Content
85%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 well-engineered router skill: obsessive about safety gates (site confirmation, env confirmation, dry-run, deployment gate, verification) with excellent progressive disclosure — all 19 referenced files exist, navigation is one level deep, and loading discipline is explicitly instructed. The main weaknesses are mild redundancy (the tcb-deploy warning and access.md routing rows are repeated) and the absence of one end-to-end executable example in the body itself.
Suggestions
Consolidate the three mentions of avoiding `tcb deploy` (Core Principle 8, the `references/app.md` routing row, and the intro) into a single authoritative statement, and merge the duplicate `references/access.md` routing rows into one.
Add one short end-to-end executable example (login with site flag → `tcb env use` → a representative deploy command with real flags) in the Quick workflow so the most common path is complete without opening a reference.
Trim the Minimum self-check checklist to items not already enforced by Core Principles, or drop the checklist in favor of a pointer to the principles, to recover tokens without losing information.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with non-obvious operational facts ("Absence from help ≠ deprecated", "tcb --help itself is a hand-maintained static list") and avoids explaining concepts Claude already knows. It falls at 4 rather than 5 because of redundant repetition: "do not recommend `tcb deploy`" appears in Core Principle 8, the routing row for `references/app.md`, and the description; `references/access.md` occupies three routing rows; and the self-check checklist largely restates the Core Principles. | 4 / 5 |
Actionability | Concrete, executable commands are present throughout ("tcb login --intl true|false", "tcb config set isIntl true|false", "tcb env use <envId>", "tcb docs search \"<keyword>\"", "--dry-run", "check the exit code (`$?`)") plus a five-step Quick workflow. It is not a 5 because the body never shows one complete executable example session (e.g., a full login → env use → function deploy invocation with real flags) — the actual syntax is deferred to references, leaving minor gaps for the most common path. | 4 / 5 |
Workflow Clarity | The Quick workflow is a clearly sequenced five-step process with explicit validation checkpoints ("confirm envId with the user", "with `--dry-run` for destructive ops", "Verify the result with the corresponding `list` / `detail` command"), a feedback loop for error recovery ("If a command fails after 2-3 attempts, check the exit code (`$?`), read the error message, consult `tcb docs search`, and try a different approach"), and a Minimum self-check checklist. This matches the 5 anchor's sequence + validation + recovery loop + checklist; the destructive-operation validation cap does not apply because dry-run and verification are both mandated. | 5 / 5 |
Progressive Disclosure | The body is a pure overview/router: a 20-row routing table maps every task to one of 19 reference files, all of which exist in the bundle, and the reference index enumerates them for reachability. References are one level deep (cross-links inside reference files are lateral routing hints, e.g. access.md pointing a storage-ACL task to storage.md, not deferred content), and the "How to use this skill" section explicitly enforces loading discipline: "Load only the reference files that match the user's task" and "Stop loading more context once you have the workflow and command syntax". This matches the 5 anchor's clear overview with well-signaled one-level references. | 5 / 5 |
Total | 18 / 20 Passed |