Works With Agents MCP Server
# Works With Agents MCP Server
> **⚠️ Maintenance mode.** Feature-complete, not actively developed. For the Works With Agents tender-compliance MCP server, see [workswithagents.dev](https://workswithagents.dev). This package is frozen at v1.0.2 and is no longer receiving updates.
14 MCP tools for AI agent infrastructure.
```bash
pip install wwa-mcp
```
## Usage
Add to your MCP client config:
```json
{
"mcpServers": {
"wwa": {
"command": "python3.11",
"args": ["-m", "wwa_mcp.server"]
}
}
}
```
## Tools
| Tool | Tier | Description |
|------|------|-------------|
| `wwa_facts_query` | 1 | Query shared agent knowledge base |
| `wwa_facts_list_categories` | 1 | List fact categories and counts |
| `wwa_pitfalls_lookup` | 1 | Search known failure patterns |
| `wwa_skills_search` | 1 | Search agent skill library |
| `wwa_skills_list_categories` | 1 | List skill categories |
| `wwa_handoff_create` | 2 | Create agent-to-agent handoff |
| `wwa_handoff_verify` | 2 | Validate handoff completeness |
| `wwa_handoff_validate_schema` | 2 | Schema validation against Handoff Protocol |
| `wwa_blueprint_search` | 2 | Search blueprint specifications |
| `wwa_blueprint_validate` | 2 | Validate against a blueprint |
| `wwa_trust_score_calculate` | 3 | Compute agent trust score |
| `wwa_identity_verify` | 3 | Ed25519 signature verification |
| `wwa_sla_validate` | 3 | Validate against SLA tiers |
| `wwa_compliance_check` | 3 | Compliance-as-Code validation engine |
All specifications CC BY 4.0 — [workswithagents.dev](https://workswithagents.dev/specs/)
TDQS
Scored across 22 tools
Most tools target clearly distinct resources and actions, with the wwa_ prefix helping to group them. However, wwa_handoff_verify and wwa_handoff_validate_schema both validate handoffs and could be confused, as could wwa_bid_search and wwa_bid_list despite their differing purposes.
All tools share the wwa_ prefix, and the majority follow a noun_verb pattern (e.g., wwa_bid_search, wwa_sla_validate). The main deviation is wwa_generate_key, which uses verb_noun ordering, and a few compound names like wwa_facts_list_categories add minor complexity.
With 22 tools, the server is on the heavier side (16-25 tools per the rubric) but not extreme. The wide range of domains (tenders, bids, facts, skills, handoffs, blueprints, etc.) makes the count feel somewhat bloated, though each tool does have a distinct role.
Several resources have incomplete lifecycles: bids can be indexed and retrieved but not updated or deleted; facts and skills only support querying/list categories, with no create/update/delete; handoffs have create and validation but no retrieval or management operations. These gaps are notable but not critical, as the server may focus on specific workflows rather than full CRUD.