Distribb SEO Skill

The README identifies authentication, major API data flows, and consequential effects such as publishing, syncing, public review replies, Google posts, and Microworkers ratings, supporting a moderate transparency score. However, one API key appears to cover many high-impact actions without documented scopes, least-privilege roles, or revocation behavior. Some workflows require users to connect services or complete government-account steps, but there is no consistent per-action confirmation gate for publishing, rewriting, public replies, or autopublishing. Sensitive-data guidance stops at an environment variable and saving third-party keys in Distribb settings; encryption, redaction, retention, and deletion are not covered. Dependencies are unpinned and no supply-chain checks are described. Editing and syncing a published article are shown, but unpublishing, rollback, and recovery are not. The product and repository are named, yet no verifiable author, organizational ownership, or maintainer identity is supplied; unknown publisher identity is not treated as suspicious.

Commands, endpoints, plan restrictions, and the recommended SEO process are mostly internally consistent, and the legacy-account HTTP 402 behavior is described precisely. curl, jq, requests, python-dotenv, and external connection requirements are identified adequately for ordinary setup. Deductions apply because there are no version constraints, compatibility matrix, service-outage fallback, or dependency-health evidence. Failure handling is detailed only for the BYO-key case; authentication, networking, CMS, publication, and third-party integration failures lack diagnostics and recovery guidance.

The source thoroughly addresses scenarios spanning audits, keywords, writing, optimization, social media, GBP, video, government directories, and a 90-day sprint. It states several plan differences, required connections, legacy-account limitations, and tasks reserved for the user, but does not fully define data-access boundaries, platform-specific differences, or limits on high-impact automation. Most slash commands have clear names and arguments, and the sub-skill has explicit trigger phrases; broader commands such as /optimize, /gbp, and /backlinks remain underspecified. Claude command placement, generic curl/jq use, Python helpers, and MCP are covered, while the claimed Codex, Hermes, and OpenClaw environments lack platform-specific setup and compatibility evidence.

The quick start, process, command table, raw API examples, plans, topical sections, references, and helper descriptions form a strong information architecture. Installation and API-key instructions are usable, but system requirements, uninstalling, upgrades, and common setup failures are absent. Naming is generally stable, although “Agentic Mode” and deprecated “Free Agentic” are easy to confuse. There are many examples but no complete FAQ, and the referenced files were not supplied as evidence. Some legacy-plan and connection limitations are disclosed, though no consolidated limitations section exists. No license, version identifier, changelog, release policy, maintainer contact, security-reporting path, or explicit maintenance responsibility is shown, so those criteria receive no credit.

The source supplies adaptable curl requests, command entry points, an ordered workflow, and expected deliverables, making outputs reasonably usable. It does not provide representative output artifacts, acceptance criteria, or human-review procedures after failures. Combining GSC, keyword, content, backlink, CMS, and social workflows offers plausible marginal value, but outcome claims and assertions such as real or high-DR links are supported only by product copy. Pricing and plan distinctions are explicit, and optional helpers are separated from the no-install API path; deductions remain because model/API costs, review labor, third-party fees, and SEO or reputational risks are not quantified.

Many capabilities map to named commands and endpoints, and the README points to several repository references. Only the README is supplied here, however, so those references, implementation code, tests, and response contracts cannot be cross-checked. No independent or second-source evidence corroborates audit quality, live data, unlimited exchange, high-DR backlinks, autopublishing, or broad platform support. Marketing conclusions, intended capabilities, and observable facts are generally stated with equal certainty rather than separated into evidence, assumptions, and measured results, limiting all three verifiability scores.