القائمة
遇到任何 bug、测试失败或异常行为时使用，在提出修复方案之前执行；要求先定位根因、再做最小修复、最后用新鲜证据验证。
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Use this skill when the Agent needs to decide where a capability, rule, workflow, runtime anchor, script, job, document, persona change, plugin asset, or repository maintenance item should live. It provides a layered map of the Agent's durable assets so requests to “落实到 all”, “整理自身框架”, or “细化自身” do not bloat memory or scatter files.
Use this skill when the user asks to 自检自身、重新蒸馏、进化版 Agent、增强回去、避免死循环处理、能力增益评估，或在瘦身与增强之间取舍。它先盘点已处理对象、能力风险和真正增益点，再决定增强、修复、拆分、瘦身或停止，不把行数减少当成默认目标。
Use this skill when the Agent needs an executive-control layer before acting: classify user intent, decide direct execution vs read-only inspection vs button confirmation vs clarification, avoid over-asking, and keep work moving to the smallest safe completed state. Trigger especially when the user says 继续、落地、放开手脚、接管、自己判断、按方案推进，or complains about typing confirmations that should be buttons.
Use this skill when the Agent needs to read feedback signals before deciding whether to act, correct itself, distill a lesson, or ask for clarification. It treats user wording, repeated failures, tool errors, missing evidence, and activity context as sensory inputs, then routes to the right organ without overreacting.
Use this skill when the Agent needs to turn experience into a repeatable growth loop: observe signals, decide whether learning is needed, distill reusable lessons, place them into the right asset layer, verify the change, and keep future behavior lighter rather than heavier. Trigger when the user asks to 主动成长、持续进化、经验闭环、复盘后改进, or when repeated failures show a missing habit rather than a one-off bug.
Use this skill when the Agent needs to distinguish safe self-improvement from high-risk mutation. It guards credentials, destructive actions, MoviePilot core files, downloads, restarts, external service changes, and over-broad memory edits while still allowing authorized low-risk /config/agent and /config local plugin work to proceed.
| version | 3 |
| name | systematic-debugging |
| description | 遇到任何 bug、测试失败或异常行为时使用，在提出修复方案之前执行；要求先定位根因、再做最小修复、最后用新鲜证据验证。 |
| allowed-tools | execute_command read_file list_directory task subagent_task |
Use this skill whenever something is failing, flaky, inconsistent, or surprising. The goal is to fix the root cause with evidence, not to stack guesses.
Core rule:
No root-cause evidence, no fix claim.
Use for:
Do not use it to delay exact low-risk work that already has a known fix and verification path.
Read the failure
Reproduce or confirm state
Localize the layer
Identify where the chain breaks: configuration, authentication, scheduler, plugin, API, database, downloader, transfer, media recognition, filesystem, or Agent skill/router.
For multi-component flows, check boundaries in order and stop at the first unsupported assumption.
Parallel boundary probe: when the failure could span multiple independent
layers (e.g. site + downloader + scheduler + logs), dispatch parallel
read-only probes via subagent_task instead of serial tool calls:
| Suspected layer | Subagent | Probe |
|---|---|---|
| Site/auth | resource-searcher | query_sites + test_site |
| Downloader | download-diagnostician | query_downloaders + query_download_tasks |
| Scheduler/workflow | system-diagnostician | query_schedulers + query_workflows |
| Plugin/config | system-diagnostician | query_plugin_config + query_plugin_capabilities |
| Code/log | moviepilot-explorer | read source files + grep logs |
| Media/library | media-researcher | query_library_exists + query_transfer_history |
Synthesize all subagent results privately, then form one hypothesis from the combined evidence. Fall back to direct tool calls if subagent infrastructure fails.
Compare with a working reference
Form one hypothesis
Apply the smallest safe fix
Verify fresh
When the failure is an explicit error message, stack trace, command failure, or log exception, use this compact intake before hypothesizing:
Do not ask the user to paste more context if the current error already names a safe next diagnostic command or authoritative MoviePilot query.
After a recurring error is fixed, preserve only reusable prevention value:
A replay tag is useful only when it helps future routing or diagnosis; never create a second error database inside memory.
SKILL.md, memory/router references, stale platform terms, frontmatter, and minimal structure assertions.When the issue is performance, slowness, repeated retries, large output, or context bloat:
Do not import external benchmark frameworks unless the project already uses them or the user explicitly asks.
Treat regression evidence as a debugging input when:
Debug regressions by comparing current evidence with the last known working command, config, route, or file version. If no baseline exists, create a minimal baseline before changing behavior.
When there are many logs or repeated failures, group by stable fingerprints before fixing:
For MoviePilot Agent work, lightweight grep/log grouping is enough unless the user explicitly asks to integrate an external error platform.
When reporting debugging work, include: