Real-interface campaign reports and reproducible defects. These document observed behavior and its limits; they are not runtime modules.
| File | Role |
|---|---|
| efficiency-defaults-and-install-v0.1.14.md | Omitted-setting defaults, official provider/daemon journeys and compiler-wrapper installation boundary. |
| execution-efficiency-i0-cost-baseline.md | I0 completion contract, seven real-provider cost-ledger cases, independent usage reconciliation, diagnostic isolation, and bounded qualification evidence. |
| execution-efficiency-i1-indexed-projections.md | I1 completion, exact reference/index correspondence, measured large-history lookup work, real provider continuation and bounded recovery evidence. |
| execution-efficiency-i2-checkpoint-pages.md | I2 completion, default-enabled verified page reuse, syscall/write measurements, corruption/interruption recovery and real accepted product journeys. |
| execution-efficiency-i3-provider-layout.md | I3 completion, default once-only OpenCode schemas, real selection/correction/ProductRunner comparisons, full cached-input accounting and measured limits. |
| execution-efficiency-i4-decision-views.md | I4 completion, exact decision views, originating request dependencies, v6 reader guard and declared progress, native closure fixes, recovery and real provider qualification. |
| execution-efficiency-i5-plans-waits.md | I5 completion, sequential plans, owned waits, live control and permission revisions, exact publication/recovery evidence and paired native-provider observations. |
| execution-efficiency-i6-freshness.md | I6 shared full-capture qualification, platform fallback, real managed continuation/rollback, exact identity correspondence and bounded local savings. |
| pipeline-handoff-real-2026-10-26.md | Issue #159 promise/receipt checks, actual Codex/DeepSeek handoffs, read-only boundaries, and OpenCode SQLite quota failure/recovery. |
| task-continuity-real-2026-10-26.md | Issue #160 lost failed-read exchange, 24 real tool/context recovery journeys, source-admission boundaries, and live Codex/DeepSeek continuation. |
| read-argument-repair-real-2026-10-27.md | Issue #160 upstream argument correction, real OpenCode/DeepSeek fault prompting, baseline host rejection, repeated-error recovery, and Codex control. |
| red-team-phase8-real-2026-10-24.md | Final adversarial containment, corruption recovery, actual-provider brownfield repair, clean source-build/install/launch, and source-publication campaign. |
| red-team-phase8-evidence.json | Exact Phase 8 market, source-build, hostile replay, state recovery, brownfield, artifact, original-preservation, and check identities. |
| red-team-phase8-acceptance-ledger.md | Bounded A1–A14 final-program mapping with explicit platform, adapter, coverage, and claim limits. |
| red-team-phase7-real-2026-10-24.md | Complete TUI/WebUI campaign projection, source build/install/enable journey, restart recovery, failure-path UX, stable exports, and unchanged-original proof. |
| red-team-phase7-evidence.json | Exact Phase 7 host, source, package, run, export, restart, failure-path, and original-preservation identities. |
| red-team-phase6-real-2026-10-23.md | Copy-only remediation, four-category replay, independent patch review, live control recovery, original-byte proof, delivery artifacts, and source-market build campaign. |
| red-team-phase6-evidence.json | Exact Phase 6 candidate, patch, replay, recovery, source-build, report, bundle, and unchanged-original identities. |
| red-team-phase5-real-2026-10-23.md | Autonomous security-family routing, known and unknown disclosure discovery, host-canary probes, independent skeptical adjudication, context renewal, and report publication campaign. |
| red-team-phase5-evidence.json | Exact Phase 5 provider, package, run, report, classification, reuse, and 25 observed scenario records. |
| red-team-phase4-real-2026-10-23.md | Source catalog, bounded local build, disabled installation, TUI/WebUI actions, compatibility, cancellation, failure and restart recovery campaign. |
| red-team-phase4-evidence.json | Exact Phase 4 source/build/package identities, interface receipts, recovery records and 19 observed scenarios. |
