AI_README.md -- entry point; step 0b says check board.worklog first.AI_PORTAL.md -- what this repo is, and what the publication-staging tree is not.labtalk/ai_portal/AI_ENGINEERING_STANDARDS_SEED_V1.md -- contracts, regression doctrine,
definition of done, house conventions, git hygiene.docs/ai-friendly/AI_ROLES_TAXONOMY_V1.md -- which kind of AI you are talking to.| Lane | What it is | Evidence | Newest run (return here) |
|---|---|---|---|
AIF-045 |
Identity/RBAC/authorization management for the AI Portal, 2026-07-21 | **ACTIVE — core stack implemented, committed, and runtime-proven through 2d-3.** M0/M1 strong-ID domain and deny-precedence resolver; nine-table identity DBF persistence; APH-5 save/reload/compare proof; DBF boot adoption with degraded read-only fallback; USER admission/request/approve/grant/revoke/delete; low-privilege unauthenticated `member.public` startup; principal + acting-member session model; human credential and owner-issued rotatable AI/service token authentication. Real enforcement is wired into the AI dev-tools gate, `!` host shell, and SFTP; `DOTTALK_ALLOW_HOST_COMMANDS`/network policy remain independent final constraints. `SECURITY LOGIN` is explicitly a legacy diagnostic selector and grants no USER/RBAC access. Commits include `c59f688ca`, `b44a32fe2`, `0d4b9b407`, `be021e8b7`, `cf977b160`, `6b350ffed`, `2a89cc5c3`, `bfa6bb0fd`, `8afac8f63`. Open: M4 registry import, M5 portable YAML bootstrap, M6 teaching loop. Dev-stage, not promoted. | AIPR-20260722-005 |
AIF-046 |
Scan-evaluator optimization session `AIPR-20260722-007`, 2026-07-22 | **M0–M4 complete, committed dev-only; mirror has M0 only; not promoted/pushed** — parent `project.x64base.runtime` | AIPR-20260722-007 |
AIF-050 |
AI run traceability governance session `AIPR-20260722-007`, 2026-07-22 | **M0–M7 landed as candidates dev-only; M8 (Agent-Sync + peer) remaining; not promoted — parent `project.x64base.runtime`.** Owner: member.derald · steward/author: member.ai.claude.cowork (assigned 2026-07-22). **Records who did the work, in which run, reachable how — separate from who owns/commits.** Motivation is measured, not asserted: git attributes 100% of AI work to the maintainer (he commits); the report-audit envelope traces only to product level (`model`/`session` = `not_exposed`); and planner≠implementer is invisible (identity/RBAC was **ChatGPT-planned, Cowork-implemented, Derald-owned**). **Delivered:** the five-role attribution contract (owner/committer/author/planner/attestor); the RUN registry `ai_runs.yaml` seeded with real runs + a lane→newest-run return index (`current_by_lane`); the universal per-file `@dottalk.file` contract (a new AIF-042 KIND) with a runnable census/coverage tool — measured **1009 source files, 227 (22%) carry `@dottalk.usage`, 782 (78%) invisible to the harvest today**; the `ai-report-audit-v2` envelope (`authored_by`/`planned_by` vs `owner`/`committer`, `run_id`/`chat_handle`/`handle_binding`); an advisory closeout validator (run: 68 closeouts, all v1); and portal doctrine ("Ownership and Authorship", "Prove the Bottleneck First"). **Live proof of need:** this lane was first mislabeled AIF-047 and collided on-branch with a *parallel* Cowork session's AIF-047 (HELP UX) — caught by `git log`, renumbered 047→048→050 as three parallel Cowork sessions each claimed the number in turn; the exact failure the lane removes, observed while writing it. chat_handle for every seeded run is `MAINTAINER_ATTESTED` (platform stamps the session id `not_exposed`). Dev-only, candidate, not promoted. | AIPR-20260725-001 |
