Conciseness

The body is lean and assumes competence: every line carries Open-SWE-specific facts Claude cannot know ('The old sandbox is detached, not deleted', 'Never write a secret to disk — the proxy injects git auth per run'). Nothing explains concepts Claude already knows and nothing is padded, matching the score-5 anchor rather than the score-4 anchor's 'minor instances of over-explanation'.

Actionability

Guidance is fully executable for an instruction-only skill: exact tool invocations with parameters ('publish_workspace', 'recreate_sandbox(source="base")', 'snapshotctl(action="edit")', 'background_task("status", task_id)'), exact paths ('/open-swe/environment/logs/', '/workspace/<repo>', 'OPENSWE_WORKSPACE_REPOS'), and concrete script requirements ('set -euo pipefail', install rg, gh). The common cases (edit / fork / from scratch) are each covered with specific next actions, matching the score-5 anchor.

Workflow Clarity

Multi-step processes are clearly sequenced with explicit validation checkpoints and feedback loops: 'when it works you call publish_workspace', the nightly replay 'replaces the image only if it exits 0', and on failure 'Fix the script with snapshotctl(action="edit") … then rebuild again' while 'the previous image stays in place'. This matches the score-5 anchor (explicit validation, error-recovery loops) rather than score 4, whose checkpoints would be merely 'most' present.

Progressive Disclosure

Sections are well-organized and clearly navigable (routing table, per-topic sections, no nested references), and no bundle files exist so nothing can be offloaded. However, the ~85-line body exceeds the under-50-line simple-skill case and keeps all reference-style detail (refresh stages, log locations, snapshot naming) inline, fitting the score-4 anchor ('good structure; most content appropriately placed; minor organization gaps') better than score 5.

Specificity

The description enumerates concrete actions — 'create one, edit or fork an existing one, start from scratch, what setup_script and update_script are for, why a nightly refresh failed, where the build logs are, and how to read a rebuild in progress' — comprehensively covering the domain's operations. It exceeds the score-4 anchor ('minor gaps in coverage') because every major capability of the skill is explicitly named.

Completeness

It explicitly answers 'what' ('How an Open SWE workspace's sandbox image works and how to change it') and 'when' ('Read this whenever someone asks about workspaces, snapshots, sandbox images, or a refresh, whether or not this is an admin thread'). Both are present with concrete trigger phrases, matching the score-5 anchor exactly; the score-4 anchor requires a 'when' that 'could be more explicit', which does not apply here.

Trigger Term Quality

Trigger terms are comprehensive and natural: 'workspaces, snapshots, sandbox images, or a refresh', plus 'nightly refresh', 'rebuild', and 'build logs', with synonyms (sandbox image / snapshot / environment) covered. These are exactly the words a user in this environment would say when needing the skill, matching the score-5 anchor rather than the score-4 anchor's 'a few natural terms missing'.

Distinctiveness Conflict Risk

'Open SWE workspace's sandbox image' defines a clear niche with distinct triggers (workspaces, snapshots, refresh) and minimal conflict risk with other skills. It is well above the score-4 anchor ('minor overlap risk with closely related skills') because no other common skill would claim these triggers.