| red-team-phase3-real-2026-10-23.md | Contained Linux jobs, trusted canary oracle, resource enforcement, dependency preparation, cancellation and recovery real-interface campaign. |
| red-team-phase3-evidence.json | Exact Phase 3 candidate identities, resource policy, job receipts and 13 observed scenarios. |
| red-team-phase2-real-2026-10-23.md | Scoped source, durable evidence, recovery, selective invalidation, and honest partial-report real-interface campaign. |
| red-team-phase2-evidence.json | Exact candidate identities and 24 observed Phase 2 source/evidence scenarios. |
| open-items-real-2026-10-22.md | Sequential issue/PR integration, released-versus-fixed IPC behavior and native TUI evidence; macOS campaign excluded. |
| configuration-upgrade-real-2026-10-21.md | Released CLI collision, shared generated-format migration, actual bootstrap/terminal restart, concurrency and recovery evidence. |
| configuration-upgrade-evidence.json | Binary/source identities and concrete outcomes for 43 generated-configuration upgrade scenarios. |
| model-capacities-real-2026-10-21.md | Route/model capacity resolution, shared output headroom, configured overrides, real catalogs, and live inference evidence. |
| model-capacities-evidence.json | Final executable hashes, actual catalog capacities, request metadata, and durable-memory readbacks. |
| read-loop-recovery-real-2026-10-20.md | Twelve read-loop findings, additional historical/private-source defects, and native/generated/live-provider recovery evidence. |
| read-loop-recovery-evidence.json | Final executable identity, 1,120 real API actions, independent boundary counts, and actual official-provider/CLI results. |
| structural-runtime-performance-2026-10-20.md | Post-fix runtime measurements, metadata snapshot scaling repair, and remaining source-grounded performance targets. |
| structural-runtime-real-2026-10-20.md | Native behavior evidence for publication, replay caching, deferred history, client boundaries, and plugin diagnostics. |
| bundled-skills-real-2026-10-20.md | Canonical skill migration, native exact-byte readers and automatic package/resource rebuild lifecycle. |
| opencode-parser-trace-real-2026-10-20.md | Repairs for all four v0.1.5 review findings: argument integrity, ambiguity/duplicate rejection, actual CLI memory and append/recovery evidence. |
| read-reuse-real-2026-10-20.md | Versioned visible-range reuse, archive restoration, changed-target refresh, retained task progress, and real API/live-provider evidence. |
| opencode-rejected-turn-diagnostic-real-2026-10-20.md | Retained rejected OpenCode replies, the local private-trace reader, and the loop's refusal/retry narration, with real-run measurement and scripted replay evidence. |
| credential-store-wsl-real-2026-10-20.md | Real provider-settings replay for an unreachable and a reachable WSL Secret Service, with the credential-store guidance and removal evidence. |
| opencode-wsl-interop-real-testing-2026-10-20.md | Windows OpenCode reached through WSL interop pinned as the runtime, actionable discovery failure, and real pre/post provider-setup evidence. |
| provider-progress-real-2026-10-19.md | Live 220-second wait diagnosis and real terminal/provider acceptance, reasoning, generation and cancellation feedback replay. |
| tui-conveniences-real-2026-10-19.md | Real terminal search, picker, clipboard, editor, queue, shortcuts, themes, and attachment recovery campaign. |
| incremental-context-real-2026-10-19.md | Incremental exchange selection and document reuse: exact before/after API behavior, freshness, permissions, pressure and recovery evidence. |
| checkpoint-cache-real-2026-10-19.md | Actual checkpoint/read reduction, official Codex cache reuse, current-context authority, invalidation and recovery campaign. |
| tui-formatting-recall-real-2026-10-19.md | Actual terminal Markdown and saved-message recall/search campaign, including delayed replies, reconnect, paging and draft preservation. |
| tui-feedback-real-2026-10-19.md | Observable contract and real-interface evidence for visible steering, file edits, and formatted terminal output. |
| tui-visible-work-real-2026-10-19.md | Corrective campaign for full useful tool output, live published summaries, and resumed-task plan reuse. |
