morluto/rea

Fork 0

mirror of https://github.com/morluto/rea.git synced 2026-11-05 06:45:22 +00:00

[GH-ISSUE #4444] [Bug] build_reconstruction_obligation_ledger reports a missing review record or duplicate record IDs as unreadable_output ("run rea doctor") #1762

New issue

Closed

opened 2026-10-16 21:09:59 +00:00 by bsakaguchi · 2 comments

bsakaguchi commented

2026-10-16 21:09:59 +00:00

Owner

Originally created by @Rinkia on GitHub (Oct 16, 2026).
Original GitHub issue: https://github.com/morluto/rea/issues/4444

Before submitting

I searched the open and closed issues for a related report (obligation ledger, unreadable_output, "Reviewed obligation Evidence is missing"). Sibling of #3219, which was fixed by 3a34ad37b for non-canonical bundles only. Open PR #3288 adds a test for that same non-canonical case and doesn't change the error mapping.
I removed credentials, tokens, private paths, and sensitive target data.

Area

Evidence, artifacts, or reconstruction

Observed behavior

Two caller-input mistakes make build-reconstruction-obligation-ledger / build_reconstruction_obligation_ledger fail as an internal error:

{"error":"Application workflow failed","code":"unreadable_output","category":"execution_failure",
 "message":"Analysis returned an unreadable result. Run `rea doctor` and review the reported details; repeating the unchanged request is not expected to succeed.",
 "retryable":false}

There are no details, so the caller can't tell which input is wrong.

A review_evidence_ids entry that isn't in evidence_bundle.records. This is the case a user hits by following docs/reconstruction-obligation-ledgers.md literally: paste one of the "Reviewed obligation examples" (sample ID ev_aaaa…) into the "Complete CLI request" from the same page.
An Evidence bundle that contains the same record twice.

evaluate_reconstruction_coverage returns invalid_request / invalid_input for the equivalent inputs, with the reason in details.issues:

{"code":"invalid_request","category":"invalid_input",
 "details":{"issues":[{"path":[],"reason":"invalid_value","message":"Evidence bundle contains duplicate record IDs"}]}}

and Coverage workspace Evidence is missing: ev_… for a missing cited record.

Cause (source-derived on main @ d72d856f5): deriveReconstructionObligationCandidates throws TypeError("Reviewed obligation Evidence is missing: <id>") (src/application/ReconstructionObligationCandidates.ts:326), and bundle parsing throws TypeError("Evidence bundle contains duplicate record IDs") (src/domain/evidenceBundle.ts). buildReconstructionObligationLedgerEvidenceValidated (src/application/ReconstructionObligationLedgerService.ts) maps only NonCanonicalEvidenceBundleError and z.ZodError to workflowInputError. Every other TypeError becomes AnalysisProtocolError, which is presented as unreadable_output. ReconstructionCoverageService.ts maps cause instanceof TypeError || cause instanceof z.ZodError to workflowInputError. Other caller-reachable TypeErrors in the same path, such as the residual-unknown graph checks in evidenceBundle.ts, would presumably be affected the same way (not tested).

Reproduction

Minimal request: an empty bundle plus one reviewed obligation whose review_evidence_ids is ["ev_" + "a"×64]. The other fields are the docs' example fields with required_case_kinds: ["positive"].

rea build-reconstruction-obligation-ledger /tmp/min12.json --json

Same result on rea-agents@6.5.0, the main @ d72d856f5 CLI and the main stdio MCP tool. For case 2, append a copy of evidence_bundle.records[0] to an otherwise valid request.

Expected behavior

invalid_request / invalid_input with the TypeError message as an issue, as evaluate_reconstruction_coverage already does. That means mapping caller-input TypeErrors in the ledger service the same way, or throwing typed input errors at those two sites.

Impact and scope

Anyone trying the documented example, and anyone whose bundle and reviewed obligations drift apart. The message sends them to rea doctor, which doesn't help.
Tested: the two cases above on 6.5.0 and main. Everything else I tried in the ledger was classified as documented: verifier fail → failed, a contradiction → contradicted, a duplicate binding → unowned, a missing dependency → blocked, out-of-scope kept as a disposition, and static or inferred Evidence flagged as weak-*-authority so process and runtime claims stay open.