AIF-052 |
AI-BBS agent-server lane (M1/M4) built + runtime-observed | runtime-observed | AIPR-20260725-001 |
AIF-053 |
NET egress-security + M3 Argon2id token crypto built + runtime-observed | runtime-observed | AIPR-20260725-001 |
AIF-054 |
AI-BBS M6 standalone daemon + The Lounge room | runtime-observed | AIPR-20260725-001 |
AIF-055 |
AI-BBS guest leave-a-message (member.guest + Guestbook) | runtime-observed | AIPR-20260725-001 |
AIF-056 |
AI Portal onboarding hardening: engineering-standards / definition-of-done seed | source-defined | AIPR-20260725-001 |
AIF-057 |
AI-BBS worklog/handoff simplex drop-point (design note) | source-defined (small end built) | AIPR-20260725-001 |
AIF-058 |
AI roles taxonomy: agents vs Ollama vs GPTbase | source-defined | AIPR-20260725-001 |
AIF-059 |
Git-lock "Hot Potato" commit-coordination lane (design note) | design-intended -- not started | AIPR-20260725-001 |
AIF-060 |
Agency model + reports: name the capacity/accountability frame the system already implements | source-defined | AIPR-20260725-001 |
AIF-061 |
Memo/WAL atomicity: close the last store; correct the stale 'stubs' comment on a shipped WAL | design-intended | AIPR-20260725-001 |
AIF-062 |
Evidence layer versioning: blanket `*.log` was hiding the entire proof registry | runtime-observed | AIPR-20260725-001 |
AIF-069 |
EXPORT SDF fixed-width output, Codex 2026-07-27 | closed_runtime_observed | AIPR-20260727-001 |
AIF-074 |
SQLSEL product lane -- gold-standard SQL over the Open Relationship Platform | **CHARTERED -- plan of record under review; no engine source changed; zero runtime evidence.** Maintainer-assigned lane run through Portal onboarding (task packet first). Findings: the early SQL group is abandoned pre-engine-maturity work published as `supported` (SQ-01: `SQLERASE` discards its FROM table and deletes from the current area, 0 corpus users; SQ-04: corpus scripts call `SQLSEL SELECT ... FROM ...`, an interface never built -- the parser greps clean of SELECT/FROM); `SHOW` (199 uses) and `sqlnorm` (5 external consumers) are live and excluded from the group. **Eleven owner rulings** (R1-R11, ledger in the lane doc): start-new-keep-SQLSEL; SQL mode with modal SELECT alias (`SET MODE SQL\ | AIPR-20260729-001 |
AIF-078 |
Workspace qualifier namespace depth: decide the qualifier's depth BEFORE AIF-074 P4.1 authors it | review_needed -- design only, NO engine source changed, zero runtime evidence | AIPR-20260730-001 |
AIF-086 |
AI Systems Integration SDLC, Codex 2026-08-03 | M0 accepted; M1 owner-review candidate; local visibility prototype runtime-observed; no publication | AIPR-20260803-003 |
| Lane | What it is | Record |
|---|---|---|
AIF-001 |
AI Baby Bootstrap Card | docs/agents/AI_BABY_BOOTSTRAP_CARD.md |