| local-shutdown-2026-10-18.md | Broken advertised shutdown command, concurrent-stop recovery, real CLI replay, and builder installation repair. |
| local-shutdown-evidence.json | Source/executable identities and real CLI scenario outcomes for local shutdown. |
| codex-notifications-real-2026-10-18.md | Actual Codex advisory/model-access failure, decoder repair, live provider/TUI replays, and builder installation recovery. |
| codex-notifications-evidence.json | Exact source/binary identities, original/fixed decoding, live provider/TUI receipts, and retained-run reattachment. |
| automation-apparmor-real-2026-10-18.md | Actual Ubuntu release-installer failure, managed-browser namespace policy repair, and real sandbox replay evidence. |
| automation-apparmor-evidence.json | Native Ubuntu consumer hashes, exact AppArmor policies, unchanged restrictions and owned renderer observations. |
| research-quality-rubric-v2.md | Prospective, frozen semantic acceptance criteria for independent research; distinct from source matching and UI operation. |
| research-quality-rubric-v2.json | Exact structured version of the frozen research-quality criteria and evidence boundaries. |
| deep-research-provider-retry-evidence.json | Actual 125-second official-client timeout, automatic live-provider retry, exact marker and owned-process cleanup. |
| deep-research-ea-real-2026-10-17.md | Neutral-topic research baseline defects, context/ledger/coverage repairs and real replay evidence. |
| deep-research-inquiry-real-2026-10-17.md | User-rejected shallow report, frozen substantive research rubric, adaptive-inquiry repairs and real native/live-provider replay evidence. |
| deep-research-continuous-evidence.json | Exact package/host identities, focused report verdict and bounded native/TUI/GUI evidence for continuous research, steering, pause/recovery and immutable report reuse. |
| effective-altruism-research-report-2026-10-17.md | Earlier checked neutral-topic output later rejected as shallow; retained baseline, not the accepted controversy remake or product guidance. |
| deep-research-repair-evidence.json | Compact receipts for real host context retention, native follow-up recovery, CLI/HTTP behavior and the repaired live research campaign. |
| automation-setup-evidence.json | Compact native runtime, installer, launch and actual rendered-source receipts. |
| automation-setup-real-2026-10-17.md | Automatic browser/Python provisioning, native installation, repair, concurrency and live rendered-page evidence. |
| module-market-real-2026-10-17.md | Native Rust plugin market campaign, observed defect, bounded evidence and replay paths. |
| module-foundation-evidence.json | Compact native hashes, CLI counts, real browser actions, provider/job output identities and recovery receipts. |
| module-foundation-real-2026-10-17.md | Bundled/installed module lifecycle, real role isolation, terminal/browser/provider journeys, recovery defects and plugin process cleanup. |
| web-search-rust-real-2026-10-17.md | Zero-setup built-in Rust search, actual public queries, cancellation, migration and terminal/model evidence. |
| web-tools-real-2026-10-17.md | Keyless search, exact native source reads, real browser rendering, role/permission routing and retained defect replays. |
| reviewer-startup-real-2026-10-17.md | Startup/settings review-model selection, saved role routing, repair/offline behavior, and actual terminal/provider evidence. |
| retained-state-growth-real-2026-10-16.md | Removed retained-state quotas, paged migration, indexed scope inventory, reproduced failures, and real native/provider evidence. |
| token-budgets-real-2026-10-16.md | Local token-budget stop, complete working-state rendering, configuration compatibility, copied archive recovery, and live deployment evidence. |
| binary-evidence-real-2026-10-16.md | NUL-bearing PostgreSQL diff failure, binary-aware rendering, exact historical evidence recovery, and real replay evidence. |
| in-place-runtime-churn-real-2026-10-15.md | Live PostgreSQL candidate/checkpoint failures, independent original/fixed terminal replays, and builder installation recovery. |
