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.
An exemplary orchestration skill: a situation-driven loop with explicit done-state, copy-paste-ready commands with exact fields and expected responses, and disciplined gating of every risky action. The weaknesses are minor: a few over-stuffed sentences that trade clarity for density, and a bundle with one orphaned file and some second-level reference chains.
Suggestions
Link references/progress.md from somewhere (or remove it) — it is currently unreachable from SKILL.md and every other reference file.
Flatten the second-level reference chains: surface triggers.md, building-blocks.md, and customize.md directly from SKILL.md (e.g. from the 'Suggested next steps' or a 'What else you can ask for' section) rather than only via starters.md.
Split the longest sentences for readability — invariant 10 (Commands) and invariant 1 pack several rules into single 80–100 word sentences; separate sentences or bullets would preserve the same rules at lower reading cost.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Nearly every line carries Tessl-specific, non-inferable knowledge (CLI quirks, API fields, timing behavior like 'keeps answering `missing` for up to five minutes'), with no padding about concepts Claude already knows. Not 5: a few passages are over-stuffed and would be clearer split up — e.g. the ~100-word invariant 10 sentence and the long parenthetical in invariant 1 — which is density rather than leanness. | 4 / 5 |
Actionability | Fully executable throughout: exact commands ('tessl code repo update <owner/repo> --org <org id> --review-enabled --review-trigger every-push', 'gh api -X PATCH repos/<owner>/<repo> -F has_issues=true'), exact JSON bodies ('{"workspaceId": ..., "sourceType": "git", "sourceUri": ...}'), exact response fields to read ('usage.credits.state', 'meta.issuesEnabled', 'data.attributes.status') and expected values for each. Common cases, including error and refusal paths, are covered with the exact command to run. | 5 / 5 |
Workflow Clarity | The situation loop is explicit — 'Read where things stand, take the first situation that applies, do its next action, and look again' — with a clear done-state, ordered situations 1–13, and validation checkpoints at each turn ('Confirm the Loops setup reported ... are listed and active', 'confirm with tessl code repo view ... --json'). Error-recovery feedback loops are present ('When it refuses, say the access can take a few minutes to show, and try again'; 'When either check gives no clear answer ... say you could not check it and go on'), and resume behavior is specified ('when starting and when resuming'). Risky operations (push, merge, turning on Issues) are gated on the person's explicit yes, so no validation cap applies. | 5 / 5 |
Progressive Disclosure | SKILL.md is a genuine overview with well-signaled, purpose-stated one-level references, and every anchor link verified against the bundle (all four headings in set-up-in-tessl.md exist). Not 5: the actual bundle structure has gaps — references/progress.md is orphaned (linked from neither the body nor any reference file), and triggers.md, building-blocks.md, customize.md, and confirm-loops.md are only reachable via second-level links through starters.md and testing.md, so some content sits two levels deep. | 4 / 5 |
Total | 18 / 20 Passed |