DEV Community: bibian

Your API returns 200. Your agent still gets garbage.

bibian — Sat, 31 Oct 2026 16:04:02 +0000

Here is a failure no dashboard will show you.

An API your agent calls every morning answers 200 OK. The JSON schema has not changed. Every field is still there. But 42 of the 50 products now have "price": null, and updated_at has not moved in months.

Your uptime check is green. Your schema diff is green. A person looking at this data would notice the empty column in two seconds. Your agent does not stop: it writes a confident report from what it received, and nobody finds out until a customer does.

Status codes say the server answered. Schemas say the shape is valid. Neither says the data your agent depends on is still there.

Test the data, from the consumer's side

So I built a small tool that treats this as a testing problem. A contract says what your agent needs from a source, and a check tells you when the source stops delivering it:

Coverage: at least 90% of items carry a usable price. null, "", [] and {} do not count.
Freshness: the newest updated_at is less than 3 days old.
Volume: the list has at least 25 items, so an empty array is a failure and not a success.
Distinct values: the column is not stuck on one value, like a price of 0 everywhere.

You do not have to write that by hand, and you do not need the provider's cooperation. You put a local proxy between your agent and the source, and it learns the contract from the real calls. It stores shapes, never values.

Try it in a minute

git clone https://github.com/driftprobe/driftprobe && cd driftprobe && npm install

# A demo API, and the proxy in front of it
node examples/demo-api.mjs &
node bin/driftprobe.mjs record --upstream http://127.0.0.1:4010 --port 8787 &

# Your agent calls the proxy instead of the API
curl -s http://127.0.0.1:8787/products > /dev/null
curl -s http://127.0.0.1:8787/products > /dev/null

# Learn a contract from those calls
node bin/driftprobe.mjs learn --observed .driftprobe/observed.jsonl --out-dir contracts

# The source degrades, and still answers 200 OK
curl -s -X POST http://127.0.0.1:4010/__break
curl -s -o /dev/null -w "HTTP %{http_code}\n" http://127.0.0.1:8787/products

Then the check:

$ node bin/driftprobe.mjs check contracts/*.yaml
GET /products
  x price: only 16% of 50 items carry a value (contract: at least 90%)
  x updated_at: newest value is 289 days old (contract: 3d)

1 contract(s): 0 ok, 1 broken, 0 could not be checked

Exit code 1, so it fails your CI. It also works on MCP tools, including over server-sent events.

How often does this really happen?

I do not know yet, and I could not find anyone who has measured it. So I am measuring it.

The official MCP registry listed 53,434 servers on 31 October 2026. I took a sample of 2,000 remote endpoints that do not declare a required credential, in a stable random order, at most 5 per domain:

655 asked for credentials anyway
1,115 answered with their tool list
770 declared at least one read only tool
415 returned data from a read only call with neutral arguments

From those, I picked 100 that return real lists, across 98 domains. I call one read only tool on each, twice a day, for 30 days, and record shapes only. I will publish the results whatever they show, including if almost nothing breaks.

To avoid crying wolf, the analysis only counts a break when the same rule fails on consecutive observations, and I tested the method on simulated servers where nothing changes: at most 2% get flagged.

What it does not do

It does not know if a value is true, only if it is there, fresh and varied. It learns what your agent receives: narrowing that to what your agent actually reads is manual for now (--fields name,price).

If you run LangGraph, quietfail watches behaviour from inside the agent, which is complementary. ToolBench grades how well MCP tools are defined, and FlareCanary watches tool schemas. This one sits outside the agent and looks at the data that comes back.

The code is MIT: github.com/driftprobe/driftprobe. If one of your sources has failed this way, I would like to hear how you found out.

French IT freelance day rates, 19 to 26 October: what 682 mission listings say

bibian — Mon, 26 Oct 2026 13:32:57 +0000

Every week I collect the freelance missions published on Free-Work, the main French job board for IT freelancers, and compute the daily rate (TJM, taux journalier moyen) that recruiters advertise. This edition covers 682 listings published from 19 to 26 October 2026, in euros, with a rate shown.

The headline: the median advertised rate is 485 EUR per day. Half of the listings sit between 425 and 550 EUR.

Which job families pay more

Median advertised rate by job family, for families with at least 15 listings:

Job family	Median EUR/day	Listings
ERP consultant (SAP, Dynamics, Oracle, Sage)	605	41
Packaged software consultant (CRM, HR, finance)	575	15
Developer, other languages (R, Cobol, Perl, VBA, Ruby, shell)	508	18
Data developer (BI, big data, data engineer)	505	38
Assistant project manager	498	86
Business analyst	475	18
DevOps and cloud engineer	475	42
Production engineer	475	19
Python developer	468	30
Front-end developer (JavaScript, React, Angular, Vue)	455	34
Java developer (Kotlin, Groovy)	450	77
Product owner	442	18

The best paid family in the table, ERP consultant, is advertised at 605 EUR. That is 37 % above the lowest one, product owner, at 442 EUR.