| tui-reconnect-real-2026-10-15.md | Terminal transport recovery, exact Unicode draft retention, old-binary reproduction, and independent real replay. |
| handshake-stall-real-2026-10-15.md | Long-history prompt-planning stall: live profiling, exact brownfield snapshot replay, worker isolation and first-attempt attachment. |
| startup-recovery-real-2026-10-15.md | Cold-start timeout, large-archive storage migration, folder identity and interrupted-socket recovery evidence. |
| guided-design-real-2026-10-15.md | Interactive interview, durable Markdown, read-only/stale updates, implementation handoff, and browser readability evidence. |
| second-instance-real-2026-10-15.md | Second-launch daemon replacement: root cause, real replay evidence, and limits. |
| opencode-malformed-output-real-2026-10-15.md | Issue 127: real malformed model replies, bounded correction, Go diagnostics, and provider isolation. |
| composer-modes-real-2026-10-13.md | Document: Composer navigation and mode explanations — GitHub #121 / #122. |
| computer-automation-campaign-2026-10-11.md | Document: Browser and native GUI real-testing campaign. |
| computer-automation-defects-2026-10-11.md | Document: Computer automation: confirmed real defects. |
| context-resume-real-2026-10-13.md | Document: Fixer context preparation repair — October 13, 2026. |
| daemon-disconnects-2026-10-12.md | Document: Long-history daemon stalls and disappearing disconnect diagnostics. |
| document-reading-2026-10-11.md | Document: Larger document reading: real campaign and defects. |
| history-recovery-2026-10-11.md | Document: Oversized context and durable-history recovery. |
| in-place-directory-real-testing-2026-10-13.md | Document: In-place directory and memory recovery: real testing, 2026-10-13. |
| macos-command-path-real-2026-10-13.md | Document: macOS installer command-path failure. |
| opencode-runtime-real-testing-2026-10-13.md | Document: OpenCode runtime: real-interface campaign, 2026-10-13. |
| proactive-bug-discovery.md | Document: Proactive bug discovery. |
| prompt-skills-real-2026-10-14.md | Document: Prompt cues and Rust reliability real testing. |
| readline-real-2026-10-14.md | Document: Readline TUI editing real testing — issue 126. |
| real-testing-provider-restart-2026-10-11.md | Document: Provider settings restart and historical process recovery. |
| real-testing-skill-and-coverage-loop-2026-10-11.md | Document: Real-testing skill and missing-check-contract loop repair. |
| release-0.0.6-defects.md | Document: Release 0.0.6 real defects. |
| rsi-f0-real-2026-10-12.md | Document: RSI E3/F0 real execution campaign — 2026-10-12. |
| rsi-production-real-2026-10-13.md | Document: Native RSI production repair and real campaign — 2026-10-13. |
| rsi-real-testing-2026-10-12.md | Document: RSI real testing — 2026-10-12. |
| runtime-friction-2026-10-12.md | Document: Long-run runtime friction: real testing. |
| test-retirement-scope-restoration-2026-10-10.md | Document: Test-retirement scope restoration — October 10, 2026. |
| Folder | Role / next waypoint |
|---|---|
| evidence/ | Retained real-interface observations grouped by boundaries, framing, varied inputs, mutations, and recovery campaigns. |
| reproducers/ | Retained drivers/consumers for replaying actual program defects with disposable state and independently stated expectations. |
Use each report together with its named reproducer, source identity, and independent observation boundary. Separate historical results from current behavior, keep new mutable artifacts outside the repository unless they are intentionally retained evidence, and route implementation changes through the linked code owner.
Navigation: Parent waypoint · Repository map.
Keep this waypoint aligned when local files or child responsibilities change. Descriptions are orientation, not a substitute for reading the current implementation.
Deep Research delivery evidence records focused native replay and separates the v1.2 delivery failure from pending report quality.
Grounding efficiency documents task-scoped observation/design reuse, prompt delivery, compaction, metrics, and actual replay evidence.
Single grounding check records the enforced designer handoff, durable readiness, actual provider replay, and target-specific source safety.