# Current remote qualification and historical results

## Current OpenCode catalog: 2 passed scenarios, 7 failed scenarios, 235 unrun

The denominator remains **244 scenarios / 35 areas / 191 frozen capabilities**. Nine unique scenarios have actual remote measurements: **2 passed, 7 failed**. Ten attempts were retained: **2 passed attempts, 8 failed attempts**. Backlog's final correction is a second attempt on the same scenario, never an additional scenario. Historical local passes remain separate. Authored callbacks and the235 unrun scenarios are not passing E2E evidence.

## Latest eight-case AWS wave: 2 passed, 6 failed

[Actual run40575715287](https://github.com/paperclipai/paperclip/actions/runs/40575715287) used exact source `1c72e2d74e77f4761d64a5bf571684fa6ccb2311`, production baseline `50b932a142216331b8d476042b454bc123739c40`, OpenCode/OpenRouter `openrouter/deepseek/deepseek-v4.1-flash-1017`, embedded PostgreSQL and separate verified AWS fleet hosts. Maximum concurrency4; automatic retries0; original queries, bounds and independent oracles unchanged. All eight cleanups passed. Unpriced dollars and automatic hello usage remain unknown, not zero.

**Actually passed on the current remote lane:**

- `agents.change-manager`:9/9 recorded checks, one native run; actual saved reporting line, attribution and preserved unrelated state.
- `companies.rename-company`:10/10 recorded checks, one native run; same company renamed, memberships/work preserved and actual actor attribution.

These pass measurements are bound to authenticated build artifact12767846870 (ZIP SHA256 `8498afb301f5a1292a354900f9a503dfafc0542f2c0632727d21d9fa3876096f`), retained native binary bytes and actual source/checksum-restore receipts; production paths remain byte-identical to50b. They do not qualify all other features or the historical baseline.

**Failed in this wave:**

- `tasks.create-backlog`: final allowed correction failed; creating a todo/self-assigned task queued unwanted work and exceeded the declared run bound. One source execution and a queued row are not two completed native executions. Its original setup failure remains retained; the original unknown hello cause is not retroactively called401. No further unchanged retry.
- `agents.hire-denied`: setup failed before canonical input,0 runs; public permissions fixture lacked required boolean `canAssignTasks`. No model denial outcome was measured.
- `tasks.comment-with-reference`: settled observation failed `one-attributed-note`; no passing note outcome claimed.
- `inbox-search.archive-without-grant`: an observed provider run failed; original failure classification and evidence retained.
- `inbox-search.show-my-work`: an observed provider run failed; original failure classification and evidence retained.
- `tasks.ambiguous-task`: timed out waiting for actual clarification; no invented clarification or passing choice outcome.

This standard presentation uses original result bytes through a closed public projection. Raw transcripts/logs, private account data and credentials are not published. The report's2/8 selection rate is distinct from the whole catalog's2/244 current scenario rate.

## Delegation pilot — actual canonical request, overall FAILED

[Run40572484517](https://github.com/paperclipai/paperclip/actions/runs/40572484517), initial ordinal1, source `2f68f4ff3c4be3ff608823e556cc000eb6948a06`, production `50b932a142216331b8d476042b454bc123739c40`, exact OpenCode/OpenRouter `openrouter/deepseek/deepseek-v4.1-flash-1017`, ordinary embedded PostgreSQL on a verified AWS fleet host. The original result remains FAILED (`transient_infrastructure`), with an observed CEO native-session timeout at the unchanged120-second bound. No evidence-backed correction or unchanged retry is claimed.

Exactly one canonical human request was persisted. The CEO created the requested task; a later mutation assigned Morgan. Six CEO tool executions completed, including a document write, but no terminal protocol frame arrived before the timer. Morgan never admitted. The second retained run row stayed queued and was cancelled during cleanup; it is not a second native execution. The announcement, worker provenance, completion reply and usable saved-result link were not proven.

The finally observation recorded **4/14 checks passing** (one human request, one assigned task, captured request, agent-created task). This is partial creation evidence, not a returned completed feature grade or a passing scenario. Actual aggregate usage was **165536 tokens**; unpriced dollars and automatic hello usage remain unknown, not zero. Cleanup passed. Actual framework interval19:58:41.606–20:01:18.853UTC. Artifact12766536663 SHA256 `2785814c8483b0f47218f99dcac2dc4b86b41eabb73cdb72cf98c65cad19c523`; raw transcripts/logs and private identifiers are excluded from this public presentation.