Originally created by @Rinkia on GitHub (Oct 16, 2026). Original GitHub issue: https://github.com/morluto/rea/issues/4444 ### Before submitting - [x] I searched the open and closed issues for a related report (obligation ledger, unreadable_output, "Reviewed obligation Evidence is missing"). Sibling of #3219, which was fixed by 3a34ad37b for non-canonical bundles only. Open PR #3288 adds a test for that same non-canonical case and doesn't change the error mapping. - [x] I removed credentials, tokens, private paths, and sensitive target data. ### Area Evidence, artifacts, or reconstruction ### Observed behavior Two caller-input mistakes make `build-reconstruction-obligation-ledger` / `build_reconstruction_obligation_ledger` fail as an internal error: ```json {"error":"Application workflow failed","code":"unreadable_output","category":"execution_failure", "message":"Analysis returned an unreadable result. Run `rea doctor` and review the reported details; repeating the unchanged request is not expected to succeed.", "retryable":false} ``` There are no `details`, so the caller can't tell which input is wrong. 1. **A `review_evidence_ids` entry that isn't in `evidence_bundle.records`.** This is the case a user hits by following `docs/reconstruction-obligation-ledgers.md` literally: paste one of the "Reviewed obligation examples" (sample ID `ev_aaaa…`) into the "Complete CLI request" from the same page. 2. **An Evidence bundle that contains the same record twice.** `evaluate_reconstruction_coverage` returns `invalid_request` / `invalid_input` for the equivalent inputs, with the reason in `details.issues`: ```json {"code":"invalid_request","category":"invalid_input", "details":{"issues":[{"path":[],"reason":"invalid_value","message":"Evidence bundle contains duplicate record IDs"}]}} ``` and `Coverage workspace Evidence is missing: ev_…` for a missing cited record. **Cause** (source-derived on `main` @ `d72d856f5`): `deriveReconstructionObligationCandidates` throws `TypeError("Reviewed obligation Evidence is missing: <id>")` (`src/application/ReconstructionObligationCandidates.ts:326`), and bundle parsing throws `TypeError("Evidence bundle contains duplicate record IDs")` (`src/domain/evidenceBundle.ts`). `buildReconstructionObligationLedgerEvidenceValidated` (`src/application/ReconstructionObligationLedgerService.ts`) maps only `NonCanonicalEvidenceBundleError` and `z.ZodError` to `workflowInputError`. Every other `TypeError` becomes `AnalysisProtocolError`, which is presented as `unreadable_output`. `ReconstructionCoverageService.ts` maps `cause instanceof TypeError || cause instanceof z.ZodError` to `workflowInputError`. Other caller-reachable `TypeError`s in the same path, such as the residual-unknown graph checks in `evidenceBundle.ts`, would presumably be affected the same way (not tested). ### Reproduction Minimal request: an empty bundle plus one reviewed obligation whose `review_evidence_ids` is `["ev_" + "a"×64]`. The other fields are the docs' example fields with `required_case_kinds: ["positive"]`. ```sh rea build-reconstruction-obligation-ledger /tmp/min12.json --json ``` Same result on `rea-agents@6.5.0`, the `main` @ `d72d856f5` CLI and the `main` stdio MCP tool. For case 2, append a copy of `evidence_bundle.records[0]` to an otherwise valid request. ### Expected behavior `invalid_request` / `invalid_input` with the `TypeError` message as an issue, as `evaluate_reconstruction_coverage` already does. That means mapping caller-input `TypeError`s in the ledger service the same way, or throwing typed input errors at those two sites. ### Impact and scope - Anyone trying the documented example, and anyone whose bundle and reviewed obligations drift apart. The message sends them to `rea doctor`, which doesn't help. - Tested: the two cases above on 6.5.0 and main. Everything else I tried in the ledger was classified as documented: verifier `fail` → `failed`, a contradiction → `contradicted`, a duplicate binding → `unowned`, a missing dependency → `blocked`, `out-of-scope` kept as a disposition, and static or inferred Evidence flagged as `weak-*-authority` so process and runtime claims stay open.

bsakaguchi commented

2026-10-16 21:10:00 +00:00

Author

Owner

@amyanger commented on GitHub (Oct 16, 2026):

I'd like to tackle this one if it is okay!

@amyanger commented on GitHub (Oct 16, 2026): I'd like to tackle this one if it is okay!

bsakaguchi referenced this issue

2026-10-17 21:01:32 +00:00

[GH-ISSUE #2383] [Tracker] Audit analysis and debugging coverage for the TIOBE top 100 languages (October 2026) #794

bsakaguchi closed this issue

2026-10-17 21:03:26 +00:00

bsakaguchi commented

2026-10-17 21:03:27 +00:00

Author

Owner

@matbrgz commented on GitHub (Oct 16, 2026):

Cross-referencing with sibling obligation ledger report #3370 ([Bug] Reconstruction obligation ledgers lose semantic unknowns and report complete coverage). Both issues isolate validation and outcome projection defects in build_reconstruction_obligation_ledger.

Consolidated under Cluster V (Reconstruction Obligation Ledgers & Ledger Integrity) in tracker #4330.

@matbrgz commented on GitHub (Oct 16, 2026): Cross-referencing with sibling obligation ledger report #3370 (`[Bug] Reconstruction obligation ledgers lose semantic unknowns and report complete coverage`). Both issues isolate validation and outcome projection defects in `build_reconstruction_obligation_ledger`. Consolidated under **Cluster V (Reconstruction Obligation Ledgers & Ledger Integrity)** in tracker #4330.

bsakaguchi referenced this issue

2026-10-31 00:55:14 +00:00

[PR #1762] [MERGED] fix(javascript): honor package exports before sibling files #4802

bsakaguchi referenced this issue

2026-10-31 00:55:29 +00:00

[PR #1803] [CLOSED] fix(javascript): retain unbound require semantic module links #4827