AIF-002 |
Contract lifecycle chat-to-contract rule | docs/contracts/CONTRACT_LIFECYCLE_V1.md |
AIF-003 |
LabTalk SelfDoc first lab | /code/ccode/labtalk/labs/self_documenting_systems/LAB_SELFDOC_COMMENTS_TO_CONTRACTS_v0.md |
AIF-004 |
AI Friendly lane proposal | docs/ai-friendly/AI_FRIENDLY_LANE_MANIFEST_V1.md |
AIF-005 |
New-AI assimilation portal/book | docs/ai-friendly/AI_ASSIMILATION_PORTAL_V1.md |
AIF-006 |
Claude session 2026-07-13 (hosted, no repo access); re-evaluated 2026-07-14 with repo access | docs/agents/CURRENT_TARGET.md |
AIF-007 |
Same session; withdrawn proposal | AI_PORTAL.md |
AIF-008 |
MCC databuild session 2026-07-14 | docs/maintenance/SESSION_CLOSEOUT_MCC_DATABUILD_2026-07-14.md |
AIF-009 |
MCC databuild session 2026-07-14 | labtalk/ai_portal/LOCAL_ACCESS_AGENT_CHECKLIST_V1.md |
AIF-010 |
MCC databuild session 2026-07-14 | AI_README.md |
AIF-011 |
MCC databuild session 2026-07-14 | doc missing SESSION_CLOSEOUT_MCC_DATABUILD_2026-07-14.md |
AIF-012 |
Maintainer resume, 2026-07-14 | labtalk/LABTALK_DEVELOPER_PROFILE_v0.md |
AIF-013 |
Maintainer unique-system discussion, 2026-07-14 | labtalk/docs/co-development/recursive_coproject_model_v1.md |
AIF-014 |
Maintainer build continuation, 2026-07-14 | docs/maintenance/XBASE_OPTIONAL_INDEX_ARCHITECTURE_DECISION_V1.md |
AIF-015 |
Maintainer product/build separation request, 2026-07-14 | docs/maintenance/X64BASE_ENGINE_EDITION_SEPARATION_PLAN_V1.md |
AIF-016 |
Cold-clone journey certification, 2026-07-15 | docs/maintenance/SESSION_CLOSEOUT_CLONE_JOURNEY_CERTIFICATION_2026-07-15.md |
AIF-017 |
Pinocchio stress-test lane, 2026-07-15 | docs/maintenance/PINOCCHIO_STRESS_TEST_PLAN_V1.md |
AIF-018 |
Messaging Normalization lane, 2026-07-15 | docs/maintenance/MESSAGING_NORMALIZATION_LANE_PLAN_V1.md |
AIF-019 |
Build-editions review, 2026-07-14; renumbered during development reconciliation 2026-07-15 | BUILDING.md |
AIF-020 |
Codex task `019f627a-4bbd-7b53-9d02-18cab01f91fb`, 2026-07-15 | labtalk/ai_portal/AI_REPORT_AUDIT_CONTRACT_V1.md |
AIF-021 |
Claude Cowork report `AIPR-20260716-001`; Codex corrective audit `AIPR-20260716-002`, 2026-07-16 | docs/maintenance/SESSION_CLOSEOUT_CLAUDE_MESSAGING_CORRECTIVE_AUDIT_2026-07-16.md |
AIF-022 |
Claude audit response + Codex two-path proof, 2026-07-16 | docs/maintenance/DOTSCRIPT_COMMENT_PREFIX_EXECUTION_PATH_DRIFT_V1.md |
AIF-023 |
Table-buffer WAL/durability session `AIPR-20260719-003`, 2026-07-19 | docs/maintenance/SESSION_CLOSEOUT_TABLE_BUFFER_WAL_2026-07-19.md |
AIF-024 |
Maintainer directive, 2026-07-19 | AI_PORTAL.md |
AIF-025 |
Website catalog-drift session `AIPR-20260719-004`, 2026-07-19 | docs/maintenance/SESSION_CLOSEOUT_WEBSITE_SOURCE_DERIVED_CATALOG_HARVEST_2026-07-19.md |
AIF-026 |
Engine limits session `AIPR-20260719-005`, 2026-07-19 | docs/maintenance/SESSION_CLOSEOUT_ENGINE_LIMITS_AND_CORRECTIONS_2026-07-19.md |
AIF-027 |
Engine limits session `AIPR-20260719-005` follow-ups, 2026-07-19 | docs/maintenance/RECNO64_END_TO_END_64BIT_ADDRESSING_LANE_V1.md |
AIF-028 |
RECNO64 M3 buffer-proof run, 2026-07-19 | src/cli/cmd_replace.cpp |
AIF-029 |
