# Tru8 Evidence Research

> Source-traced evidence research for AI agents. We organise the evidence; you decide.

- name: io.github.SamYatesSmith/tru8
- version: 1.0.4
- connection class: R1 (autonomous)
- trust: 93/100
- quality: 90/100 (strong)
- owner: not claimed
- descriptor: https://api.protogrid.dev/v1/servers/io.github.SamYatesSmith%2Ftru8
- tools: https://api.protogrid.dev/v1/servers/io.github.SamYatesSmith%2Ftru8/tools

**An agent can connect on its own if it already holds `X_API_KEY`, `X_API_KEY`.** Remote endpoint behind a static API key or header. The registry never sees the value; the agent fills the placeholder.

## Connection (mcpServers)

```json
{
  "mcpServers": {
    "tru8": {
      "type": "http",
      "url": "https://api.trueight.com/mcp/",
      "headers": {
        "X-API-Key": "${X_API_KEY}",
        "Authorization": "${X_API_KEY}"
      }
    }
  }
}
```

Placeholders to fill: `${X_API_KEY}`, `${X_API_KEY}`.

## Trust

- hygiene: 100
- liveness: 97
- freshness: 85
- provenance: 90
- drivers: +repository +namespace-matches-repo +website +reachable +uptime-93% ~updated-71d-ago
- Derived from observable signals (official registry feed and our own credential-free probes); no code audit performed.

## Quality 90/100 (strong)

### protocol: 75 (weight 25)

- warn `protocol.modern`: newest supported version is 2025-06-18; 2026-07-28 not supported, older versions are deprecated until 2027-07-28
- n/a `protocol.stateless`: only applies to 2026-07-28 servers
- pass `protocol.transport`: streamable HTTP
- n/a `protocol.list_ttl`: only applies to 2026-07-28 servers

### authorization: n/a (weight 15)

- n/a `auth.prm`, `auth.as_metadata`, `auth.cimd`: no remote uses OAuth
- n/a `auth.secret_in_url`: no templated URL

### tool hygiene: 93 (weight 30)

- pass `tools.descriptions`: every tool has a description
- warn `tools.description_length`: 1 tools have descriptions over 1,024 characters, which crowds the context
- pass `tools.schemas`: every tool has an object input schema
- pass `tools.annotations`: 3 of 3 tools declare readOnlyHint or destructiveHint
- pass `tools.directory_hints`: every tool declares readOnlyHint, destructiveHint, idempotentHint and openWorldHint
- pass `tools.token_cost`: about 1,236 tokens to load every tool
- pass `tools.api_dump`: tools are not a one-to-one API dump

### stability: 100 (weight 15)

- pass `stability.changes`: 1 tool changes in 30 days
- pass `stability.rug_pull`: no tool changed its meaning under the same name

### dependencies: 100 (weight 15)

- pass `deps.known_vulns`: no known advisory in 30 resolved packages
- pass `deps.mcp_sdk_version`: mcp 1.30.0, no known advisory
- pass `deps.resolvable`: tru8-mcp@1.0.3 resolved: 30 packages

- tools: 3, about 1,236 tokens to load them all
- tool set hash: 76e7e2499625cc3e050ec490ad3b4721
- tools last changed: 2026-10-12T12:05:26.702Z
- badge: [![protogrid quality](https://protogrid.dev/badge/io.github.SamYatesSmith/tru8.svg)](https://protogrid.dev/servers/io.github.SamYatesSmith/tru8)
- Checks run on what our credential-free, read-only probes observe; tools are never called and no code audit is performed.

