Content
65%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A highly actionable CLI reference with copy-paste commands and clear examples, weakened by redundancy between the command table, detailed sections, and Key Principles, and by inline content that could be deferred to reference files. Non-destructive workflows are well sequenced but lack explicit validation checkpoints.
Suggestions
Remove duplication: drop the repeated "Single-profile only" / "-o json" points from Key Principles or from the command table, keeping each fact in one place to tighten conciseness.
Add a brief validation/confirmation step to the investigation workflow (e.g., check the artifact list is non-empty before calling `artifacts get`, or confirm the chat ID was returned before the follow-up) to reach the validated workflow anchor.
Move the model catalog and detailed artifact spill/decompression behavior into a reference file (e.g., references/olly-models.md) referenced one level deep, leaving the overview lean and improving progressive disclosure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient and free of concept padding Claude already knows, but it repeats content: the command table restates the commands later expanded in "Chat Commands"/"Artifacts", and "Key Principles" reiterates points already made ("Single-profile only", "-o json for scripting"). It could be tightened, so it sits at the 2 anchor rather than the lean 3. | 2 / 3 |
Actionability | Commands are fully executable and copy-paste ready ("cx olly ask", "cx olly artifacts list -o json | jq '.[0].id'", "--chat-id", "--model", "--timeout"), with concrete examples matching the copy-paste-ready anchor. | 3 / 3 |
Workflow Clarity | The "Investigate an issue" example gives a clear linear sequence (ask -> capture chat ID -> follow up -> list/get artifacts), but there are no explicit validation checkpoints or feedback loops, fitting the "sequence present but checkpoints missing or implicit" anchor rather than the validated 3. | 2 / 3 |
Progressive Disclosure | The ~125-line body is well-sectioned but entirely inline with no bundle files or one-level-deep references to defer detail (e.g., model catalog, artifact spill behavior), matching the "some structure but content that could be separate is inline" anchor; the under-50-line simple-skill exception does not apply. | 2 / 3 |
Total | 9 / 12 Passed |