Content
63%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-organized, token-efficient overview with clearly sequenced workflows, explicit output contracts, and sensible failure-handling rules. Its main weaknesses are executable incompleteness — the bundled Python tools it depends on are missing from the bundle and no run commands appear in the body — and reference paths that lead to nonexistent files.
Suggestions
Ship the missing `tools/edgar.py` and `tools/market.py` in the bundle (or remove the 'bundled Python tools' claims and reference paths) — the /analyze, /score, and /compare workflows currently fail at step 1.
Add concrete invocation examples to the body (e.g. `python3 tools/edgar.py AAPL` and `python3 tools/market.py AAPL`) so the workflow is executable from SKILL.md alone.
Enumerate the 'minimum required EDGAR and market fields' for /score inline or point to the exact section of references/scoring.md that defines them.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~95-line body is lean: numbered command steps, tight execution rules, and output shapes, with no explanation of concepts Claude already knows (no 'what is EDGAR' or finance tutorials). Minor trimmable bits remain — the intro line duplicates the frontmatter description and the branded footer/compliance sections are overhead — so it sits just below the 'every token earns its place' anchor. | 4 / 5 |
Actionability | Steps are concrete in shape (named files, named references, explicit output sections), but the executable layer is broken: `tools/edgar.py` and `tools/market.py` are referenced throughout yet no tools/ directory exists in the bundle, and the body contains no actual invocation commands or enumeration of the 'minimum required fields' for /score. This is 'some concrete guidance but incomplete; missing key details' rather than mostly-executable level 4. | 3 / 5 |
Workflow Clarity | Each command has a clear numbered sequence, explicit output-format checklists, and an error-recovery rule ('If a tool call fails, state exactly what data is missing and continue with available inputs. Do not hallucinate missing figures'). The operations are read-only research so the destructive-operation cap does not apply, but validation is implicit rather than an explicit checkpointed step, keeping it below level 5. | 4 / 5 |
Progressive Disclosure | The three references (methodology.md, scoring.md, edgar-guide.md) exist, are clearly signaled in dedicated sections, and are one level deep — but scoring against the actual bundle shows the two most load-bearing referenced paths (`tools/edgar.py`, `tools/market.py`) do not exist anywhere in the bundle, and the body never mentions the four bundled PNG assets. Navigation is well-signaled for what exists, but the broken tool references are more than a 'minor organization gap', matching level 3 better than level 4. | 3 / 5 |
Total | 14 / 20 Passed |