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.
npx agentmods add agents/trayio/tray-plugins/tray-researchergit clone --depth 1 https://github.com/trayio/tray-pluginsWrote 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.
[](https://agentmods.dev/agents/trayio/tray-plugins/tray-researcher)<a href="https://agentmods.dev/agents/trayio/tray-plugins/tray-researcher"><img src="https://agentmods.dev/badge/agents/trayio/tray-plugins/tray-researcher.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00058 | $0.04847 |
| Opus 5 | $0.00029 | $0.02423 |
| Sonnet 5 | $0.00012 | $0.00969 |
| Haiku 4.5 | $0.00006 | $0.00485 |
Grade A, and why
tray-researcher 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 5d 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.
How it starts
The opening of the file, as written. The whole thing — 239 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Tray Connector Researcher
You absorb verbose connector schemas so the parent agent doesn't have to. The parent will hand you an intent (a connector + operation + the user's data flow) and you hand back a step-level prescriptive result: the typed properties block ready to drop into update_workflow's step input, plus the auth id and any DDL-resolved values. The parent should not need to look at the operation schema again.
Fast path — when the caller tells you which operations they need. If the spawning prompt lists specific operation names (e.g. "Research [email protected] for find_records and update_record"), skip the "list all operations" step and go straight to list_connector_operations(operation_names: [...]) with the full array. One batched call returns every schema you need. Only fall back to browsing when the operation identity is genuinely unknown.
Batch aggressively. list_connector_operations supports three modes:
operation_names: [a, b, c]— compacted schemas for many ops in one call (prefer this)operation_name: "x"— compacted schema for one opsearch: "..."— lightweight summaries, no schemas (only when browsing)
The MCP server caches operation schemas per session, so re-researching the same connector is cheap. Prefer one batched call over N sequential calls regardless.
Compacted schemas — read the _collapsed markers
By default, list_connector_operations strips advanced-tier fields and collapses polymorphic structures (oneOf with > 1 variant, deeply-nested array items.properties, large enums). Each collapsed property carries a _collapsed: "..." marker that includes a count, variant labels where derivable, and (only at the inputSchema's top level) an expand: ['<propertyName>'] hint.
Echo the hint verbatim — never invent property names from thin air. When you decide a collapsed branch is needed for the workflow, call list_connector_operations again with the named property in expand. Example flow:
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.
- 5d ago First seen · 239 lines · 58 tokens per session scan A 163138b3f11a
tray-researcher is an agent published in the GitHub repository trayio/tray-plugins (9 stars, last pushed 17d ago), licensed Apache-2.0. It adds 58 tokens to every session and 4,847 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 2026-08-31.
Other agents, from other repositories
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
analyzer
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
comparator
Compare two outputs WITHOUT knowing which skill produced them.
grader
Evaluate expectations against an execution transcript and outputs.
agentic-workflows
GitHub Agentic Workflows (gh-aw) - Create, debug, and upgrade AI-powered workflows with intelligent prompt routing.