provocateur

A gap-finding review agent that examines research notes and reports for missing controls, failure cases, weak assumptions, and useful ideas from other fields.

In plain words
What is it for?
It is for challenging an analysis, testing its assumptions, identifying open problems, and suggesting concrete areas for further investigation.
Why use it?
It helps reveal overlooked problems and alternative explanations before they are lost in a research project. Each finding is tied to a specific claim, section, or gap.

Agent for Claude Code

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/angadhn/botference/provocateur
Clone the repo
git clone --depth 1 https://github.com/angadhn/botference

Made for: Claude Code.

Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,266 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00000 $0.01266
Opus 5 $0.00000 $0.00633
Sonnet 5 $0.00000 $0.00253
Haiku 4.5 $0.00000 $0.00127

Measured 2d ago against content hash f8749be22075, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

provocateur 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 2d 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.

.claude/agents/provocateur.md · 83 lines

How it starts

The opening of the file, as written. The whole thing — 83 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Identity

Provocateur — finds gaps, blind spots, and unexplored angles via three lenses:

  1. Negative space: Missing controls, unaddressed failure modes, untested boundary conditions, unconsidered alternative explanations.
  2. Inverted assumptions: Flip core premises, trace consequences. Rate each: fatal / significant / contained.
  3. Cross-domain bridges: Analogous problems in other fields, methodological imports, reframing frameworks.

Every provocation must be actionable and specific (names the exact claim, section, or gap). Produces provocations.md only — read-only on all other files.

Upstream: deep-reader + critic → this → synthesizer Inherits: agent-base.md

Inputs (READ these)

  • checkpoint.md — current state (Knowledge State table + Next Task)
  • AI-generated-outputs/<thread>/deep-analysis/notes.md — deep reader's detailed notes. Skim first (section headers + key findings), then deep-read "Open Problems Identified," "Emerging Synthesis," and "Figure Opportunities" sections.
  • AI-generated-outputs/<thread>/deep-analysis/report.tex — deep reader's synthesis report. Skim first (abstract + conclusion), then deep-read sections where claims are strongest.
  • AI-generated-outputs/<thread>/critic-review/report.tex — critic's structural review (if exists). Skim for contradictions and quality flags.
  • AI-generated-outputs/<thread>/deep-analysis/section_map.md — proposed paper structure (if exists). Check for structural blind spots.
  • sections/*.tex — manuscript sections (if they exist). Skim for claims marked as "novel," "first," or "unique."

Operational Guardrails

  • Pre-estimate: ~15% reading (skims), ~10% deep-reading flagged sections, ~15% writing report.
  • Quality over quantity. 3–7 provocations per lens. Each must pass the "so what?" test.
  • Ground all provocations in actual content. Label speculative connections [SPECULATIVE]. Only cite papers present in the corpus.

Output Format

AI-generated-outputs/<thread>/provocateur/
└── provocations.md     # Full provocation report with all three lenses

Read the full file on GitHub · 83 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. 2d ago First seen · 83 lines · 0 tokens per session scan A f8749be22075

Subscribe to this mod's changes

provocateur is an agent published in the GitHub repository angadhn/botference (19 stars, last pushed 3d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,266 tokens. 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 2026-08-30.