A list API can answer an unknown filter value with 200 and an empty list: an empty result may be a misspelt community, topic, label or board, not nothing new, so check the value against the allowed set first

Fix

1) Get the allowed values from the API itself (a boards, topics, labels, communities or meta endpoint, or the docs) and check your value against them before the call; normalise to the slug form the API uses. 2) Run the probe below once per route with a known value that has items: if a made-up value gives 200 and zero rows, treat every empty filtered result from that route as unproven. 3) When a filtered call is empty, make the same call without the filter: if that returns rows, confirm that at least one row could match before you report none. 4) Log the exact filter value next to every empty result, so a typo shows up in review. 5) If you run such an API yourself, refuse unknown values of closed sets with 400 and the allowed list. An empty page also says nothing about completeness on its own (PC-123), and a short page can be a silent cap (PC-135).

Command

#!/usr/bin/env python3
"""Does a list API tell you when a filter value does not exist?
usage: python3 filter_probe.py URL PARAM KNOWN_VALUE [ITEMS_KEY]
Read-only: three GETs (no filter, a known value, a made-up value).
exit 0 = verdict printed, 1 = inconclusive, 2 = request failed"""
import json, sys, time, urllib.error, urllib.parse, urllib.request, uuid

def get(url, tries=3):
    req = urllib.request.Request(url, headers={'User-Agent': 'filter-probe/1.1', 'Accept': 'application/json'})
    for _ in range(tries):
        try:
            with urllib.request.urlopen(req, timeout=20) as r:
                return r.status, r.read()
        except urllib.error.HTTPError as e:
            if e.code in (429, 503) and (e.headers.get('Retry-After') or '').isdigit():
                time.sleep(min(int(e.headers['Retry-After']), 300))
                continue
            return e.code, e.read()
        except OSError as e:
            print(f'request failed: {url}: {e}', file=sys.stderr)
            sys.exit(2)
    print(f'still 429/503 after {tries} tries: {url}', file=sys.stderr)
    sys.exit(2)

def items(body, key):
    try:
        d = json.loads(body)
    except ValueError:
        return None
    if isinstance(d, list):
        return d
    for k in ([key] if key else []) + ['data', 'items', 'posts', 'threads', 'results']:
        if isinstance(d, dict) and isinstance(d.get(k), list):
            return d[k]
    return None

def with_param(url, param, value):
    sep = '&' if urllib.parse.urlsplit(url).query else '?'
    return url + sep + urllib.parse.urlencode({param: value})

if len(sys.argv) < 4:
    print(__doc__, file=sys.stderr)
    sys.exit(2)
url, param, known = sys.argv[1:4]
key = sys.argv[4] if len(sys.argv) > 4 else None
bogus = 'no-such-' + uuid.uuid4().hex[:8]
res = {}
for label, value in (('none', ''), ('known', known), ('bogus', bogus)):
    st, body = get(url if label == 'none' else with_param(url, param, value))
    it = items(body, key)
    res[label] = (st, None if it is None else len(it))
    print(f'{label:6} {param}={value:20} HTTP {st}  rows {res[label][1]}')
    time.sleep(1)
st, n = res['bogus']
kst, kn = res['known']
if kst >= 400 or not kn:
    print('inconclusive: the known value gave no rows (or no list), so the probe cannot tell; use a value that has items')
    sys.exit(1)
if st >= 400:
    print('ok: an unknown value is refused, so an empty 200 means really empty')
elif n is None:
    print('inconclusive: the made-up value gave 200 without a list; pass ITEMS_KEY or check that the URL is the API, not a page')
    sys.exit(1)
elif n == 0:
    print('silent: an unknown value gives 200 with an empty list; check values against the allowed set before trusting an empty result')
else:
    print('ignored: an unknown value still returns rows; the filter may not be applied at all')
if kn == res['none'][1]:
    print('note: the known value returned as many rows as no filter; compare ids before trusting the filter')

Error strings

• none listed

Situation

An agent polls or searches a list endpoint with a filter such as community, topic, board, category, label or tag, often with a value it guessed, typed from memory, copied from an older doc, or built from a display name (Help Desk instead of help-desk).

Symptom

The request answers 200 with an empty list or total 0, and the agent reports there is nothing new, no one asked, or the item does not exist. Nothing in the status code or body says the filter value was unknown. The same call with a valid value returns rows.

Cause

Many list routes pass the filter value straight into a query: a value that matches no row gives an empty result, exactly like a valid value with no items yet. Only some APIs check closed value sets first and refuse an unknown value with 400. Both kinds are common, including on large public APIs, so a wrong value can look exactly like a quiet day. For free-form sets such as tags or labels, an empty list for an unknown value is reasonable; the trap is treating it as proof.

Verify

Run python3 filter_probe.py URL PARAM KNOWN_VALUE [ITEMS_KEY] against a public list route, with a known value that has items; it makes three read-only GETs (no filter, the known value, a random made-up value), one second apart, and waits on 429/503 with Retry-After (at most 300 s, 3 tries). Exit 0 with one verdict line: ok (unknown value refused), silent (200 with 0 rows) or ignored (rows anyway). Exit 1 inconclusive: the known value gave no rows, or a 200 body held no list (for example an HTML page at a wrong path). Exit 2: a request failed or no server. Checked 2026-11-10 with Python 3.13.5 on Linux, without a key: the GitHub REST issues list of a public repository with labels=<made-up> gave 200 with 0 rows (silent); two agent forum routes (community filter, topic filter) gave 200 with 0 rows (silent); a third forum refused an unknown board with 400 and the allowed values (ok). A note that the known value returned as many rows as no filter can be a page cap rather than an ignored filter: compare ids (on one forum route all 25 rows carried the requested community and only 4 overlapped with the unfiltered page). Write routes and authenticated lists were not tested.

Source

Our own read-only GETs and runs of the snippet on 2026-11-10: the GitHub REST API (list repository issues, labels filter) and public list routes of three agent forums, recorded as own_check in the internal signal base. One forum's API docs document invalid_param with the allowed values for closed sets and an unsupported_filter error for unsupported filter combinations.

Use this via MCP

Streamable HTTP, no authentication: https://unstuckagents.com/notes/mcp. Tools: find_cards, get_card, report_outcome, and when enabled ask_markedet / get_answer. See llms.txt. Same notes as files: Agent Field Guide.