blindoracle

ops.agent-workforce-report

Deterministic Agent Workforce Early-Warning Report over YOUR OWN Claude Code transcript bundle, computed with the method of Anthropic's ‘How employment is changing at Anthropic’ (2026-10-16). Delivered as a written Markdown report (output, also at /v1/services/result/{job_id}?format=md) with one row per metric: your value | Anthropic's figure | what it means, plus the structured metrics JSON. Metrics: agent:human hours (40 h/week parity, pro-rated to the observed span when the bundle is short, and measured engaged time), human-wait share with the > 30 min wait exclusion, reading load (words/day at 250 wpm), meta-work share (deterministic path rule), skill use; and, when you also send them, R6 code survival by author class from git_log, decision latency from decisions, and the R5 standing-instruction inventory from instructions. Anything not supplied is listed with the export command. Zero LLM, zero network; inputs are processed in memory and only their sha256 is retained — no message text, paths, shas, subjects or ids are echoed. The full report carries a signed ProofOfAuditReport (kind 30105) binding the report hash and your bundle hash, and sha256(audit_hash|input_sha256|proof_id) is anchored in a 0-value Base transaction after the response (status at /v1/services/result/{job_id}/anchor, free). Pass tier: diagnostic for the R1+R2 subset, or buy the $0.25 ops.agent-workforce-diagnostic rung first. Bundle cap: 4 MB per request on this endpoint. Sending more than 4 MB of transcripts, or do not want to send them at all? Run the free local extractor https://craigmbrown.com/blindoracle/tools/workforce_extract.py (sha256 34f33389a564555e4d4476b86982feff10ca53eb6143bbc38139798d8dd09417) and send its summary instead of records: counts only, no message text, any number of days, identical numbers. The report then states that it was computed from a buyer-run extraction and whether your extractor matches the published hash; the attestation proves that input produced this output, not that the input is complete. (x402: $25.00 USDC per call)

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": "ops.agent-workforce-report"
  }
}

Input Schema

Name	Required	Description	Default
`task`	No	Task/query text for this SKU

Schema Changelog

Changes observed during successful MCP inspections.

AddedOct 25, 2026

TDQS

A4.2/5.0

Behavior4/5

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

Beyond the annotations it discloses pricing ($25.00 USDC x402 per call), the 4 MB request cap, in-memory processing with only sha256 retained, the signed ProofOfAuditReport (kind 30105) and the post-response Base-chain anchor with a status endpoint. The one soft spot is the 'zero network' claim, which sits awkwardly beside openWorldHint=true and the on-chain anchoring step; it is a tension, not a contradiction, since the computation itself is local.

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

Conciseness3/5

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

The purpose is front-loaded in sentence one and the structure moves sensibly from method to metrics to pricing, but the prose is heavily overloaded: an embedded extractor URL with its full sha256, an anchor status endpoint, and parity/pro-rating minutiae crowd out the core invocation guidance. Several clauses could be moved to linked documentation without losing anything an agent needs to call it.

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?

For a complex, paid, no-output-schema tool, the definition covers what is returned (Markdown report in `output`, structured `metrics` JSON, a retrieval path), the attestation guarantees, the input size limit, the fallback path, and the privacy contract. An agent has enough to decide, prepare inputs, and interpret results without guessing.

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

Parameters2/5

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

The schema exposes exactly one field, `task`, described circularly as 'Task/query text for this SKU', while the description directs the caller to pass `tier: diagnostic`, `records`, `summary`, `git_log`, `decisions`, and `instructions` — none of which exist in the schema. The semantic detail is rich and genuinely useful about what data to send, but an agent cannot map any of it onto the actual parameter surface.

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 first sentence names a specific verb+resource (deterministic early-warning report over your own Claude Code transcript bundle), states the method it follows, and enumerates the metrics produced. It also names the sibling ladder explicitly (ops.agent-workforce-diagnostic, `tier: diagnostic` for the R1+R2 subset), so an agent can separate the two without opening either schema.

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

Usage Guidelines5/5

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

It gives explicit routing conditions: send the bundle here if under 4 MB, otherwise run the free local extractor and send `summary` instead of `records`, or buy the $0.25 diagnostic rung first. It also states when the R6/R5/decision-latency metrics are available (only when `git_log`, `decisions`, `instructions` are supplied) and that missing inputs are listed with the export command.

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

All blindoracle tools Inspect

Try in Browser

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.