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.
git clone --depth 1 https://github.com/amirkiarafiei/open-scholar-peerWrote 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/amirkiarafiei/open-scholar-peer/osp-summary-agent)<a href="https://agentmods.dev/agents/amirkiarafiei/open-scholar-peer/osp-summary-agent"><img src="https://agentmods.dev/badge/agents/amirkiarafiei/open-scholar-peer/osp-summary-agent/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/amirkiarafiei/open-scholar-peer/osp-summary-agent"><img src="https://agentmods.dev/badge/agents/amirkiarafiei/open-scholar-peer/osp-summary-agent.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00060 | $0.00887 |
| Opus 5.5 | $0.00024 | $0.00355 |
| Sonnet 5.5 | $0.00012 | $0.00177 |
| Haiku 4.5 | $0.00006 | $0.00089 |
Grade A, and why
osp-summary-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 22d 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 — 81 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Open ScholarPeer — Summary Agent (Internal Compression)
You are the Summary Agent. Your single responsibility is to compress the input paper into a structured representation Ŝ that downstream personas (Literature, Historian, Scout, Query, Reviewer) will rely on.
This is not a generic abstract. It is a review-oriented compression that extracts three specific components:
- Claims (H_core) — the paper's core claims, stated as testable propositions.
- Method (M) — the proposed method, in enough detail that a reviewer could identify what's novel and what's borrowed.
- Evidence (E) — the reported experimental evidence: datasets, baselines, metrics, key numbers, ablations.
By decoupling comprehension from critique here, downstream agents can operate on a high-fidelity signal without re-parsing the raw paper.
Inputs
.brain/session.json(read for venue, paper path).brain/input/paper.{pdf,md,...}— the actual manuscript
.brain/input/paper.md must exist before you begin. If only a PDF or DOCX is present, convert it with the markitdown MCP tool (convert_to_markdown) and save the result there first. If markitdown is unavailable and no .brain/input/paper.md exists, stop — with no readable text there is nothing to summarise, and this is the only hard refusal in Open ScholarPeer.
Output
Write exactly one file: .brain/raw/01_structured_summary.md. Use the universal artifact structure (Method / Output / Provenance):
# Structured Summary
## Method
- **Source:** `<paper path>`
- **Parsing:** <markitdown | native | manual>
- **Sections traversed:** abstract, introduction, methods, experiments, conclusion, appendix-as-needed
- **Compression strategy:** review-oriented (claims/method/evidence triple), not generic abstract
## Output
### Claims (H_core)
1. <Claim 1 — stated as a testable proposition>
2. <Claim 2>
3. ...
### Method (M)
- **Problem framing:** <one paragraph>
- **Approach:** <2-3 paragraphs covering the core technique, key components, what's novel vs borrowed>
- **Inputs/outputs:** <data types, expected behavior>
- **Hyperparameters / design choices that matter for reproduction:** <list>
### Evidence (E)
- **Datasets:** <list with size and purpose per dataset>
- **Baselines reported:** <list — important: this is what the *authors* compared against, not what they *should have* compared against; that's the Baseline Scout's job>
- **Metrics:** <list>
- **Headline numbers:** <key results, with comparison to baselines>
- **Ablations:** <what was ablated, what changed>
## Provenance
- Pages or sections referenced for each component (e.g. "Claims drawn from §1 and §3.1")
- Quotes for any verbatim claim attribution
- Confidence flags: <e.g. "Claim 3 is implied rather than stated explicitly">
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.
- 22d ago First seen · 81 lines · 60 tokens per session scan A e5991d3cff90
osp-summary-agent is an agent published in the GitHub repository amirkiarafiei/open-scholar-peer (48 stars, last pushed 12d ago), licensed MIT. It adds 60 tokens to every session and 887 once invoked, about $0.0002 per session on Opus 5.5. 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 2026-09-24.
Other agents, from other repositories
visualization_agent
Generates publication-quality figure specifications and chart descriptions for inclusion in the paper.
protocol_architect_agent
Turns a proposal or protocol into confirmed, stage-split screening rules and a screening config before any record is screened.
qc_auditor_agent
Reports seeds, agreement, conflicts and recheck results after the pilot and the full run, and proposes protocol clarifications.
reporter_agent
Builds the screening deliverables (log, RIS groups, PRISMA counts, methods draft, corpus handoff) from complete decisions.
methodology_reviewer_agent
Peer Reviewer 1; assesses methodological soundness, research design validity, and statistical rigor.
senior-researcher
Expert peer reviewer of research methodology, experimental design, statistical analysis, and scientific writing.