Content
92%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.
A tight, well-sequenced verification workflow with concrete commands, checklists, and a fix-and-re-run feedback loop. The only notable gap is the unspecified mechanism for discovering the project's lint/type-check/test commands in Step 3.
Suggestions
In Step 3, add a concrete discovery step for the project's check commands, e.g. 'If unknown, check package.json scripts / Makefile / CI config for lint, typecheck, and test entries'.
Replace the 'pattern' placeholder in the grep example with a realistic example pattern so the command is copy-paste ready.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean throughout: no explanations of concepts Claude already knows, and every section is executable instruction, checklist, or compressed guidance ('the index is a pointer, not the goal'). Every token earns its place. | 5 / 5 |
Actionability | Mostly executable with concrete commands (git diff, get_context.py --mode packages, cat spec index), but Step 3 ('Run the project's lint, type-check, and test commands') gives no way to discover those commands, and 'grep -r "pattern" src/' is a placeholder rather than a ready command. | 4 / 5 |
Workflow Clarity | Clear Step 1-6 sequence with explicit validation checkpoints ('Fix any failures before proceeding'), conditional-scope guidance ('Skip this step if your change is confined to a single layer'), checklists, and a fix-and-re-run feedback loop ('Re-run project checks after fixes') — all non-destructive, so no cap applies. | 5 / 5 |
Progressive Disclosure | A single self-contained file with well-organized sections and no nested references; the only pointers (.trellis spec index and its guideline files) are one level deep and clearly signaled ('read the spec index and follow its Quality Check section'). Content is appropriately inlined at this length. | 5 / 5 |
Total | 19 / 20 Passed |