## Latest repaired-credential diagnostic — overall FAILED, outside the244

[Run40561571573 attempt2](https://github.com/paperclipai/paperclip/actions/runs/40561571573/attempts/2) is the authorized same-run rerun after the exact CI environment credential was replaced with direct human approval. It reused the original successful same-run build at source `2f68f4ff3c4be3ff608823e556cc000eb6948a06`, production baseline50b932 and unchanged OpenCode/OpenRouter `openrouter/deepseek/deepseek-v4.1-flash-1017`. It ran on actual AWS On-Demand/x86_64 with ordinary embedded PostgreSQL. No global fleet setting, product assertion, model or source changed for the rerun.

**The overall result remains FAILED:** `permanent_infrastructure`, **“Required evidence missing: final-state.png, snapshots/api-state.json.”** The evidence packaging failure is not relabeled as a pass. No more health or feature operation is implied by this report.

Independent retained receipts prove the single current-key request returned **HTTP200**, `recognized: true`, `expired: false`, key-budget availability true (no amount retained). Exactly **one normal lifecycle hello passed**, the agent was ready and its config matched. All three setup-health checks passed: ordinary readiness, one ordinary hello, and zero chat/task work. The real API reported0 issues,0 conversations,0 comments and0 heartbeat runs. Cleanup passed. Key-budget availability does not prove account credit or all model access; automatic hello tokens/cost are unknown outside heartbeat accounting.

This functional setup proof is outside244, not a feature scenario pass or a new scenario ordinal. Original attempt1 [HTTP401 report](https://d1p6rlowie26tp.cloudfront.net/runner-e2e/campaigns/gha-40561571573-1-auth-report/) and artifact12763471080 remain unchanged and separate. Attempt2 authenticated artifact12764569482 SHA256 `368659db7e0bb016914e98db997919890be40c347011b3fc138af3c497cf95e8`. No credential values, account identities, provider response bodies, raw logs or screenshots are published.

The prior [embedded health report](https://d1p6rlowie26tp.cloudfront.net/runner-e2e/campaigns/gha-40553618239-1-report-r2/) preserves source78de/run40553618239 and the original “User not found.” hello failure. These diagnostic operations remain distinct from the two failed catalog initials and from historical local passes.

## Historical local Codex:17 original initial passes

These are original local one-time measurements of provider `codex`, model `gpt-5.6-sol`, production baseline8373 with original per-attempt source/definition identities. They are **not current OpenCode or updated-production qualification**. Across158 attempted IDs:17 initial passes,141 initial failures,159 retained measurements including one historical correction, and86 unrun. [Immutable historical report](https://d1p6rlowie26tp.cloudfront.net/runner-e2e/campaigns/historical-local-codex-chat-catalog-20261027-158measured-86unrun/) and [original summary](https://d1p6rlowie26tp.cloudfront.net/runner-e2e/campaigns/historical-local-codex-chat-catalog-20261027-158measured-86unrun/summary.md). Unpriced cost is unknown, not zero.

- access.private-resource
- agents.agent-api-key
- assistant-connections.decline-consent
- chat-channels.provider-disabled
- companies.archive-company
- companies.rename-company
- inbox-search.archive-partial-failure
- instance-operations.instance-settings
- projects.project-work
- questions-and-approvals.accept-one-proposal
- questions-and-approvals.ambiguous-yes
- runs-adapters.list-model-options
- steering.preserve-draft
- task-coordination.reject-dependency-cycle
- tasks.create-backlog
- tasks.delegate-announcement
- tasks.website-task

## Six incomplete or blocked journeys

These remain incomplete/blocked despite registration and prerequisite probes; no passing E2E coverage is claimed.

