Development record
A debt nothing lists is a debt nobody pays. Every shipped intent in a managed repository carries an owed fidelity review, and no surface counts them.
A debt nothing lists is a debt nobody pays. Every shipped intent in a managed repository carries an owed fidelity review, and no surface counts them.
Observed in an autonomous run: twenty receipts owed in that repository, none ingested, and nothing in any status render, lint row or listing that says so. The obligation is real and by design — a shipped intent owes a fidelity review, that is what the receipt is for — but it accrues silently. There is no verb that answers "what is outstanding", so the only way to learn the number is to enumerate the shipped intents and check each for an ingested verdict, which is exactly the work the debt makes expensive and which nobody does unprompted.
Corroborating evidence from abcd's own repository on the same day: three receipts were owed here, and they were discharged only because a human asked what was outstanding. The tool that defines the obligation does not track it for itself either.
This is a designed obligation that silently accumulates, which is a sharper failure than a missing convenience. A debt with no counter grows until someone looks, and by then discharging it is a project rather than a step — twenty audits is a day's work, three is ten minutes. The receipts already exist as records, so the count is a directory read.
Wanted: a row in the status render or in abcd lint that names the number of owed fidelity reviews and the intents they belong to, and a listing verb that enumerates them. Related and filed separately: two defects that make an owed review expensive to discharge once found — the request carries no provenance hashes that ingest nevertheless requires, and it asks the host for a delivered diff range it has no mechanism to supply.
Corroboration (2026-09-18, Gropius managed-repo session gropiusllm-56, relayed
to abcd-17). Second managed repository, at v0.9.0, after twelve fidelity
audits in one day: abcd spec close prints "fidelity review OWED, receipt
rcp-…" at the close, and nothing afterwards lists the receipts still owed, so
the session closed the loop by grepping the decision log for receipt ids. Its
ask is a read-only listing, abcd intent audit --owed, so a session can find
what is outstanding without enumerating shipped intents by hand. Same shape as
the filing; the number this time was twelve, all paid, found by grep.
Grounds
- pursued: we expect a count of owed fidelity reviews on the bare status surfaces to be enough to make the debt get paid, because the debt was invisible rather than resisted, and a session that was asked what was outstanding discharged three in one sitting; it is shown wrong if the count is rendered and the debt still accumulates, which would mean visibility was not the constraint
- pursued: we expect a count of owed fidelity reviews on the bare status surfaces to be enough to make the debt get paid, because the debt was invisible rather than resisted; it is shown wrong if the count is rendered and the debt still accumulates, which would mean visibility was not the constraint
Frontmatter
| id | iss-2609100509537730 |
|---|---|
| type | issue |
| lifecycle | resolved |
| severity | major |
.abcd/work/issues/resolved/iss-2609100509537730-a-debt-nothing-lists-owed-fidelity.md
open on GitHub ↗ · commit history ↗
References1
- itd-2609150819445595A shipped intent's owed fidelity review is listed, so the debt is paidrelated
mentions1
- iss-2609252038344132Two low notes from the owed review: (1) intent audit --json echoes a dead-letter reason's attacker-supplied text verbatim, so a quoted-back token shaped like 'Raw payload retained at .../.work.local/...' puts a local-tier-looking string in the output (the real retention path is cut correctly; internal/core/intent/owed.go:120-147), and the test's .work.local substring check proves the fixture, not the property; (2) the resolution of iss-2609100509537730 appends a near-duplicate pursued grounds bullet after its corroboration prose.