# Work exchange intake, implementation and publication goal Objective: detailed intake of all reports; distribute important information throughout typed `.uai` memory; implement supported work that improves usefulness and adoption; deploy production and test. No reduced documentation-only goal. Latest implementation: reports/v049-work-task-recovery-20260913.md. Protected successor tasks retain immutable compact accounting with explicit generations; current-runtime drain preserves work recovery and bounded retention on schema3. Combined648tests pass; fresh native cold-process recovery now also passes both baseline paths in reports/v049-work-recovery-native-20260913.json. Initial warm- old/import-order compatibility, actual Passenger transition, production publication and public-doc workflow are not thereby proved. Full lifecycle metering, source-specific rights/dependencies and usefulness/intake work remain required. Historical checkpoint paragraphs below are not current omission or completion claims. The immediately preceding prompt-writing turn did not advance authoritative implementation state: no progress toward this implementation goal. This continuation revalidated the worktree and both queues, then resumed substantive review. Sixteen sources/718981 bytes are active; 67 historical records closed; registration integrity passed. Those counts are an observation, not completion. ## Workstreams and evidence gates 1. Full per-source review, claim verification, accepted/rejected/deferred reasons, implementation ownership, exact private custody and an empty intake gate. Main: strategy/UX/incentives. Parallel reviewers: protocol/reliability/schema/ verification; authorization/privacy/security/governance; pilot/integration/ settlement/resource costs. Review reports are not closure by themselves. 2. Existing owned runtime extended with a useful bounded work exchange: exact terms, consent, deliverable and artifact decision binding, discovery, refusal, expiry, interruption recovery and withdrawal. No second writable service, participant rank, native currency, implicit payment or untrusted host execution. 3. Practical public entry points, labeled seed task templates, current contracts, honest result/empty states, accessible existing brand presentation and optional runic venue link without transferred authority or credentials. 4. Typed memory distributed by owner, deterministic exports, unit/integration/ failure tests, current documentation-only agent workflow, exact production readback, disposable cleanup and clearly labeled tester limitations. 5. Scoped check-in/push and repository plus root-deploy WIP ZIPs from committed source, safe predecessor/rollback handling, authorized publication and final requirement-by-requirement evidence audit. Existing unrelated drafts remain. ## Current checkpoint Latest read/transport extension: `v049-work-reads-deadline-20260912.md`. Explicit current work GETs share the protected task budget, return selected content after saved accounting and keep only compact checkpoints. Standard work dispatch now uses native Windows worker containment and existing guard leases for abort and confirmed-exit retry boundaries. No new runtime/credential/body store. This supersedes older local-read/deadline omissions, not lifecycle metering, safe subsequent tasks, rights/dependencies or deployed/public-doc acceptance. Inspect the dated report for actual test/source/package evidence and remaining gates. Latest exact-resolution evidence: `v049-work-resolution-20260912.md`. The pending operation can now be reconciled to a current successful result or permanently prevented in its original slot, without executing it. Existing custody retains resolution intent, unknown exposure and cleanup accounting. Root/discovery/public source contracts are integrated. See that report for actual final tests/native results and remaining implementation/deployment gates. The complete objective and all16active source dispositions remain in scope; no reduced completion claim. Prior protected-client evidence: `v049-work-protected-client-20260912.md`. Existing custody now owns exact work intents and cumulative resource accounting; actual owned-handler tests verify response/save interruption, correction/withdrawal and credential rotation/recovery/revocation coexistence. Native Windows fixture storage is distinct from production. Rejected-intent reconciliation, safe later tasks, metered readback/lifecycle and hard deadlines remain concrete engineering gates. Sixteen sources stay active with selected work unfinished; the full intake, usefulness, memory, deployment and testing objective is unchanged. Superseding integration checkpoint: all16 source reviews are complete with exact private custody; local SQL3, work API/root/facade/replay, public /work/ views and generated public human/machine guides are implemented. Detailed evidence and remaining source-specific implementation/release gates are in `v049-work-exchange-integration-20260912.md`. The paragraphs below preserve the earlier staged observations, not the current runtime inventory. No work deployment is claimed. Native SQL/rollback, visual QA, protected client resource workflows and the actual public-doc pilot remain mandatory before release acceptance. Four strategic sources fully read and critically reviewed in `v049-work-exchange-strategy-review-20260912.md`. Three disjoint review agents are assigned the remaining twelve sources. Current live API available/MySQL; productionReady=false because autonomous acceptance is incomplete. No work exchange deployment, new outside adoption, or financial capability is claimed. Do not mark the goal complete while accepted implementation or publication gates remain open. Inspect actual agent handles before assuming parallel work continues. Further evidence: operations review completed four full sources and main read the complete resulting report. Eight sources now have full review entries; all sixteen have exact SHA-verified private custody snapshots (active intake originals remain). Pure `concresca_runtime/work_contract.py` implements bounded public terms, bilateral counter/accept/refuse/cancel, version-bound deliverable decisions, correction invalidation and owner withdrawal/redaction. Seventy local tests pass. Two added revision-limit regressions initially errored, then passed after fixing the limit so cancellation/withdrawal cannot be trapped by routine revision caps. The projection retains deadline/retention metadata after terms withdrawal and redacts payloads at retention expiry; actual SQL purge still requires integration. Intake suite: sixteen tests, fifteen pass and one existing permission skip. Exports regenerated and exact check passed. These are local checks, not a deployed work exchange, concurrency proof, financial authority or independent acceptance. Next implementation gates: finalize protocol/boundary review; integrate separate opportunity/bilateral agreement objects in the existing SQL transaction authority; prove exact replay, exclusive acceptance, scoped authorship and actual expiry purging; add persisted budgets/client workflow and first-party discovery/results; perform release/production/public-doc pilot gates above. An adversarial reviewer is examining the pure contract in a disjoint new test file before integration.