Content
35%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill is a verbose, abstract orchestration template for incident response that reads more like a process document than actionable Claude instructions. It extensively describes concepts Claude already understands (SRE practices, postmortems, observability) without providing concrete, executable guidance. The workflow structure is reasonable but lacks the validation checkpoints critical for high-stakes incident response operations.
Suggestions
Remove explanatory content Claude already knows (severity level definitions, incident types, SRE concepts, extended thinking block) and focus on the specific decision logic and tool invocations unique to this workflow.
Replace abstract 'Use Task tool with subagent_type=...' instructions with concrete, executable examples showing actual tool calls with realistic inputs and expected outputs.
Add explicit validation checkpoints between phases (e.g., 'Before proceeding to Phase 2, confirm: mitigation is active, user impact is reduced, incident doc is created') to create proper feedback loops for this high-stakes workflow.
Move configuration details (severity levels, incident types, coordination protocols, success criteria) into separate reference files and link to them from the main skill to improve progressive disclosure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is heavily verbose with extensive explanations Claude already knows (SRE concepts, what observability is, what a postmortem is). The extended thinking block is unnecessary filler. Severity level definitions, incident types, and coordination protocols are standard knowledge that don't need this level of detail. The description itself is redundantly 'incident response incident response'. | 2 / 5 |
Actionability | Despite listing many steps, the guidance is entirely abstract — it describes what to do conceptually ('Use Task tool with subagent_type=...') but provides no executable code, no actual commands, no real tool invocations, and no concrete examples. The prompts are templates with $ARGUMENTS placeholders but no demonstration of actual usage. There's no copy-paste ready content. | 2 / 5 |
Workflow Clarity | The phases are clearly sequenced (Detection → Investigation → Resolution → Communication → Postmortem) with numbered steps, which is good. However, there are no explicit validation checkpoints or feedback loops — step 3 mentions rollback assessment but doesn't define criteria for when to rollback vs proceed. The deployment step mentions rollback triggers but doesn't specify them. For a destructive/high-stakes workflow like incident response, the lack of explicit validation gates caps this at 3. | 3 / 5 |
Progressive Disclosure | The skill references `resources/implementation-playbook.md` but no bundle files are provided, so we can't verify it exists. The massive amount of content (13 detailed steps, configuration, success criteria, coordination protocols) is all inlined in a single file when much of it (severity definitions, communication templates, coordination protocols) could be split into separate reference files. Some structure exists via phase headers but the file is monolithic. | 3 / 5 |
Total | 10 / 20 Passed |