01Safety
30.0 / 30
What changed in the harness
Selection accuracy 100→100, token cost up 2%, unconfirmed writes 100%→100%.
Category breakdown
Where the score comes from.
Earned points across the four signals Gradable measures. Safety and Legibility are scored out of 30; Economics and Discoverability are scored out of 20.
02Legibility
28.8 / 30
03Economics
17.1 / 20
04Discoverability
20.0 / 20
Highest-impact fix
Estimated gain +3 pointsReduce the tool-definition and task token payload
Consolidate overlapping tools and trim repetitive schema text while preserving behavior, parameter, and return semantics.
Description evidence
Defects and rewrites.
3 defects found across the exposed tool descriptions. Suggested rewrites make purpose, inputs, boundaries, and returns easier for an agent to understand.
| Tool | Defect types | Suggested rewrite |
|---|---|---|
search_mcp_servers |
params_unexplained |
Search Remote OpenClaw's directory of 13,870+ MCP servers for Claude Code, OpenClaw, Hermes, Codex, and Cursor. 'query' is the search text (e.g. 'postgres database'); 'limit' caps the number of results returned. Returns names, descriptions, links, and popularity. |
search_skills |
params_unexplained |
Search Remote OpenClaw's directory of 4,384+ agent skills for Claude Code, OpenClaw, Hermes, and Codex. 'query' is the search text (e.g. 'pdf' or 'react'); 'limit' caps the number of results returned. Returns names, descriptions, links, and install commands. |
search_plugins |
params_unexplained |
Search Remote OpenClaw's directory of Claude Code / OpenClaw / Hermes / Codex plugins. 'query' is the search text (e.g. 'testing'); 'limit' caps the number of results returned. Returns names, descriptions, and links. |
Selection evidence
Confusable tool pairs.
3 pairs where similar names or overlapping descriptions may send an agent toward the wrong tool.
| Tool A | Tool B | Confidence | Why they collide |
|---|---|---|---|
search_skills |
search_plugins |
medium | Skills and plugins are related extension concepts for the same platforms; a query like 'find something for pdf handling' could plausibly be satisfied by either a skill or a plugin, causing the agent to pick the wrong search tool. |
search_mcp_servers |
search_plugins |
low | MCP servers and plugins are distinct enough (server integration vs. packaged extension) that most tasks clearly indicate one, though a vague 'find a postgres tool' request could theoretically be routed to either. |
search_mcp_servers |
search_skills |
low | MCP servers and skills serve different purposes (external integrations vs. task capabilities), so confusion is less likely, but generic queries like 'search for react support' could ambiguously map to either. |
Compare the field