feat(okf): trust_tier derives three tiers from the actor prefix

Co-Authored-By: Claude <claude-opus-5>
This commit is contained in:
Kjell Tore Guttormsen 2026-09-02 20:17:04 +02:00
commit a00cde6444
2 changed files with 125 additions and 0 deletions

View file

@ -371,6 +371,53 @@ class SkippedLink:
reason: SkipReason reason: SkipReason
#: A concept's trust level, derived from its ``verified`` actors (SPEC §5.3), lowest to highest.
#: Derived, never stored: OKF records objective signals and refuses to persist a subjective score,
#: so this is a reading of the actors and not a field any document carries.
TrustTier = Literal["unverified", "machine-confirmed", "human-reviewed"]
#: The ONE actor prefix that raises a concept to the human tier (SPEC §7 actor convention).
_HUMAN_ACTOR_PREFIX = "human:"
def trust_tier(entries: tuple[dict[str, str], ...] | None) -> TrustTier:
"""Derive the trust tier from ``verified`` entries (SPEC §5.3).
``None`` or empty ``unverified``; any actor whose ``by`` STARTS WITH ``human:``
``human-reviewed``; otherwise ``machine-confirmed``.
**Prefix, never substring.** ``by: bot/human:2`` is a machine actor whose identifier merely
contains the token, and a substring test would promote it minting a human sign-off nobody
gave, which is the fabricated-provenance defect one level down.
**An entry that names NO actor is REFUSED, never tiered.** "Otherwise ⇒ machine-confirmed"
would derive a trust level from an entry that identifies nobody. ``decode_flow_value`` already
refuses that shape when it reads a ``verified`` value; this function is public, so it ASSERTS
the invariant at its own door instead of assuming its caller came through that one. No new
exception type: the named refusal for this condition belongs to the decoder, and a second class
here would imply a second rule.
**Entry count is not a tier.** SPEC §5.3 derives the tier from the actor prefix alone, so a
two-entry machine list stays ``machine-confirmed``.
Gated by ``tests/test_provenance_decoder_loadbearing.py``."""
if not entries:
return "unverified"
actors: list[str] = []
for entry in entries:
actor = entry.get("by", "").strip()
if not actor:
raise ValueError(
f"a verification entry {entry!r} names no `by` actor — SPEC §5.2 makes it "
"required, and deriving a trust tier from an entry that identifies nobody would "
"mint the provenance it claims to read"
)
actors.append(actor)
if any(actor.startswith(_HUMAN_ACTOR_PREFIX) for actor in actors):
return "human-reviewed"
return "machine-confirmed"
#: WHY a provenance value could not be read. The tokens name the SHAPE the value is written in #: WHY a provenance value could not be read. The tokens name the SHAPE the value is written in
#: and NOTHING else — the same discipline ``SkipReason`` carries. A block sequence and a block #: and NOTHING else — the same discipline ``SkipReason`` carries. A block sequence and a block
#: mapping are both CONFORMANT OKF (SPEC §5.2 writes ``verified`` in exactly those forms); they are #: mapping are both CONFORMANT OKF (SPEC §5.2 writes ``verified`` in exactly those forms); they are

View file

@ -431,3 +431,81 @@ def test_the_offending_text_is_carried_verbatim() -> None:
assert isinstance(result, okf.UnreadableProvenance) assert isinstance(result, okf.UnreadableProvenance)
assert result.value == " - { by: human:a, at: 2026-01-01T00:00:00Z }" assert result.value == " - { by: human:a, at: 2026-01-01T00:00:00Z }"
assert result.file == str(path) assert result.file == str(path)
# --- Step 5: trust tier from the actor prefix -------------------------------------------------
def test_no_entries_at_all_is_unverified() -> None:
"""SPEC §5.3: "No ``verified`` key ⇒ unverified". ``None`` and the empty tuple say the same
thing here nothing was verified and both must reach the same floor."""
assert okf.trust_tier(None) == "unverified"
assert okf.trust_tier(()) == "unverified"
def test_a_machine_actor_gives_machine_confirmed() -> None:
assert okf.trust_tier(({"by": "process:finance-nightly", "at": "2026-01-01T00:00:00Z"},)) == (
"machine-confirmed"
)
def test_a_human_actor_gives_human_reviewed() -> None:
assert okf.trust_tier(({"by": "human:ktg", "at": "2026-01-01T00:00:00Z"},)) == "human-reviewed"
def test_the_tier_keys_off_the_PREFIX_never_a_substring() -> None:
"""``bot/human:2`` is a MACHINE actor whose identifier happens to contain ``human:``.
A substring test would promote it to the human tier inventing a human sign-off that never
happened, which is the fabricated-provenance defect this work removes, one level down.
"""
assert okf.trust_tier(({"by": "bot/human:2", "at": "2026-01-01T00:00:00Z"},)) == (
"machine-confirmed"
)
def test_an_entry_naming_no_actor_is_refused_never_tiered() -> None:
""" "Otherwise ⇒ machine-confirmed" would mint a tier out of an entry that names nobody.
Step 3 already refuses such an entry when it decodes a ``verified`` value; this door ASSERTS
that rather than assuming it, because ``trust_tier`` is public and its caller may not have come
through the decoder.
"""
with pytest.raises(ValueError) as excinfo:
okf.trust_tier(({"at": "2026-01-01T00:00:00Z"},))
assert "by" in str(excinfo.value)
def test_a_human_LAST_entry_still_resolves_to_human_reviewed() -> None:
"""AMENDMENT C — the ordering IS the arm.
``any(...)`` and "the last entry wins" agree on every single-entry case AND on a human-FIRST
list. They disagree on exactly one shape: process first, human last. That is the one fixture
that can tell the specified rule from the measured second-entry-wins defect, so it is the one
written here.
"""
process_first_human_last = (
{"by": "process:finance-nightly", "at": "2026-01-01T02:00:00Z"},
{"by": "human:ktg", "at": "2026-01-02T09:00:00Z"},
)
assert okf.trust_tier(process_first_human_last) == "human-reviewed"
# CONTROL — human FIRST, process LAST. Without it, a function that always answered
# `human-reviewed` would satisfy the arm above.
human_first_process_last = (
{"by": "human:ktg", "at": "2026-01-02T09:00:00Z"},
{"by": "process:finance-nightly", "at": "2026-01-01T02:00:00Z"},
)
assert okf.trust_tier(human_first_process_last) == "human-reviewed"
# CONTROL — two machine actors stay machine-confirmed, so the tier is not simply a function
# of the entry COUNT (SPEC §5.3 derives it from the actor prefix alone).
assert (
okf.trust_tier(
(
{"by": "process:a", "at": "2026-01-01T02:00:00Z"},
{"by": "process:b", "at": "2026-01-02T02:00:00Z"},
)
)
== "machine-confirmed"
)