Hesper Atlas Evidence

Rolling walk-forward evidence

get_walk_forward_evidence

Read-onlyIdempotent

Use when a user asks for stricter time-ordered evidence, weak years, benchmark underperformance, or a check against look-ahead. Returns the committed cache-frozen annual walk-forward artifact. Each year's candidate is selected using only earlier observations and then traded for the next calendar year with next-open execution and costs. Returns the content-derived evidence id, input hash manifest, aggregate results, and paged per-symbol/year rows. Weak periods and underperformance versus buy-and-hold are retained. Historical validation, not live performance or advice.

Input Schema

TableJSON Schema

Name	Required	Description
`limit`	No	Rows to return, default 25, max 100.
`offset`	No	Rows to skip, default 0.
`symbol`	No	Ticker symbol, e.g. NVDA. Case-insensitive.
`include_inputs`	No	Include all committed input-file hashes; default false for compact responses.

Output Schema

TableJSON Schema

Name	Required	Description	Default
`as_of`	No
`source`	No
`caveats`	No
`is_live`	No
`citation`	No
`data_type`	No
`disclaimer`	Yes
`provenance`	No
`source_url`	No
`last_updated`	No
`calculation_version`	No
`methodology_version`	No

TDQS

A4.5/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already mark the tool read-only, idempotent, and non-destructive, and the description adds substantial context beyond that: the artifact is 'cache-frozen,' candidates use 'only earlier observations,' execution is 'next-open' with costs, and weak periods or underperformance are deliberately retained. This gives agents a clear model of what the operation does and does not do.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is front-loaded with usage triggers and stays compact despite covering purpose, methodology, output contents, and limitations. Every sentence contributes useful selection or invocation information without fluff.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the optional parameters, rich annotations, and an output schema, the description covers the full context an agent needs: when to invoke, what methodology guarantees are built in, what the response contains, and what the result is not. No critical gap is apparent.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Input-schema coverage is 100%, with each parameter already described including defaults, bounds, and case-insensitivity. The tool description does not add parameter-level meaning beyond the schema, so the baseline of 3 is appropriate.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description names a specific verb and resource: it 'Returns the committed cache-frozen annual walk-forward artifact' and ties the tool to concrete user intents like 'stricter time-ordered evidence, weak years, benchmark underperformance, or a check against look-ahead.' It also distinguishes itself from live performance or advice, which separates it from related validation or performance tools.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It explicitly opens with 'Use when a user asks for...' and gives four concrete trigger conditions. It also states an important boundary: 'Historical validation, not live performance or advice.' However, it does not name a sibling alternative or say when to prefer another evidence-related tool, so it stops short of the strongest possible routing guidance.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All Hesper Atlas Evidence tools Inspect

Try in Browser

RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

TDQS

A4/5.0

Disambiguation3/5

The five historical-performance tools (get_track_record, get_walk_forward_evidence, get_heldout_evidence, get_ledger_stats, list_closed_trades) occupy heavily overlapping territory, and an agent could easily grab the wrong evidence artifact. The detailed 'use when' openers help considerably, but the boundaries between replay, walk-forward, and validation are subtle enough that misselection risk remains real.

Naming Consistency4/5

All tools follow a clean snake_case verb_noun pattern, with get_ reserved for single artifacts/reports and list_ for enumerable collections. The convention is slightly loose—get_changes_since and get_ledger_stats are more list-like than get_-like, and the verbs don't always signal collection size—but overall the pattern is predictable and readable.

Tool Count4/5

At 15 tools, the server sits at the upper boundary of a well-scoped surface, and each tool does earn its place in the evidence ecosystem (current signal, lists, portfolio, context, diffs, publications, replay, validation, methodology, provenance). It is slightly heavy and could feel daunting, but nothing is redundant.

Completeness4/5

The surface covers the evidence domain thoroughly: current state, forward publication history, performance replays, two distinct validation artifacts, methodology, and hash-level provenance. Minor gaps exist—there is no tool to enumerate the full covered universe or search across tickers, forcing users to arrive with a symbol in mind—but core workflows have no dead ends.

Resources

Need Help?