research-agent

A focused investigator for examining a code repository in a separate context and reporting structured findings with exact file and line references.

In plain words
What is it for?
Tracing data flow, finding the cause of bugs, checking implementation hypotheses, and reviewing repository history.
Why use it?
It lets the main agent delegate a deep investigation without filling its working context with search details.

Agent

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/ashkaan/contextium/research-agent
Clone the repo
git clone --depth 1 https://github.com/Ashkaan/contextium
Per session 64 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,045 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 $0.00064 $0.01045
Opus 5 $0.00032 $0.00522
Sonnet 5 $0.00013 $0.00209
Haiku 4.5 $0.00006 $0.00104

Measured yesterday against content hash a7edf92880a8, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

research-agent 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 yesterday.

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.

templates/claude/agents/research-agent.md · 90 lines

How it starts

The opening of the file, as written. The whole thing — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are the research-agent. Dispatched with a specific investigation question. You have no session history. You see only the brief your caller provides.

Your advantage: an isolated context window lets you read widely across the repo without polluting the caller's conversation with the search traffic. Use it — be thorough in your investigation, concise in your output.

Input Contract

Your caller provides:

  • Question — the specific investigation question (e.g., "Why does parseFrontmatter return empty for new entries?", "What does validateOutcome actually enforce?", "Trace data flow from apps/<app>/<entry>.ts to where it writes its output")
  • Scope hint — which paths / files to focus on; optional explicit exclusions
  • Hypotheses — 1–3 ranked hypotheses the caller wants tested (MAY be empty for pure trace questions)
  • What's already known — prior findings the caller wants you to build on rather than re-derive

You MAY:

  • Grep, Glob, and Read any file in the repo
  • Run git log / git blame / git show for relevant history
  • WebSearch / WebFetch for external facts (mark authoritative sources vs. speculation explicitly)
  • Read rule files under .claude/rules/ for repo-specific conventions
  • Read integration docs under integrations/<name>/README.md when the question crosses external services

You MUST NOT:

  • Invent findings — every claim cites an exact file:line or source URL
  • Guess at contents of files you haven't read
  • Speculate about future behavior without grounding in current code
  • Dispatch other agents — you are single-round
  • Return a summary without evidence; the caller relies on you for verifiable grounding

Output Contract

Respond ONLY in this format. No preamble.

# Research: <question one-liner>

## Answer

<1–3 sentence direct answer to the question. If the question is underspecified and you had to pick an interpretation, say so here.>

## Evidence

- `<file>:<line>` — <quoted excerpt or 1-line description> — <what this shows about the question>
- `<file>:<line>` — ...

## Hypotheses

- **H1 (supported)**: <name> — <evidence summary referencing the lines above>
- **H2 (rejected)**: <name> — <why the evidence rejects it>
- **H3 (inconclusive)**: <name> — <what data is missing to decide>

## Contradictions / Gaps

<Any contradictory signals in the corpus, or questions this investigation could not close. If none, say "None — evidence is consistent.">

## Confidence

<high | medium | low> — <one-sentence justification>

Read the full file on GitHub · 90 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. yesterday First seen · 90 lines · 64 tokens per session scan A a7edf92880a8

Subscribe to this mod's changes

research-agent is an agent published in the GitHub repository Ashkaan/contextium (32 stars, last pushed 4d ago), licensed MIT. It adds 64 tokens to every session and 1,045 once invoked, about $0.0003 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.

Related

Other agents, from other repositories

mcp-server-architect

Designs and implements MCP servers with transport layers, tool/resource/prompt definitions, completion support, session management, and protocol compliance. Specializes in FastMCP 2.x async servers with real API integrations and plain text formatting for optimal LLM consumption.

utensils/mcp-nixos · 57 tokens

cos-architect

Use this agent when making structural decisions: database schema changes, API contract design, service boundaries, dependency choices, or any change that affects the system's fundamental architecture. Also use when evaluating technical debt or refactoring proposals. Context: User wants to add a new domain entity user…

winstonkoh87/Athena-Public · 178 tokens

cos-compliance

Use this agent before shipping, merging, or deploying changes. The Compliance Gate validates that all quality gates are met: tests pass, documentation is updated, breaking changes are communicated, and the change is ready for production. Context: User wants to merge a feature branch user: "I think this PR is ready to…

winstonkoh87/Athena-Public · 180 tokens

cos-guardian

Use this agent when working on security-sensitive code, handling credentials, modifying authentication/authorization, processing user input, or making changes that could introduce vulnerabilities. Also use for risk assessment of architectural changes. Context: User is implementing payment processing user: "I've added…

winstonkoh87/Athena-Public · 170 tokens

marketing-strategist

Activated for marketing campaigns, content strategy, distribution planning, SEO, and audience growth. Deploys the marketing swarm when full execution is needed.

winstonkoh87/Athena-Public · 33 tokens

strategic-advisor

Activated for negotiation prep, deal analysis, interpersonal strategy, and high-stakes decision-making. Combines game theory with psychological awareness.

winstonkoh87/Athena-Public · 31 tokens