red-team

red-team is a command for Claude Code from equity-rigor/us-equity-research. It costs 74 tokens per session (1,457 once invoked), scanned A, original, MIT.

A review command for checking an investment memo about a US-listed company. It runs 20 verification checks and grades the memo using categories for common research and modelling errors.

In plain words
What is it for?
Use it to review an existing ticker-specific memo and its structured data files. It reports failed checks, an overall score, and practical changes needed to reach a higher score.
Why use it?
It helps identify calculation mistakes, missing evidence, and weaknesses in an analyst's argument before the memo is shared or used for an investment decision.

Command for Claude Code

Written for Claude Code: allowed-tools in frontmatter.

Runs only inside its plugin — its command needs a path that Claude Code sets for a plugin’s own hooks and for nothing else. Install the plugin, not this.

Part of the us-equity-ic-rigor plugin — 1 skill, 2 commands shipped together

Good fit Use it to review an existing ticker-specific memo and its structured data files. It reports failed checks, an overall score, and practical changes needed to reach a higher score.

Compare 6 commands from other repositories ↓
Install

Getting it into your agent

This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.

Claude Code
/plugin marketplace add equity-rigor/us-equity-research
Claude Code
/plugin install us-equity-ic-rigor

Made for: Claude Code.

Or install us-equity-ic-rigor, the plugin that ships this one along with the rest of its 1 skill, 2 commands.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for red-team

README.md
[![agentmods](https://agentmods.dev/badge/commands/equity-rigor/us-equity-research/red-team.svg)](https://agentmods.dev/commands/equity-rigor/us-equity-research/red-team)
Your own site
<a href="https://agentmods.dev/commands/equity-rigor/us-equity-research/red-team"><img src="https://agentmods.dev/badge/commands/equity-rigor/us-equity-research/red-team.svg" alt="Measured on agentmods" height="20"></a>
Per session 74 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,457 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00074 $0.01457
Opus 5 $0.00037 $0.00728
Sonnet 5 $0.00015 $0.00291
Haiku 4.5 $0.00007 $0.00146

Measured 8d ago against content hash c47142eb9776, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade A, and why

red-team scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

us-equity-ic-rigor/commands/red-team.md · 58 lines

How it starts

The opening of the file, as written. The whole thing — 58 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Load the us-equity-ic-rigor skill and run Phase 4 (PM red-team) on the existing memo for the ticker provided.

If a ticker is provided, look for outputs/{TICKER}_IC_memo.md + outputs/{TICKER}_structured.json + outputs/{TICKER}_manifest.json (v0.5.0+ consolidated convention). For grandfathered v0.4.x memos, also accept the split-file form with separate outputs/{TICKER}_scenarios.json + outputs/{TICKER}_source_tags.json sidecar files. If any required file is missing, instruct the user to run /us-equity-ic-rigor:ic-memo {TICKER} first — DO NOT silently re-run Phase 0-3.

If a target score is provided (e.g. "8.5" or "9.0"), use it as the gate threshold. Otherwise default to ≥8.5 per D20.

Workflow per references/pm-redteam-rubric-us.md Phase 4:

Step 1 — Mechanical gate sweep: Run all 20 ${CLAUDE_PLUGIN_ROOT}/scripts/verify_*.py against the structured JSON outputs. Report pass/fail per gate with the specific line/section that failed. The 20 gates:

  • G1 — EPS × multiple rows multiply exactly (no rounding lies)
  • G2 — Segment GM reconciles to consolidated within tolerance (±50bp industrials / ±5bp banks NIM per D24)
  • G3 — SOTP monotonicity (NI ≤ Pre-Tax ≤ OP ≤ GP for industrials; NI ≤ Pre-Tax ≤ PPNR ≤ Total Revenue for banks per D24)
  • G4 — Scenario probabilities sum to exactly 1.00
  • G5 — Bear EPS bridge reconciles to base via line-item delta walk
  • G6 — No unsourced specifics (every number tagged S1-S5 or Pending)
  • G7 — Headline conditionality: if any top-3 anchor is S3 or weaker, headline must be source-conditional
  • G8 — No mixed GM definitions (consolidated vs segment vs adj-ex-one-time used consistently within a row); GM taxonomy box T1-T5 present
  • G9 — What-would-reverse triggers have numerical denominators (not "deterioration in X" but "X declines >Y% over Z quarters")
  • G10 — Anchor weighting impact table present + A0 tail map shifts sum to zero per tail event
  • G11 — Non-GAAP figures accompanied by GAAP reconciliation (US-specific, B11)
  • G12 — SBC included in FCF or flagged explicitly excluded + buyback-offset-to-SBC ratio (US-specific, B12)
  • G13 — Barra factor exposure stated as numeric z-scores (Value/Quality/Momentum/Growth/Size/Low-Vol/Liquidity, B13)
  • G14 — Capacity / ADV / days-to-exit at 10/20/30% participation stated + max_position_constrained_by_adv (B14)
  • G15 — Consensus variance declared with sized scenario impact OR headline self-labels "consensus-anchored" (v0.2.0, B15)
  • G16 — Bank discipline (AOCI + CET1 + NIM + stress capital) when sector=Banks (v0.2.0, B16)
  • G17 — Earnings revision velocity disclosed when n_analysts ≥ 5 (v0.2.0, B17)
  • G18 — Quant overlay cross-document consistency: prose factor mentions match structured block ±0.2 (v0.3.0)
  • G19 — Plugin 1 → Plugin 2 provenance manifest with SHA-256 file-hash integrity (v0.3.0; caps memo at 7.5 on fail or hand_authored=true)
  • G20 — View defensibility: for any score >8.5, three conjunctive conditions (differentiation magnitude ≥8pp, S1/S2 evidence on load-bearing variance, surviving variance_attack with isolation + model diversity); v0.4.0 graduated rigor for >9.0 (v0.3.0+)

If a verifier script cannot be reached (file not found, permission denied, import error), DO NOT degrade to LLM-analytical gate evaluation. STOP, report the error to the user with the exact command and exact error message, and follow the SKILL.md anti-degradation preamble protocol.

Step 2 — Rubric score: Score the memo 1-10 across all B1-B17 bug categories from pm-redteam-rubric-us.md. Report the lowest-scoring band as the bottleneck. Memo passes iff overall ≥ target (default 8.5). Mechanical-math gate failures (G1, G3, G15, G16) cap the score at 7.0; revision velocity (G17) caps at 7.5; provenance (G19) caps at 7.5 on fail or hand_authored=true; view defensibility (G20) caps at 8.5 on fail.

Step 3 — Push-from-N-to-N+1 fixes: For each bug category scoring below target, produce a specific actionable edit (filename, line number, before/after diff). Order by score-gain-per-effort.

Read the full file on GitHub · 58 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 8d ago First seen · 58 lines · 74 tokens per session scan A c47142eb9776

Subscribe to this mod's changes

red-team is a command published in the GitHub repository equity-rigor/us-equity-research (4 stars, last pushed 1mo ago), licensed MIT. It adds 74 tokens to every session and 1,457 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.