Content
75%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 a well-structured, actionable workflow with concrete CLI commands and explicit validation checkpoints for a batch artifact-generation process. It falls just short of top marks due to a duplicated constraint note, a thin error-recovery loop, and the absence of a concrete example artifact or reference structure.
Suggestions
Remove the duplicated 'context/rules are constraints for you, not file content' note by stating it once in the Artifact Creation Guidelines section and referencing it from step 4a.
Add an explicit error-recovery feedback loop for the validation step, e.g. 'if a file is missing or status is not done, fix the issue and re-run openspec status before proceeding'.
Include one short concrete example of a generated artifact (e.g. a minimal tasks.md skeleton) so artifact authoring is copy-paste ready rather than fully delegated to the CLI template.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is efficient with copy-paste commands and no padding about concepts Claude already knows, but the constraint that 'context' and 'rules' are for the agent and not for the file is stated twice (step 4a and the Artifact Creation Guidelines), a minor redundancy that keeps it below the lean anchor 5. | 4 / 5 |
Actionability | Provides concrete, executable CLI commands (openspec new, status, instructions) and specifies the JSON fields to parse, but the actual artifact-file authoring is delegated to the CLI's dynamic template output with no concrete example artifact shown, leaving a minor gap versus fully copy-paste-ready anchor 5. | 4 / 5 |
Workflow Clarity | Steps 1-5 are clearly numbered and sequenced with validation checkpoints (re-run openspec status after each artifact, verify each file exists), but the error-recovery path when verification fails is thin and not fully spelled out, short of the explicit feedback-loop anchor 5. | 4 / 5 |
Progressive Disclosure | Content is well-organized into clearly headed sections (Input, Steps, Output, Artifact Creation Guidelines, Guardrails) and appropriately self-contained since detail comes from the CLI, but with no bundle files and ~100 lines of inline content there is no one-level-deep reference structure to reach anchor 5. | 4 / 5 |
Total | 16 / 20 Passed |