Content
86%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is an exemplar of lean, command-driven skill writing: concrete CLI invocations with full payloads, a clearly sequenced task lifecycle with failure paths, and no wasted tokens. The only real gap is unstated provenance for the claim_token and a few undefined criteria ('unblocked assignment'), which slightly limit copy-paste executability.
Suggestions
State where the claim_token comes from, e.g. 'Use the claim_token returned by claim-task in subsequent transition commands.'
Define what makes an inbox assignment 'unblocked' or instruct the worker to skip assignments lacking a pending task file.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and efficient with zero padding: every section ('When to use', 'Inputs and preconditions', 'Operational steps', 'Exit and evidence') carries operational content, no concepts Claude already knows are explained, and every token earns its place. | 5 / 5 |
Actionability | Steps provide copy-paste-ready `omx team api` commands with complete JSON payloads and exact state-file paths, which is mostly fully-executable guidance. Minor gaps: the `claim_token` required in step 6 is never sourced (presumably the claim-task output, but this is unstated), and 'take the first unblocked assignment' leaves the blocked/unblocked criterion undefined, keeping it just below the fully-executable anchor. | 4 / 5 |
Workflow Clarity | The 8-step sequence is clearly ordered with a claim-before-edit checkpoint, explicit failure paths ('Complete or fail through the lifecycle API', 'release-task-claim only to requeue a blocked task'), and defined observability requirements. Minor validation gaps remain: no instruction to confirm the ACK or claim-task command succeeded before proceeding, and no error-recovery loop if a transition fails. | 4 / 5 |
Progressive Disclosure | The skill is under 50 lines with well-organized sections and no need for external reference files, so per the simple-skill guideline it scores 5 on structure alone; the single external pointer (`AGENTS.md#durable-runtime-invariants-canonical-ssot`) is clearly signaled and one level deep. | 5 / 5 |
Total | 18 / 20 Passed |