- access.invite-teammate
- access.manage-membership
- activity.routine-audit
- agents.restore-instructions
- budgets-costs.subscription-fees
- skills.enable-for-agent

## All244 current remote statuses

### Passed current remote scenarios (2)

- agents.change-manager
- companies.rename-company

### Failed current remote scenarios (7)

- tasks.create-backlog (two retained failed attempts; correction budget exhausted)
- tasks.delegate-announcement
- agents.hire-denied (setup before canonical)
- tasks.comment-with-reference
- tasks.ambiguous-task
- inbox-search.show-my-work
- inbox-search.archive-without-grant

### Unrun current remote scenarios (235), grouped by area

#### access (6)

- access.cli-authorization
- access.invite-teammate
- access.manage-membership
- access.other-company
- access.private-resource
- access.sign-in

#### activity (6)

- activity.agent-feedback
- activity.audit-change
- activity.company-overview
- activity.find-runs
- activity.routine-audit
- activity.work-timeline

#### agent-chat (6)

- agent-chat.chat-disabled
- agent-chat.conference-room
- agent-chat.multiple-handoffs
- agent-chat.remember-and-follow-up
- agent-chat.reopen-conversation
- agent-chat.track-delegated-work

#### agents (14)

- agents.agent-api-key
- agents.ambiguous-cloud-code
- agents.choose-primary
- agents.configure-required-agent
- agents.edit-instructions
- agents.existing-grok-bot
- agents.hire-claude-code
- agents.hire-needs-account
- agents.hire-needs-approval
- agents.managed-profile
- agents.pause-agent
- agents.restore-instructions
- agents.resume-agent
- agents.retire-agent

#### app-permissions (6)

- app-permissions.ask-before-action
- app-permissions.grant-one-agent
- app-permissions.new-quarantined-action
- app-permissions.read-allowed-write-off
- app-permissions.revoke-account
- app-permissions.test-as-agent

#### assistant-connections (6)

- assistant-connections.connect-external-assistant
- assistant-connections.consent-read-only
- assistant-connections.decline-consent
- assistant-connections.device-login
- assistant-connections.revoke-personal-grant
- assistant-connections.use-scoped-client

#### budgets-costs (6)

- budgets-costs.explain-hard-stop
- budgets-costs.repair-cost-recording
- budgets-costs.resolve-budget
- budgets-costs.set-budget
- budgets-costs.spend-breakdown
- budgets-costs.subscription-fees

#### cases (8)

- cases.browse-cases
- cases.case-history
- cases.cases-disabled
- cases.close-case
- cases.create-case
- cases.handoff-same-task
- cases.related-work
- cases.update-case-facts

#### chat-channels (8)

- chat-channels.channel-file
- chat-channels.connect-slack
- chat-channels.link-sender
- chat-channels.outbound-email
- chat-channels.provider-disabled
- chat-channels.repair-endpoint
- chat-channels.reply-in-thread
- chat-channels.reply-to-email

#### cli-operations (6)

- cli-operations.choose-cli-context
- cli-operations.create-from-cli
- cli-operations.export-document-cli
- cli-operations.feedback-bundle
- cli-operations.inspect-run-cli
- cli-operations.owned-dev-instance

#### companies (8)

- companies.agent-observability
- companies.archive-company
- companies.configure-decision-model
- companies.export-company
- companies.fast-responses
- companies.import-preview
- companies.prompt-injection-guard
- companies.switch-company

#### connection-setup (13)

- connection-setup.channel-prerequisites
- connection-setup.choose-provider
- connection-setup.notion-agent-unassigned
- connection-setup.notion-connected
- connection-setup.notion-declined
- connection-setup.notion-expired
- connection-setup.notion-missing
- connection-setup.notion-wrong-account
- connection-setup.return-from-login
- connection-setup.save-and-resume
- connection-setup.secure-api-key
- connection-setup.unreachable-mcp
- connection-setup.wrong-person-resolves

#### developer-labs (4)

- developer-labs.component-reference
- developer-labs.gated-lab
- developer-labs.interaction-fixture
- developer-labs.long-thread-performance

#### documents-artifacts (7)

- documents-artifacts.find-artifact
- documents-artifacts.private-download
- documents-artifacts.produce-download
- documents-artifacts.resolve-feedback
- documents-artifacts.restore-document
- documents-artifacts.revise-delivery
- documents-artifacts.write-plan

#### execution-environments (5)

- execution-environments.add-computer
- execution-environments.assign-computer
- execution-environments.choose-capable-target
- execution-environments.prepare-image
- execution-environments.test-computer

#### gateways-profiles (5)

- gateways-profiles.connect-client
- gateways-profiles.create-gateway
- gateways-profiles.define-profile
- gateways-profiles.inspect-gateway-error
- gateways-profiles.rotate-token

