual sign-in versus business-key or approved offline activation before execution. Redact all keys, account identifiers, activation artifacts, and sensitive endpoint details.
Do not edit projects on cloud/network storage, disable security software, force undocumented command flags, or launch duplicate exporters to manufacture concurrency.
Return baseline and candidate metrics, bottleneck, change, before/after delta, quality checks, safe worker limit, rollback, and monitoring threshold.
| Condition | Response |
|---|---|
| Workload is not reproducible | Freeze inputs and preset before drawing conclusions. |
| Disk saturation | Reduce workers and stage active files on approved local storage. |
| Output differs | Reject the optimization and restore the baseline configuration. |
| Crash rate increases | Rollback immediately and preserve redacted diagnostics. |
The example below shows the minimum redacted evidence expected from a successful invocation of this operator workflow.
workload=training-20m; workers=1->2; duration=-28%; peak_mem=82%; quality=pass; decision=keep