Loading ToolBench

Reading the current corpus…

oneshot MCP server · F · ToolBench

← Browse

oneshot

github.com/shyamsridhar123/oneshot

AI-Powered Professional Services Engagement Platform with multi-agent social media orchestration

2026-07-28+STREAMABLE_HTTP10 tools

48/100

Definition Quality53

Protocol Readiness52

Supportability36

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence · evidence: Streamable HTTP

Current-spec patterns detected

Summary

OneShot is an agent-framework-based social media command center with 10 tools. All tools have basic descriptions (120-180 chars average), but input schemas are either absent or inferred from middleware files. Parameter descriptions are minimal or missing. No documented output schemas. Tools appear to operate on social media analysis workflows but lack the structured input validation and LLM-optimized descriptions needed for reliable agent reasoning. The codebase shows tool references in middleware/advisor/analyst files but does not expose explicit MCP tool registration with JSON Schema, schemas are inferred from docstrings. This caps per-tool scores significantly and prevents high definition scores.

Tools (10)

analyze_hashtagsread onlysource verified57/100

Analyze hashtag performance and recommend optimal hashtag strategy.

calculate_engagement_metricsread onlysource verified47/100

Calculate engagement metrics and provide quantitative analysis.

get_brand_guidelinesread onlysource verified42/100

Retrieve brand guidelines and style standards.

get_content_calendarread onlysource verified55/100

Retrieve the content calendar and scheduled posts.

get_past_postsread onlysource verified55/100

Retrieve past social media posts and their engagement metrics.

recommend_posting_scheduleread onlysource verified52/100

Generate posting schedule recommendations based on engagement patterns.

search_competitor_contentread onlysource verified55/100

Search and analyze competitor social media content.

Issues (7)

CRITICALtool ↗

No visible MCP tool registration with explicit JSON Schema. Tool definitions are inferred from middleware files, not formally registered. This violates the tool registration pattern and makes schemas unverifiable.

search_trendsanalyze_hashtagssearch_competitor_contentsearch_webget_brand_guidelinesget_past_postsget_content_calendarcalculate_engagement_metricsrecommend_posting_schedulesearch_knowledge_base

CRITICALconstrained-input ↗

Input parameters lack descriptions in schema. 'platform' parameter appears in multiple tools but is not documented with valid values (linkedin, twitter, instagram, all). LLMs will guess at valid options.

search_trendsanalyze_hashtagssearch_competitor_contentget_past_postsrecommend_posting_schedule

HIGHresponse-shaper ↗

No output schemas documented. Tools return data but LLMs have no structured specification of return types, fields, or format. This blocks proper chaining and data extraction.

Recommendations

Explicitly register all 10 tools with MCP using proper JSON Schema for inputs and documented output schemas. File: backend/app/agents/middleware.py, advisor.py, analyst.py, memory.py. Add JSON Schema definitions for each tool's input/output.
Add enum constraints for 'platform' parameters. Valid values: ['linkedin', 'twitter', 'instagram', 'all']. Document this in parameter descriptions.
Document output schema for each tool. Example for search_trends: {type: 'object', properties: {trends: {type: 'array', items: {type: 'string'}}, platforms: {type: 'array'}, timestamp: {type: 'string', format: 'date-time'}}}
Expand tool descriptions to include 'When to use' guidance. E.g., search_trends: 'Use when the user wants to discover current trending topics for content inspiration. Distinct from search_web (general web search) and search_knowledge_base (internal knowledge). Returns trending topics per platform.'
Add parameter descriptions for all inputs. For 'platform': 'Target social media platform. Valid: linkedin, twitter, instagram, or all to search across all platforms.'
Specify the structure expected for calculate_engagement_metrics 'data' parameter: what fields must it contain? (e.g., dates, view_counts, click_counts, share_counts). Provide a schema example.
Add error handling guidance to tool descriptions. E.g., search_competitor_content: 'If competitor not found, returns empty results. Try search_web() with the competitor name to verify it exists first.'
Document pagination for tools returning lists (get_past_posts, search_knowledge_base). Add 'limit' and 'offset' parameters and specify max results per call.

Score history

Overall score trend

↑ 16 points across a rubric change (v1 → v2)

48/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-22	F	48	2026-07-28+	v2
2026-03-09	F	32	-	v1