#### goals (3)

- goals.align-existing-work
- goals.create-goal-tree
- goals.explain-purpose

#### inbox-search (6)

- inbox-search.archive-done
- inbox-search.archive-partial-failure
- inbox-search.find-design
- inbox-search.notification-audience
- inbox-search.restore-archive
- inbox-search.show-decisions

#### instance-operations (6)

- instance-operations.backup-and-restore
- instance-operations.diagnose-health
- instance-operations.instance-settings
- instance-operations.provision-runtime
- instance-operations.restart-service
- instance-operations.update-instance

#### navigation-preferences (6)

- navigation-preferences.broken-link
- navigation-preferences.dismiss-announcement
- navigation-preferences.edit-profile
- navigation-preferences.find-project-view
- navigation-preferences.quick-search
- navigation-preferences.sidebar-order

#### onboarding (5)

- onboarding.choose-first-runtime
- onboarding.first-work
- onboarding.new-company
- onboarding.new-instance
- onboarding.resume-wizard

#### plugins (4)

- plugins.configure-plugin
- plugins.disable-plugin
- plugins.install-plugin
- plugins.use-plugin-tool

#### projects (8)

- projects.add-repository
- projects.archive-project
- projects.create-project
- projects.private-project
- projects.project-defaults
- projects.project-work
- projects.replace-repository
- projects.share-project

#### questions-and-approvals (12)

- questions-and-approvals.accept-one-proposal
- questions-and-approvals.ambiguous-yes
- questions-and-approvals.approve-tool-once
- questions-and-approvals.decline-tool
- questions-and-approvals.defer-and-reopen
- questions-and-approvals.duplicate-answer
- questions-and-approvals.formal-hire-revision
- questions-and-approvals.multi-part-question
- questions-and-approvals.necessary-question
- questions-and-approvals.reject-and-revise
- questions-and-approvals.review-plan
- questions-and-approvals.wrong-reviewer

#### recovery (8)

- recovery.browser-reconnect
- recovery.controller-restart
- recovery.explain-stall
- recovery.post-stop-message
- recovery.retry-exhausted
- recovery.source-action
- recovery.unknown-email-outcome
- recovery.workspace-repair

#### routines (7)

- routines.create-daily-brief
- routines.manage-other-owner
- routines.pause-owned
- routines.plugin-managed-routine
- routines.read-history
- routines.run-now
- routines.webhook-trigger

#### runs-adapters (5)

- runs-adapters.inspect-failed-run
- runs-adapters.list-model-options
- runs-adapters.reset-session
- runs-adapters.upgrade-runner
- runs-adapters.validate-hire

#### secrets (6)

- secrets.ask-for-plaintext
- secrets.configure-company-secret
- secrets.least-scope-delivery
- secrets.personal-secret
- secrets.secret-proposal
- secrets.vault-reference

#### skills (7)

- skills.author-skill
- skills.enable-for-agent
- skills.explicit-restriction
- skills.find-and-install
- skills.fork-read-only
- skills.review-update
- skills.studio-test

#### status-cards (5)

- status-cards.archive-card
- status-cards.create-card
- status-cards.explain-card
- status-cards.other-authors-card
- status-cards.refresh-card

#### steering (8)

- steering.approval-delivery
- steering.discard-queued-message
- steering.edit-queued-message
- steering.interrupt-with-direction
- steering.preserve-draft
- steering.queue-follow-up
- steering.steer-active
- steering.stop-versus-pause

#### task-coordination (5)

- task-coordination.block-on-design
- task-coordination.delegate-with-context
- task-coordination.independent-review
- task-coordination.pause-tree
- task-coordination.reject-dependency-cycle

#### tasks (9)

- tasks.assignment-conflict
- tasks.attachment-brief
- tasks.completion-with-gate
- tasks.edit-properties
- tasks.organize-open-work
- tasks.private-task
- tasks.share-private-task
- tasks.website-task
- tasks.website-without-project

#### teams (4)

- teams.configure-team
- teams.find-team
- teams.install-team
- teams.preview-install

#### workspaces (7)

- workspaces.close-and-reopen
- workspaces.inspect-git
- workspaces.locate-files
- workspaces.prepare-repository
- workspaces.preview-service
- workspaces.private-repository-no-access
- workspaces.recover-lost-sandbox

This report is regenerated only from retained bytes; publication caused zero model/provider calls. Historical immutable prefixes are unchanged. No unmerged core PR contributes a passing result.