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.
A densely operational, well-sequenced orchestration skill with excellent concrete tool-call specifications and thorough failure handling, but it crams an entire multi-protocol system into one 235-line file with over-dense prose. The main workflow is highly actionable; the weaknesses are readability (run-on validation sentences) and the absence of any progressive disclosure into reference files.
Suggestions
Split the secondary sub-protocols into one-level-deep reference files — e.g. references/synapse.md for intent refinement, references/conductor.md for attention handling, references/relay-events.md for the event vocabulary — leaving the core execution loop in SKILL.md.
Rewrite the step 5 sentinel-validation rules as a short fail-closed checklist (one failure mode per bullet) instead of a single ~130-word sentence; it currently obscures an otherwise clear sequence.
Give the Synapse `ouroboros_session_signal` and `ouroboros_record_conductor_decision` calls explicit argument blocks like the other tool calls, so all mutating paths are copy-paste ready rather than prose-described.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body contains no padding or explanations of concepts Claude already knows — every line is operational protocol it could not infer — but it is 235 lines of dense prose where significant tightening is possible: step 5's sentinel-validation rules form a single ~130-word sentence enumerating eight failure modes, and the catch-up/cursor rules are repeated across steps 4-6. It fits 'mostly efficient but could be tightened' rather than 'minor instances that could be trimmed'. | 3 / 5 |
Actionability | The main path is fully concrete with copy-paste-ready argument blocks (e.g. `ouroboros_job_wait` with `timeout_seconds: 180`, `view: "summary"`, `wait_for: "attention_or_ac_change"`; `ouroboros_start_execute_seed` with conditional-argument rules). Minor gaps remain: the Synapse `ouroboros_session_signal` call (step 3) and `ouroboros_record_conductor_decision` logging are described in prose without argument blocks, so it is 'mostly executable with minor gaps' rather than fully copy-paste ready. | 4 / 5 |
Workflow Clarity | A clear numbered sequence (steps 1-8) with genuinely explicit validation checkpoints and feedback loops: fail-closed contract validation, observer-failure fallback to catch-up, and 'If the evaluate job failed or timed out... show `Next: ooo evaluate <session_id>` as the manual retry'. It falls short of the top anchor because this complex process is presented as dense paragraphs (especially step 5) rather than checklists, which obscures the sequence at points; not a 3, since checkpoints are explicit and present throughout, not implicit. | 4 / 5 |
Progressive Disclosure | The body is well-sectioned (Usage, Instructions with numbered subsections, Fallback, footer) but there are no bundle files at all — no references/, scripts/, or assets/ — and all ~235 lines are inline, including secondary sub-protocols (Synapse intent refinement, Active Conductor handling, the relay event vocabulary) that clearly belong in separate one-level-deep reference files loaded only when needed. This matches 'some structure... content that should be separate is inline'; not a 2 because the section structure is genuinely good and navigation within the file is easy. | 3 / 5 |
Total | 14 / 20 Passed |