### Tool changes

- 2026-10-12 `tru8_check` changed (description)
  - before: Evidence research for a factual claim or article URL. Submit a claim as text or paste an article URL. URLs are auto-detected and trigger article mode: the pipeline extracts claims from the page and auto-selects up to 5 for evidence research. Tiers (in fallback order): - lookup (~£0.02, instant) — cached prior analysis - consensus (~£0.03, instant) — cross-user aggregate landscape (k≥3 checks) - quick (~£0.07, ~15s) — web search + heuristic classification - full (~£0.15, ~60-90s) — web + specialist APIs, LLM classification, coverage recovery Charges based on tier actually executed, not tier requested. Identical calls within ten minutes return the same check, so a transport retry never runs — or charges — twice. Set max_tier to control maximum spend per call. Output structure: - claims[].claimMap.elements[] — verifiable sub-claims with state (supported/disputed/unresolved) - claims[].claimMap.elements[].evidenceRefs[] — evidence mapped to elements with relationship (supports/challenges/context) and reasoning - claims[].evidence[] — sources classified by tier (primary/reporting/ commentary) and type (data/official/news/analysis/opinion/academic) - claims[].claimMap.orientation — mechanical summary from element states - _meta — execution metadata: executedTier, chargedPence, limitations Per-argument guidance lives on the parameters themselves (see the Annotated/Field declarations above), which is where a client can actually read it — a docstring "Args:" block never reaches the tool's inputSchema.
  - after: Evidence research for a factual claim or article URL. Submit a claim as text or paste an article URL. URLs are auto-detected and trigger article mode: the pipeline extracts claims from the page and auto-selects up to 5 for evidence research. Tiers (in fallback order): - lookup (~£0.02, instant) — cached prior analysis - consensus (~£0.03, instant) — cross-user aggregate landscape (k≥3 checks) - quick (~£0.07, ~15s) — web search + heuristic classification - full (~£0.15, ~60-90s) — web + specialist APIs, LLM classification, coverage recovery Charges based on tier actually executed, not tier requested. Identical calls within ten minutes return the same check, so a transport retry never runs — or charges — twice. Set max_tier to control maximum spend per call. Output structure: - claims[].claimMap.elements[] — verifiable sub-claims with state (supported/disputed/unresolved) - claims[].claimMap.elements[].evidenceRefs[] — evidence mapped to elements with relationship (supports/challenges/context) and reasoning - claims[].evidence[] — sources classified by tier (primary/reporting/ commentary) and type (data/official/news/analysis/opinion/academic) - claims[].claimMap.orientation — mechanical summary from element states - _meta — execution metadata: executedTier, chargedPence, limitations A run that has not finished after several minutes returns {checkId, status, chargedPence, message} instead of the result: fetch it later with tru8_get_result (free). A run that FAILS raises an error and is not charged. Per-argument guidance lives on the parameters themselves (see the Annotated/Field declarations above), which is where a client can actually read it — a docstring "Args:" block never reaches the tool's inputSchema.

Full history: https://api.protogrid.dev/v1/servers/io.github.SamYatesSmith%2Ftru8/changes

## Dependencies

- pypi tru8-mcp@1.0.3: resolved, 29 dependencies, MCP SDK mcp 1.30.0

Dependency graphs from deps.dev; advisories from OSV.dev, including the GitHub Advisory Database and the PyPI Advisory Database; all CC BY 4.0.

## Remotes

- https://api.trueight.com/mcp/ (streamable-http, auth api_key ${X_API_KEY}, reachable true, uptime30d 0.9292035)

## Packages

- pypi tru8-mcp@1.0.3 secrets: TRU8_API_KEY

## Tools (3)

- `tru8_check`: Evidence research for a factual claim or article URL.

    Submit a claim as text or paste an article URL. URLs are auto-detected
    and trigger article mode: the pipeline extracts claims from the page
    and auto-selects up to 5 for evidence research.

    Tiers (in fallback order):
    - lookup    (~£0.02, instant) — cached prior analysis
    - consensus (~£0.03, instant) — cross-user aggregate landscape (k≥3 checks)
    - quick     (~£0.07, ~15s) — web search + heuristic classification
    - full      (~£0.15, ~60-90s) — web + specialist APIs, LLM classification, coverage recovery

    Charges based on tier actually executed, not tier requested.
    Identical calls within ten minutes return the same check, so a
    transport retry never runs — or charges — twice.
    Set max_tier to control maximum spend per call.

    Output structure:
    - claims[].claimMap.elements[] — verifiable sub-claims with state
      (supported/disputed/unresolved)
    - claims[].claimMap.elements[].evidenceRefs[] — evidence mapped to elements
      with relationship (supports/challenges/context) and reasoning
    - claims[].evidence[] — sources classified by tier (primary/reporting/
      commentary) and type (data/official/news/analysis/opinion/academic)
    - claims[].claimMap.orientation — mechanical summary from element states
    - _meta — execution metadata: executedTier, chargedPence, limitations

    A run that has not finished after several minutes returns
    {checkId, status, chargedPence, message} instead of the result: fetch it
    later with tru8_get_result (free). A run that FAILS raises an error and is
    not charged.

    Per-argument guidance lives on the parameters themselves (see the
    Annotated/Field declarations above), which is where a client can actually
    read it — a docstring "Args:" block never reaches the tool's inputSchema.
    
- `tru8_get_result`: Retrieve a previously submitted check with pre-computed analytics.

    Returns the full result including a _computed block with tier/type
    distributions, corroboration groups, diagnostic values, timeline
    analysis, element state summaries, and per-claim dispositions.

    Use this over tru8_get_result_raw when you want structured analytics
    ready for summarisation or comparison without post-processing.
    
- `tru8_get_result_raw`: Retrieve a previously submitted check without computed analytics.

    Returns claims, elements, evidence, and claim maps only. No _computed
    block. Smaller response payload. Use this when you will compute your
    own aggregations or only need specific fields from the raw data.