Skip to main content
Glama
io.github.xngg1021/academic-research-kernel

io.github.xngg1021/academic-research-kernel

by xngg1021

academic_recompute_statistics

Recompute reported statistical results—p-values, effect sizes, confidence intervals, power, and sample sizes—to verify consistency and flag mismatches in research papers.

Instructions

Recompute t/F/chi2/z p-values, two-group Cohen's d and Hedges' g, 2x2 OR/RR, CI consistency, statistical power, SD bounds, and df-implied sample sizes. Optional reported p-value comparison uses decimal round-half-up intervals; input or partial execution failures are tool errors, statistical mismatches are successful results.

Tool call template

This tool declares no required parameters. Add optional parameters from the input schema as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "academic_recompute_statistics"
  }
}

Input Schema

NameRequiredDescriptionDefault
dfNoDegrees of freedom (integer or Welch fractional)
n1NoGroup 1 size
n2NoGroup 2 size
seNoReported standard error SE (> 0)
df1NoNumerator degrees of freedom for F-test
df2NoDenominator degrees of freedom for F-test
sd1NoGroup 1 SD
sd2NoGroup 2 SD
betaNoReported regression coefficient beta
alphaNoSignificance level alpha in (0, 1), default 0.05
mean1NoGroup 1 mean
mean2NoGroup 2 mean
cell_aNo2x2 table exposed/treated event count
cell_bNo2x2 table exposed/treated non-event count
cell_cNo2x2 table control event count
cell_dNo2x2 table control non-event count
f_statNoReported F-statistic (right-tail ANOVA/regression)
t_statNoReported t-statistic
chi2_dfNoDegrees of freedom for chi-square test
df_kindNoTest family for df-to-N check
p_valueNoReported p-value
check_sdNoReported SD for bounded-scale feasibility check
ci_levelNoConfidence level in (0, 1), default 0.95
ci_lowerNoLower bound of confidence interval
ci_upperNoUpper bound of confidence interval
n_paramsNoNumber of regression parameters (including intercept) when df_kind='regression'
power_n1NoGroup 1 sample size for achieved power
power_n2NoOptional Group 2 sample size for achieved power
sample_nNoOptional sample size for bounded-scale SD check
chi2_statNoReported chi-square statistic (right-tail)
ci_methodNoOptional interval construction method (e.g. wald, clopper-pearson, wilson, bootstrap)
sample_dfNoReported degrees of freedom for N-consistency check
scale_maxNoMaximum possible value of the bounded scale
scale_minNoMinimum possible value of the bounded scale
ci_rel_tolNoRelative midpoint tolerance, default 0.02
reported_nNoReported total sample size N
ci_estimateNoPoint estimate for CI consistency check
ablation_tolNoTolerance for constant delta check, default 1e-9
ci_log_scaleNoUse geometric midpoint for ratio estimates (OR/RR/HR)
sample_ratioNoGroup size ratio n2/n1 for required sample size, default 1.0
target_powerNoTarget statistical power in (0, 1) for required sample size
effect_size_dNoCohen's d for t-test power or sample-size calculation
partition_tolNoAllowed integer rounding tolerance for partition sum, default 0
ablation_matrixNo2D numeric matrix (rows=methods/tasks, cols=settings) for constant-offset regularity check
p_value_literalNoOptional exact reported decimal/scientific literal, preserving trailing zeros; must agree with p_value and any explicit decimals.
partition_partsNoSubset counts (e.g. [train, val, test])
partition_totalNoDeclared total count for dataset split conservation check
p_value_decimalsNoOptional decimal-place precision; must agree with a supplied literal.
ablation_min_rowsNoMinimum row count for ablation delta check, default 3
bandwidth_rel_tolNoRelative tolerance for storage size comparison, default 0.01
bandwidth_max_gbpsNoHardware memory bandwidth ceiling in GB/s, default 4000.0
bandwidth_item_countNoTotal vector/item count for storage & bandwidth bound check
bandwidth_vector_dimNoEmbedding dimension for storage & bandwidth bound check
partition_share_unitNoUnit for reported shares, default auto
ablation_baseline_colNoBaseline column index for ablation delta check, default 0
bandwidth_storage_unitNoUnit for reported storage size, default auto
partition_share_tol_pctNoAllowed percentage-point tolerance for reported shares, default 0.5
partition_reported_sharesNoOptional reported subset shares or percentages
bandwidth_reported_literalNoOptional exact reported storage literal string preserving decimal precision
bandwidth_reported_size_mbNoOptional reported storage footprint in MB or MiB
bandwidth_bytes_per_elementNoBytes per element (default 4 for float32)
bandwidth_reported_decimalsNoOptional explicit decimal places for reported storage size
bandwidth_claims_full_systemNoWhether reported size claims to include full system text/metadata, default true
bandwidth_reported_latency_usNoOptional reported full-scan latency in microseconds
bandwidth_min_metadata_overhead_bytesNoOptional minimum provable non-vector metadata overhead in bytes

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.1.0

TDQS

B3.1/5.0
Behavior3/5

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

With no annotations, the description carries the full burden, and it does contribute genuinely useful behavior: input or partial execution failures surface as tool errors while statistical mismatches return as successful results, and p-value comparison uses decimal round-half-up intervals. That is exactly the kind of context an agent needs to interpret outcomes. It stops short of covering return shape, how many checks run per call, or what happens when only a subset of the 65 optional parameters is supplied.

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

Conciseness4/5

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

Two sentences, capability list front-loaded, no filler or marketing language. The enumeration is dense but each item names a distinct check, so it earns its space; the error-semantics clause in sentence two is the highest-value content and is not buried too deep.

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

Completeness2/5

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

For a 65-parameter, zero-required, no-annotations, no-output-schema tool, the description is too thin. It never explains which parameters must be supplied together to trigger a given check (e.g. p_value requires the matching statistic and df), nor what a successful response contains. An agent would have to reverse-engineer invocation patterns from the schema alone.

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?

Schema description coverage is 100%, so the schema already explains each of the 65 parameters in detail and the baseline of 3 applies. The description adds meaning only for the p-value literal/decimals comparison path and the error-vs-mismatch distinction. It does not group the parameters into the check families named in sentence one, which is where added value would matter most.

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

Purpose4/5

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

Opens with a specific verb ('Recompute') and enumerates the exact statistical quantities handled: t/F/chi2/z p-values, Cohen's d, Hedges' g, 2x2 OR/RR, CI consistency, power, SD bounds, and df-implied N. That is concrete and far from tautological. It does not differentiate itself from siblings like academic_check_percentage or academic_meta_analyze, which is the only thing keeping it from a 5.

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

Usage Guidelines2/5

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

There is no 'use when' statement, no indication of which parameter combinations activate which check family, and no named alternative for overlapping verification tasks. The second sentence is about error semantics, not usage selection. An agent knows what the tool can compute but not how to decide to call it over a sibling.

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