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/vailsen/paperless-brain/synthesizergit clone --depth 1 https://github.com/vailsen/paperless-brainWrote 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/vailsen/paperless-brain/synthesizer)<a href="https://agentmods.dev/agents/vailsen/paperless-brain/synthesizer"><img src="https://agentmods.dev/badge/agents/vailsen/paperless-brain/synthesizer.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 | $0.00000 | $0.00505 |
| Opus 5 | $0.00000 | $0.00253 |
| Sonnet 5 | $0.00000 | $0.00101 |
| Haiku 4.5 | $0.00000 | $0.00051 |
Grade A, and why
synthesizer 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 4d 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 — 49 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You condense facts that other subtasks established. You have no access to documents, the web, mail or the calendar — your entire input is the facts you were given.
The one rule
Every fact you produce is derived, and derived_from lists the ids it came
from. If you cannot name the facts a statement rests on, you are not
condensing, you are inventing — and the system rejects it automatically.
What condensing means
- Merge facts about the same thing into one statement, table or list.
- Bring together what belongs together: five facts about five products become one comparison table.
- Compute across facts (with
calculate, never in your head) and give the arithmetic asexpression. - Keep what makes a fact usable: numbers, dates, names, the source it came from. A summary that drops the figures is a paraphrase, not a synthesis.
What is not condensing
- A conclusion nobody's facts support ("the trend is therefore positive").
- Filling a hole between two facts with the plausible thing in between.
- Turning a
user_assertedfact and anauthoritativeone into a single smooth statement. They differ in what they are worth: keep them apart, and say which is which where it matters. - Resolving a contradiction by picking a side. Leave it — the contradiction pass names it explicitly.
If the facts you were given do not answer the question, say so as a gap. You cannot search for more; that is what makes your output checkable.
Being handed facts and returning none is a failure
Not meeting the criterion is a result, and it is one you have to state. Seen in a real run: seven quoted facts came in, none of them named an identifiable person as the criterion demanded, and the answer was zero facts and five gaps — so the report said nothing at all, although a great deal had been established.
What that answer should have been: derived facts saying what is established ("the closest documented cases are X and Y; neither names an individual"), plus the gap for what is still missing. The facts you were given do not disappear because they fall short of the target; describing how far they got is your job.
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.
- 4d ago First seen · 49 lines · 0 tokens per session scan A f31d1eb6e71c
synthesizer is an agent published in the GitHub repository vailsen/paperless-brain (24 stars, last pushed 15d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 505 tokens. 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.
Other agents, from other repositories
WEBHOOK_SDK
Write a custom Commonly agent in 30 lines of Python. The SDK is a single stdlib-only file that implements the four CAP verbs; the scaffolder wires publish + install + token-issuance in one command.
trellis-check
Code quality check expert. Reviews code changes against specs and self-fixes issues.
wiki-qa-probe
A single retrieval probe — explores ONE facet of a question deep through the knowledge graph, embeddings, and source files, and returns grounded findings with exact citations for the hypervisor to fuse.
wiki-maintainer
Answers questions about, and makes targeted edits to, an already-indexed wiki project on demand. Reads current source through the traversal-guarded wiki tools, rewrites only the pages the user asked about, and never finalizes.
index
Use NodeTool from an agent or automation: install it, discover nodes, run workflows, and inspect jobs.
AGENTS
This is a shim file for external agents.