Content
71%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-structured routing hub with unusually concrete operational specifics (site matrix, env vars, API action semantics) and excellent progressive disclosure into a verified reference bundle. Its main weakness is redundancy: the MCP-unavailable fallback is repeated across at least five sections, and a few named tools are referenced without any pointer to their source.
Suggestions
State the MCP-missing fallback rule once (e.g., in the 'MCP + CLI prerequisite' section) and reference it from Step 0, Workflow 2a, Activation Contract, Global rules, and Engineering constitution instead of restating it five times.
Merge overlapping guidance between 'Global rules before action' and 'Engineering constitution' (both restate the MCP-vs-CLI preference and backend-before-frontend rule).
Clarify dangling tool references — 'cloudbase-code-review', 'agent-browser', and 'searchKnowledgeBase(mode="skill")' are invoked without saying whether they are MCP tools, sibling skills, or plugin commands, which risks a dead end when they are absent from the session.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Most content is dense and non-obvious (site/env-variable specifics, GroupName vs model-id rules, EnvId alias resolution), but the MCP-missing fallback rule is restated at least five times (Step 0 bullets, Workflow 2a, Activation Contract, "Global rules", "MCP + CLI prerequisite", and "Engineering constitution"), and "Global rules before action" overlaps the "Engineering constitution" section. This fits the level-3 anchor (mostly efficient but could be tightened) better than level 2, since no space is spent explaining concepts Claude already knows. | 3 / 5 |
Actionability | Guidance is largely executable: exact endpoints ("https://tcb-api.cloud.tencent.com/mcp/v1"), exact env vars ("TCB_SITE=intl + TCB_REGION=ap-singapore"), copy-paste commands ("npx plugins add TencentCloudBase/cloudbase-plugin -y --scope user", "npx mcporter list | grep cloudbase"), and concrete MCP actions ("queryEnv(action=list, alias=..., aliasExact=true)", "manageApps(action=\"deployApp\", ...)"). Minor gaps keep it below 5: a few referenced tools are dangling in this bundle ("cloudbase-code-review", "agent-browser", "searchKnowledgeBase(mode=\"skill\")") with no pointer to where they come from. | 4 / 5 |
Workflow Clarity | The Workflow section gives a clear 3-stage sequence with explicit constraints ("Stage 2a must precede frontend code. Stage 3 is mandatory.") and a validation gate ("Run cloudbase-code-review, fix errors, declare done"), plus error-recovery loops ("After 2–3 failed attempts on the same path, stop and reroute"; serialize-then-retry-once for object writes). It sits at level 4 rather than 5 because the close-out validation depends on a tool not defined in this bundle and the self-verify steps are named at a high level rather than as explicit checkpoints. | 4 / 5 |
Progressive Disclosure | The body is a genuine router/overview: "On-demand references" lists each of the 7 bundle files with a one-line description, "Load only when needed (do not expand this entry)" signals deferral, all 7 referenced files exist in references/ and are one level deep (verified: "No npm/npx" section in tooling-fallback.md and §5 in deployment-workflow.md exist as cited), and scenario specifics are delegated to child skills ("Keep scenario-specific pitfalls in child skills — do not expand this entry file"). This matches the level-5 anchor: clear overview, well-signaled one-level-deep references, easy navigation. | 5 / 5 |
Total | 16 / 20 Passed |