Content
88%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.
This is a dense, highly operational orchestration skill: exact commands, literal templates, prioritized decision rules, and validation checkpoints throughout, with essentially no token waste or explanation of things Claude already knows. Its only structural weakness is that everything lives in one long file with some deliberate-looking but mild redundancy across sections.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is telegraphic and assumes Claude's competence: no basic concepts are explained, and every section carries operational rules rather than padding. A few spots could still be tightened, such as 'Each round is a fresh agent with no memory of the last, so the issue, the branch and the pull request are the only record' and slight redundancy between 'What done looks like' item 2 and invariant 5 ('one branch per issue, one open PR per issue'). | 4 / 5 |
Actionability | Guidance is copy-paste ready throughout: exact commands with flags ('gh issue view <n> --json number,title,body,url,state,labels,assignees,author,comments', 'git ls-remote --exit-code --heads origin tessl/<id>', 'gh pr create --base <base> --label tessl --title ...'), exact comment templates with literal marker strings, and defined placeholders (<n>, <id>, <base>). | 5 / 5 |
Workflow Clarity | The multi-step process is explicitly sequenced: a prioritized situation list, the Canonical Workflow (plan, then build), and the ordered 'Doing the work' steps (base, branch, setup, tests, before the PR, PR, link). Validation checkpoints are explicit and repeated: 'Read it again before every write', 'run it and see it fail for the reason you expect', re-running tests after merging the base, and the error-recovery loop of 'gh pr list --head <branch> --state open' before retrying a failed 'gh pr create'. | 5 / 5 |
Progressive Disclosure | The body is well sectioned with clear headers (inputs, situations, invariants, workflow, doing the work, judgment) and no buried or nested references, but it is a ~170-line monolith with no bundle files: the skill exceeds the 50-line threshold at which well-organized sections alone earn a 5, and minor cross-section overlap (done-state vs invariants) leaves small organization gaps. | 4 / 5 |
Total | 18 / 20 Passed |