fire-research

fire-research is a command for coding agents from ThierryN/fire-flow. It costs 3 tokens per session (3,027 once invoked), scanned A, original, MIT.

A recurring workflow for searching AI research papers, evaluating their relevance, and applying selected improvements to the dominion-flow project. It can focus the research on a topic or only scan without implementing changes.

In plain words
What is it for?
Use it to research a specified AI topic, score candidate papers, implement chosen improvements, review deferred candidates, or run a quicker single-agent scan.
Why use it?
It gives each session a defined research process instead of requiring the agent to rediscover how to search, score, and apply findings.

Command

Part of the fire-flow plugin — 44 commands, 15 agents shipped together

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 commands/thierryn/fire-flow/fire-research
Clone the repo
git clone --depth 1 https://github.com/ThierryN/fire-flow

Or install fire-flow, the plugin that ships this one along with the rest of its 44 commands, 15 agents.

Wrote this? Show the measurements

A badge for your README with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them.

agentmods badge for fire-research

README.md
[![agentmods](https://agentmods.dev/badge/commands/thierryn/fire-flow/fire-research.svg)](https://agentmods.dev/commands/thierryn/fire-flow/fire-research)
Per session 3 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 3,027 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 $0.00003 $0.03027
Opus 5 $0.00002 $0.01514
Sonnet 5 $0.00001 $0.00605
Haiku 4.5 $0.00000 $0.00303

Measured 3d ago against content hash fe6dbdf37b59, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

fire-research 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 3d 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.

commands/fire-research.md · 371 lines

How it starts

The opening of the file, as written. The whole thing — 371 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/fire-research

Daily AI research-to-upgrade pipeline — search papers, score findings, implement in breaths


Purpose

Systematic daily workflow for finding, evaluating, and implementing AI research improvements into dominion-flow. Eliminates the need to explain the process each session — any fresh agent follows this pipeline autonomously.

Proven track record: 7 successful executions (v3.2 through v8.0), producing 70+ implemented enhancements from 200+ papers analyzed.


Arguments

arguments:
  focus:
    required: false
    type: string
    description: "Research focus area (e.g., 'memory consolidation', 'failure taxonomy')"
    example: "/fire-research agent confidence calibration"

optional_flags:
  --scan-only: "Research and score papers but don't implement (reconnaissance mode)"
  --from-deferred: "Start from deferred candidates of last version"
  --quick: "Single agent search instead of parallel (faster, less thorough)"

Process

Step 1: Load the Research Pipeline Skill

Read the full methodology skill — it contains scoring matrices, wave templates, and citation formats refined across 7 versions:

@skills-library/methodology/RESEARCH_BACKED_WORKFLOW_UPGRADE.md

This is MANDATORY. The skill contains the paper scoring matrix, gap taxonomy, wave grouping rules, and citation format. Do not skip.

Step 2: Determine Focus Area

If arguments provided: Use as focus area directly.

If --from-deferred: Check MEMORY.md and recent handoffs for deferred candidates:

cd ~/.claude/memory
npm run search -- "deferred candidate" --type handoff --limit 5

If no arguments: Ask user:

What area should we research today?

Recent deferred candidates from previous versions:
1. [List from memory search]
2. [List from memory search]

Or describe a new focus area:
> [User input]

Step 3: Launch Parallel Research Agents

Launch 3-4 agents in a SINGLE message for true parallelism:

Agent 1: Academic Papers (arXiv, ACL, NeurIPS, ICML, ICLR — 2024-2026)
  - Search: "[focus area] AI agent 2025 2026"
  - Extract per paper: title, venue, date, key finding, measurable result,
    how it maps to our workflow, specific file/step it would modify
  - Return: Top 15 papers ranked by APPLICABILITY

Agent 2: Community Patterns + Industry Tools
  - Search: Manus AI, Replit Agent, Cursor, Devin, SWE-Agent, Claude Code
  - Focus: What do production AI coding tools do that we don't?
  - Return: Top 10 patterns with source links

Agent 3: Internal Gap Analysis (Explore subagent)
  - Read: All workflow files in the dominion-flow plugin directory
  - Classify each gap: MEMORY | REFLECTION | PLANNING | ACTION | SYSTEM
  - Return: Top 10 gaps ranked by impact

Agent 4 (optional): Failure Pattern Mining
  - Search Qdrant: debug_resolution + failure_pattern source types
  - Find recurring failures indicating systemic gaps
  - Return: Top 5 failure patterns

Read the full file on GitHub · 371 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. 3d ago First seen · 371 lines · 3 tokens per session scan A fe6dbdf37b59

Subscribe to this mod's changes

fire-research is a command published in the GitHub repository ThierryN/fire-flow (77 stars, last pushed 21d ago), licensed MIT. It adds 3 tokens to every session and 3,027 once invoked, about $0.0000 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 2026-08-30.