Content
80%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, highly actionable script for a version bump that makes excellent use of tokens, with a complete target-file list, sensible exclusions, and a lockfile refresh step. The main gap is the absence of a post-edit validation step for what is a batch operation, and the version-string formats per file are left for the model to discover.
Suggestions
Add a verification step after the edits, e.g., "Run `grep -rn '<old-version>' . --exclude-dir=.git` and confirm zero matches (excluding lock files) before running `uv lock`" — this closes the feedback loop for an 11-file batch change.
Show the expected version-string form for at least the non-obvious files (e.g., `__version__ = "0.4.11"` in `cli/planoai/__init__.py`, `APP_VERSION = "0.4.11"` in `consts.py`, the badge/URL in `Hero.tsx`) so replacements are deterministic instead of discovered per file.
Clarify bump-type semantics, e.g., what `major`, `minor`, and `patch` map to when read from `cli/planoai/__init__.py`, since the argument parsing is currently described only by example.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes competence: no explanation of what versioning is, no library comparisons, no padding — every line is an instruction, an exclusion, or a file path. Matches the "every token earns its place" anchor. | 5 / 5 |
Actionability | Concrete guidance throughout: an exact 11-file list, an explicit example ("0.4.11 → 0.4.12"), a runnable command ("cd cli && uv lock"), and explicit exclusions ("Do NOT change version strings in *.lock files or Cargo.lock"). Not 5 because the core edit operation is left implicit — the expected version-string format per file (e.g., `__version__ = "0.4.11"` in `__init__.py` vs. `version: 0.4.11` in CI YAML) is never shown, so each replacement requires discovery. | 4 / 5 |
Workflow Clarity | The sequence is clear (resolve version, with user confirmation when auto-incrementing → update all files → run `uv lock` → show summary), but this is a batch operation across 11 files with no validation step to catch missed or malformed replacements (e.g., grep for the old version afterwards). Per the rubric guideline, a batch workflow without validation/verification is capped at 3. | 3 / 5 |
Progressive Disclosure | The skill is under 50 lines, single-purpose, and needs no external references; the body is organized into clearly separated concerns (argument handling, file list, exclusions, lockfile update, summary) with no monolithic wall of text and no buried references. Per the rubric's simple-skill guideline, this qualifies for 5. | 5 / 5 |
Total | 17 / 20 Passed |