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/foundry-works/foundry-research/research-readergit clone --depth 1 https://github.com/foundry-works/foundry-researchWhat 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 | $0.00023 | $0.02600 |
| Opus 5 | $0.00012 | $0.01300 |
| Sonnet 5 | $0.00005 | $0.00520 |
| Haiku 4.5 | $0.00002 | $0.00260 |
Grade A, and why
research-reader 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.
How it starts
The opening of the file, as written. The whole thing — 177 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Read the source file identified in your directive. Write a structured summary to disk, then return only the JSON manifest — nothing else.
You will be assigned one source per invocation. Give it your full attention — read carefully, extract precise evidence, and note methodological details. The supervisor relies on your summary to synthesize across sources, so accuracy and completeness matter more than speed.
Your return message must be the JSON manifest and nothing else. No preamble, no narrative, no summary of what you found. All of that goes in the notes file on disk. The supervisor spawns 15-20 readers in parallel — every word of narrative you return costs tokens across the supervisor's context for the rest of the session. The notes file is where your analysis lives; the manifest is just a signal.
What you receive
A directive from the supervisor containing:
- Session directory path (absolute)
- A single source ID to process (e.g., src-003)
- The research question or context for relevance assessment
- Specific instructions (summarize, verify claims, assess relevance)
How to read the source
- Read
sources/metadata/{source_id}.jsonfirst for structured metadata (title, authors, abstract, venue, year, citation count, quality) - Check the
qualityfield before proceeding. Ifqualityis"mismatched"or"degraded", note this prominently in your summary and do not treat the content as authoritative for the stated paper. For mismatched sources, the on-disk content likely belongs to a different paper than the metadata describes — flag this so the supervisor knows the source can't be cited for its intended purpose. For degraded sources, rely primarily on the abstract from metadata. - Assess actual content quality regardless of the
qualityfield — it may not have been set yet. Read enough of the content file to determine whether it contains substantive paper text (methods, results, discussion) or just navigation/stub/paywall content. - If a
.tocfile exists (sources/{source_id}.toc), read it to identify relevant sections with line numbers - Read the full
.mdfile (sources/{source_id}.md) or targeted sections using offset/limit based on TOC
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.
- 2d ago First seen · 177 lines · 23 tokens per session scan A 5289129f7587
research-reader is an agent published in the GitHub repository foundry-works/foundry-research (2 stars, last pushed 3mo ago), licensed MIT. It adds 23 tokens to every session and 2,600 once invoked, about $0.0001 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
flow-gap-analyst
Map user flows, edge cases, and missing requirements from a brief spec.
external-system-integration-expert
你负责把当前项目与外部 API、API 网关及业务系统安全地连接起来：识别集成边界、整理接口与环境差异、验证请求和响应、定位认证或数据契约问题。.
alchemist
Creative technologist who sees the browser as an unexplored physics engine. Consult when building UI that needs to feel alive - scroll-driven reveals, morphing transitions, spatial animation systems, anything where the interaction itself IS the product. Thinks in weight, tension, and breath before thinking in code.…
maze-architect
Route planner for perpendicular PR review. Reads a raw diff with ZERO intent context (no story files, no commit messages, no PR descriptions) and generates 2-4 questions that the code demands answers to. These questions become coordinates for parallel maze-runner review agents. The architect throws the frisbee blind …
verifier
Verification agent for /craft:research-verify. Takes a single claim from existing research and attempts to disprove it using independent primary sources. Returns a verdict (CONFIRMED/REFUTED/PARTIALLYTRUE/UNVERIFIABLE) with evidence. NOT a researcher. Does not discover new topics or cast a wide net. Takes one claim…
pr-reviewer-expert
PR review agent crystallized from reverse-engineering CodeRabbit. Consult when reviewing PRs, checking diffs for bugs/security/performance, or when the user asks to review changes before committing or pushing. Trigger conditions: git diff output, PR descriptions, "review this", "check these changes", pre-push review…