Tierward exposes 10 tools (5 primary + 5 deprecated aliases) with minimal parameter complexity but significant documentation gaps. All tools are read-only and stateless, which is positive. However, descriptions lack LLM-optimized guidance on WHEN to use each tool vs. alternatives, parameter descriptions are sparse or missing, and output schemas are not formally documented. Tool names follow verb_noun convention (tierward_doctor_report, tierward_team_settings) but descriptions are terse (avg ~100 chars, below the 194-char baseline for A+ tools). No error handling guidance, no recovery paths, and deprecated aliases dilute the interface without clear migration messaging.
Deprecated alias for tierward_arch_audit_status. Returns the timestamp of the last `arch-audit` skill execution recorded in `.claude/session/last-arch-audit`, plus age in days.
Deprecated alias for tierward_doctor_report. Runs `tierward doctor --report` against the current project and returns the JSON compliance summary (28 checks: governance files, hooks, skill spec, drift markers).
Deprecated alias for tierward_pr_review. Reads the audit trail of `/pr-review` skill comments on a GitHub PR (verdict, severity counts, body preview).
Deprecated alias for tierward_skill_inventory. Lists every skill installed in `.claude/skills/`, with frontmatter snapshot (description, model, allowed-tools, user-invocable).
Deprecated alias for tierward_team_settings. Returns the parsed `.claude/team-settings.json` for the current project.
Returns the timestamp of the last `arch-audit` skill execution recorded in `.claude/session/last-arch-audit`, plus age in days. Useful to verify the weekly cadence is honoured.
Output schemas not documented. Tools return JSON but LLMs cannot infer field structure, types, or required fields. tierward_doctor_report returns '28 checks' but no schema shows what each check object contains.
Deprecated aliases (cdk_*) clutter the interface without migration guidance. LLMs must reason about which version to use; no deprecation notice or redirect path provided.
Descriptions lack WHEN-to-use guidance. tierward_arch_audit_status and tierward_team_settings are similar discovery tools but descriptions do not explain the distinction or when to call each.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 48 | 2026-07-28+ | v2 |
Runs `tierward doctor --report` against the current project and returns the JSON compliance summary (28 checks: governance files, hooks, skill spec, drift markers).
Reads the audit trail of `/pr-review` skill comments on a GitHub PR (verdict, severity counts, body preview). Read-only: this tool does not run a fresh review. To generate a new review, invoke the `/pr-review` Tierward skill via the CLI (the tool returns the exact invocation in `cliInvocation`). Requires the `gh` CLI to be authenticated.
Lists every skill installed in `.claude/skills/`, with frontmatter snapshot (description, model, allowed-tools, user-invocable). Custom skills are flagged separately.
Returns the parsed `.claude/team-settings.json` for the current project. The `present` flag indicates whether the file exists; when absent, the project is unrestricted by Tierward governance.
tierward_pr_review requires 'gh' CLI authentication but no error handling documents what happens if gh is not installed or user is not authenticated. No recovery guidance.
Parameter descriptions missing or minimal. tierward_pr_review accepts prNumber but description does not specify format (positive integer, range, or example). No validation constraints documented.