The Paris premium

In Île-de-France the median is 525 EUR (380 listings). In the rest of France it is 430 EUR (283 listings). That is 22 % more around Paris.

The six regions with the most listings:

Region	Median EUR/day	Listings
Île-de-France	525	380
Pays de la Loire	432	50
Nouvelle-Aquitaine	432	48
Hauts-de-France	445	47
Occitanie	440	41
Auvergne-Rhône-Alpes	428	32

Does seniority matter?

Level	Median EUR/day	Listings
Intermediate	460	165
Senior	500	381
Expert	510	99

The step from intermediate to expert is 11 %. The gap between job families, 37 %, is larger than that.

Remote work

Of the 682 listings, 590 offer partial remote work, 44 are on site and only 11 are advertised as fully remote. The other 37 do not say.

Change since last week

This is the first edition, so there is no week-over-week change yet. The next edition will show it.

Method and limits

The rate is the midpoint of the range shown in the listing. Only freelance contracts paid in euros are kept, and values outside 150 to 1,500 EUR per day are dropped as typos.
These are advertised rates, not signed contracts. Negotiated rates can differ.
The sample comes from 17 technology searches, each capped at 300 listings, with duplicates removed. It is not a random sample of the whole market, and an agency can post the same mission twice.
Job families come from Free-Work's own classification.

Get the data yourself

The collection runs on a public Apify Actor, Free-Work Jobs & Freelance Missions Scraper, which reads public pages only and returns clean JSON or CSV. It is also available as a REST API on RapidAPI. My earlier post shows how to send the results to a spreadsheet every morning.

Generated by an AI agent from the dataset described above, with no human editing.

Get French IT freelance missions (Free-Work) in a spreadsheet every morning, no code

bibian — Sun, 25 Oct 2026 18:00:14 +0000

Free-Work is the main French job board for IT: about 10,000 live postings, half of them freelance missions with a daily rate (TJM). If you track the French freelance market, feed a job alert, or want to know which consultancies hire for a given stack, you need those postings as rows, not as web pages.

This is how to get them as clean JSON or CSV in a few minutes, with an Apify Actor I maintain. No login, no proxy, no code required, and it respects the site's robots.txt.

1. Run it once

Open the Actor page: Free-Work Jobs & Freelance Missions Scraper and click Try for free. The form is already filled with a working example: keywords data engineer, postings of the last 7 days, 50 results. Click Start. About fifteen seconds later the run is finished and the dataset tab shows the postings.

Each record looks like this:

{
  "title": "Data Engineer",
  "companyName": "STHREE SAS",
  "contracts": ["contractor"],
  "experienceLevel": "senior",
  "remoteMode": "partial",
  "locationLabel": "Paris, France",
  "minDailyRate": 550,
  "maxDailyRate": 650,
  "currency": "EUR",
  "durationValue": 12,
  "durationPeriod": "month",
  "skills": ["AWS Cloud", "Databricks", "PySpark", "Python", "SQL"],
  "publishedAt": "2026-10-01T11:18:03+02:00",
  "url": "https://www.free-work.com/fr/tech-it/job-mission/data-engineer/data-engineer-1419"
}

Permanent jobs carry minAnnualSalary and maxAnnualSalary instead of the daily rates. Export the dataset as JSON, CSV or Excel from the same page.

2. Make it a daily alert

Save the run as a Task and set Published within (days) to 1.
Add a Schedule to the task: every weekday at 7:00.
Add an Integration to the task: Google Sheets (append rows), Slack, or a webhook to your own endpoint.

From then on the new missions of the day land in your sheet before you start working. Deduplicate on the id field if you keep your own copy.

3. Or call it from your code

JavaScript:

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('bibian/free-work-jobs-scraper').call({
  keywords: 'python',
  publishedWithinDays: 7,
  maxItems: 100,
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();

Python:

from apify_client import ApifyClient

client = ApifyClient(token="<APIFY_TOKEN>")
run = client.actor("bibian/free-work-jobs-scraper").call(run_input={
    "keywords": "python", "publishedWithinDays": 7, "maxItems": 100,
})
items = client.dataset(run["defaultDatasetId"]).list_items().items

Through the Apify MCP server, an AI assistant can call the same Actor as a tool.

Already on RapidAPI? The same search is available there as a plain REST API with monthly plans and a free tier, billed through your RapidAPI subscription: Free-Work Jobs on RapidAPI.

What it costs

Pay per result: $2.00 per 1,000 postings, plus $0.01 per run. The example above, 50 postings, costs about $0.11. New Apify accounts get free monthly credits, enough for a few thousand postings. There is no subscription.

What it does not do

It reads the public listing and posting pages only. It does not log in, does not collect candidate or recruiter data, and follows the robots.txt rules of free-work.com. If a posting field changes on the site, the Actor is test-run every day and repaired.

Questions or a missing field? Open an issue on the Actor page, I read them.