Loading ToolBench
PostSider MCP server - gives AI agents full access to the PostSider platform over its public API (list channels, schedule/publish posts, analytics, media).
PostSider MCP has 12 well-named tools with consistent verb_noun patterns (list_, get_, find_, create_, pause_). All tools have descriptions (avg ~150 chars, within baseline 34-392). However, input schemas are minimal: most tools accept only 1-2 simple parameters (string, number) with no enum constraints, min/max bounds, or format specifications. Output schemas are entirely undocumented, LLMs cannot predict response structure. Error handling is absent: no recovery guidance, no categorization of retryable vs fatal errors. The pause_publishing tool (IRREVERSIBLE risk) lacks confirmation/dry-run pattern. Parameter descriptions are present but terse (avg ~40 chars vs baseline 72). No batch operations despite tools like list_posts that agents might call repeatedly.
Return the next free scheduling date-time (UTC) for a channel, based on its configured posting queue. Use the returned value as the `date` for postsider_create_post when scheduling into the queue.
Get an organization-wide operational overview for an agency: clients, channels, queued posts, drafts, published posts, errors, recent errors and pending approvals. Useful for morning checks and client reporting.
Get account-level analytics for a connected channel (where the provider supports it).
Get a customer-scoped agency report with channels, queued, draft, published, error and pending approval counts.
List recent notifications for the organization (e.g. publish failures, channels needing reconnection).
No output schemas documented. LLMs cannot predict response structure (fields, types, nesting). Agents must guess what data is available after each call, risking failed downstream tool chains.
Input parameters lack constraints (enums, min/max, format, regex). E.g., 'days' parameter in get_agency_overview accepts any number with no bounds, LLM could pass 999999 or -1. No validation guidance.
postsider_pause_publishing (IRREVERSIBLE) lacks confirmation/dry-run pattern. Agents can halt all publishing with a single call, no undo via API. Should require explicit confirmation or support a dry-run mode.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 56 | 2026-07-28+ | v2 |
Get the full organization-scoped post group, including current state, scheduled time, media, channel and publish error. Use this to inspect the result of an asynchronous create or publish operation.
Get performance analytics for a single post over the last N days (where the provider supports it).
Return per-channel validation problems / missing required fields for a post, so they can be fixed before publishing.
List the social channels (integrations) connected to this PostSider organization. Returns each channel id, name and platform. Call this first to get the channel ids needed by postsider_create_post.
List configured channel groups (sets of channels that are published to together).
List posts scheduled or published within a date range (UTC). Useful for reviewing the content calendar before scheduling more.
Immediately halt ALL publishing for this organization (kill switch): no `now` or `schedule` posts can be created, and queued posts are parked to HELD instead of going out. Use when something is wrong — a PR crisis, a post on the wrong channel, a runaway automation loop. Resume is human-only (owner, dashboard); this cannot be undone via the API.
No error handling guidance. Tools return no recovery hints, error categorization, or actionable messages. LLMs cannot distinguish retryable failures from user-fixable or fatal errors.
Parameter descriptions are terse (avg ~40 chars vs baseline 72). E.g., 'days' in get_agency_overview lacks guidance on valid range, default, or interpretation. 'customer' in list_posts is optional but no hint on how to discover valid customer IDs.