Poordjaevin MCP server

by Icaro0310·io.github.Icaro0310/poordjaevin·v0.1.1·6 stars

No reviews yet

Typed answers with calibrated confidence for agents. Local-first, no API key.

A85/100grade A

Write a review Install

What users say

No reviews yet

Be the first

Safety scan

A85/100

full report

Adoption

Growing

6 stars146 downloads/wk

Reviews

Write one

Nobody has reviewed Poordjaevin yet.

If you have run it, two minutes of your experience saves the next person an afternoon.

Poordjaevin tools (10)

write = sends, deletes, buys or posts

Read from the package source without running it. The installed server may list more.

classify
Pick exactly one of `options` for `text` (optional `context`). Returns {choice, distribution, confidence, low_confidence}.
decide
Answer a batch of typed questions in one call. Each item is {"id": <name>, "type": "judge"|"classify"|"rate", ...that tool's args}. Returns {answers: [{id: <name>, ...that tool's result}]}
djaevin_calibrate
Report: mean confidence on disagreed decisions per tool and a suggested low_confidence threshold. Suggestion only — change POORDJAEVIN_LOW_CONFIDENCE manually if you agree; nothing is applied.
djaevin_usage
Model/quota monitor: decisions and measured ACP turn cost, grouped by model and by tool, read from djaevin_log.db. Observational only — it never changes behavior. A paid model shows up as cost > 0.
gate
Advisory opinion: risk that `action` moves money or destroys data (optional `kind`, e.g. "money"|"data"). Returns {risk: "low"|"medium"|"high", distribution, confidence, low_confidence, advisory: true, details}. Pure opinion — nothing is blocked, approved or enforced.
judge
Yes/no opinion on `statement` (optional `context`). Returns {answer: "yes"|"no", p_yes, p_no, confidence, low_confidence}. Advisory: a second opinion to weigh, never a verdict.
keepalive
Renew the model's keep_alive window with one trivial request.
mark_disagreement
Mark a logged decision (its `decision_id`) as wrong. With `correct_choice`, the pair input->correct is promoted to validated few-shot examples. Only mark when the right answer is obvious — disagreement on ambiguous questions has no value.
rate
Rate `text` on `scale`: a short numeric range ("1-5") or a list of ordered named levels; `criteria` says what is being rated. Returns {level, score, distribution, confidence, low_confidence}.
usage
Return local inference usage counters for this server process.

Public scan report

scanner v0.1.10 · 2026-10-14 · same rubric, same numbers if you re-run it

1 medium

Code scan33 source files scanned20/25
–Live reliabilityno gateway calls yet and no remote to proben/a
–Tool poisoningtools not inspected (local package is not executed); not countedn/a
Auth qualitylocal package, no credentials required12/15
Maintenancelast push 1 days ago15/15
Maintainer identityregistry namespace matches repository owner; GitHub account older than a year8/10

Findings (1)

mediumeval / new Function usedexec.eval

poordjaevin-0.1.1/src/poordjaevin/backends/local_nli.py: …l.to(self._device) self._model.eval() self._entail_idx = self._find…

Overall 85/100. Components that don't apply are left out of the denominator. Any critical finding is an F.Rubric Appeal a finding JSON

How Poordjaevin compares

Against every graded MCP server, from our own scans and uptime checks. Updated after each scan run.

Safety score: 85/100; higher than 70% of 48,255 servers
Tools: 10; the typical server has 7
Tools that change things: 0%; typical: 0%. Fewer write tools means less an agent can break
GitHub stars: 6; more than 85% of 28,688 servers
Downloads a week: 146; more than 70% of 12,191 servers with a package

What its tools let an agent do: 9 read · 1 change

Install Poordjaevin in Claude Code, Cursor or VS Code

claude mcp add poordjaevin -- uvx poordjaevin

Add to Cursor

What the publisher says

From the Poordjaevin repository's README, as published. We do not edit it. Read it on GitHub

Part of the DEVIN ecosystem

Track: Control · Nature: product

For: AI engineers, QA engineers

Interface: CLI / Python library / MCP server / Bridge

Path: QA engineers · step 3/3 — after devin-evals

Where this fits

Job: Control
Product: devin-judge
Packages: poordjaevin
Mode: read-only
Ecosystem: awesome-devin · registry: devin-powerups

poordjaevin

Renamed (Oct 2026): this repository moved from Icaro0310/poordjaevin to Icaro0310/devin-judge. The PyPI package, console script and MCP registry name stay poordjaevin; stars, issues and history are preserved by the redirect.

The poor man's Jev. An open source, local-first "System One" decision layer for LLM apps: typed decisions with provably calibrated confidence. No API key. No waitlist.

Community project. Not affiliated with, endorsed by, or sponsored by Cognition AI. "Devin" is a trademark of Cognition AI.

Linux · Personal Windows · Corporate Windows

Part of the awesome-devin ecosystem: the curated hub for the devin-* tools.

Your model's 0.9 is a vibe. poordjaevin's 0.9 is a measurement.

Shortened. The full README is on GitHub.

Nothing above is checked by us. What we check is on the safety report.

Poordjaevin: common questions

Is Poordjaevin MCP server safe?: Yes, by our scan: it is graded A (85/100). Read the Poordjaevin safety report
How do I install Poordjaevin?: It runs on your machine. Copy the Claude Code, Cursor, VS Code or Claude Desktop config from the install section.
Does Poordjaevin need an API key?: No secret keys are declared. It reads 1 setting from the environment.
Is Poordjaevin maintained?: The last commit was 22 days ago (2026-10-13). The latest release is v0.1.1.
How does Poordjaevin compare with other MCP servers?: safety score 85/100 (higher than 70% of 48,255 servers); tools 10 (the typical server has 7); tools that change things 0% (typical: 0%. Fewer write tools means less an agent can break). Full comparison.