Maintainer design discussion during the RECNO64/limits documentation, 2026-07-20 | docs/maintenance/RECNO64_END_TO_END_64BIT_ADDRESSING_LANE_V1.md |
AIF-030 |
Field-types investigation, 2026-07-20 | docs/maintenance/FIELD_TYPE_CODEC_LANE_V1.md |
AIF-031 |
FIELDTYPE M1 shakedown, 2026-07-19 | src/cli/cmd_replace.cpp |
AIF-032 |
Website reality audit — diagram drift, 2026-07-20 | docs/maintenance/FULLSTACK_GENERATED_DIAGRAMS_LANE_V1.md |
AIF-033 |
Website reality audit — no assembly/maintenance stream, 2026-07-20 | docs/maintenance/WEBSITE_CONTENT_MANIFEST_AND_ASSEMBLY_LANE_V1.md |
AIF-034 |
Doc/SDLC model brainstorm, 2026-07-20 | no record on disk |
AIF-035 |
MANUAL-ASSEMBLY lane, 2026-07-20 | docs/maintenance/MANUAL_ASSEMBLY_LANE_V1.md |
AIF-036 |
DotScript stop_on_error, 2026-07-20 | docs/maintenance/DOTSCRIPT_STOP_ON_ERROR_LANE_V1.md |
AIF-037 |
Representative-by-design codex, 2026-07-20 | AI_PORTAL.md |
AIF-038 |
DotScript arrays lane, 2026-07-20 | docs/maintenance/DOTSCRIPT_ARRAYS_LANE_V1.md |
AIF-039 |
PDLC student/working model, 2026-07-20 | docs/maintenance/PDLC_STUDENT_WORKING_MODEL_LANE_V1.md |
AIF-040 |
Projects/Lanes/Promotion tiering, 2026-07-20 | AI_PORTAL.md |
AIF-041 |
BETA-1 stabilization & regression validation lane, 2026-07-21 | docs/maintenance/BETA1_STABILIZATION_REGRESSION_LANE_V1.md |
AIF-042 |
Script header contract (`@script.usage`) lane, 2026-07-21 | docs/maintenance/SCRIPT_HEADER_CONTRACT_LANE_V1.md |
AIF-044 |
Build vectors — generated build-capacity authority, 2026-07-21 | docs/maintenance/BUILD_VECTORS_LANE_V1.md |
AIF-047 |
HELP command UX — not-found feedback + did-you-mean + exhaustive GIANT ALL, 2026-07-22 | docs/maintenance/HELP_COMMAND_UX_LANE_V1.md |
AIF-048 |
Full-stack documentation flush II, 2026-07-22 | docs/maintenance/lanes/full_stack_documentation/runs/DOCFLUSH-20260722-001/FULLSTACK_DOCUMENTATION_FLUSH_COMPLETE_HANDOFF_V1.md |
AIF-049 |
ArcticTalk retro TUI workbench lane, 2026-07-22 | docs/maintenance/ARCTICTALK_RETRO_TUI_WORKBENCH_LANE_V1.md |
AIF-051 |
HELP DATA phantom/mis-keyed topic keys, 2026-07-23 | docs/maintenance/HELP_COMMAND_UX_LANE_V1.md |
AIF-063 |
DDL schema PDLC and JSON/DBF parity | docs/maintenance/DDL_SCHEMA_PDLC_LANE_V1.md |
AIF-064 |
Retro VM/emulator launchpad lane | docs/maintenance/RETRO_LANE_CHARTER_20260726.md |
AIF-065 |
LMDB mapsize override: BUILDLMDB's documented size ladder is not honoured | src/cli/cmd_buildlmdb.cpp |
AIF-066 |
Locale spine survey: messaging mature, HELP preview fixture stranded, manuals greenfield | src/cli/cmdhelp.cpp |
AIF-067 |
Subcommand identity: `@dottalk.subusage` contract, generated SYSSUBCMD, derived `SET USAGE` | docs/maintenance/SUBCOMMAND_IDENTITY_CONTRACT_LANE_V1.md |
AIF-068 |
Manualgen harvest feeder: the DOCFLUSH-20260722 feeder lane, reactivated 2026-07-31 after its implementation entered version control | doc missing ASCENT_INPUT_DOCFLUSH-20260722_V1.md · claim |
AIF-070 |
xAI Grok external change package `AIPR-20260728-GROK-002` (maintainer design discussion 2026-07-27/28) | docs/maintenance/external_ai_intake/virtual_workspaces_memo_resident_2026-07-28/MANIFEST.md |
AIF-071 |
AI-portal report discoverability + audit-green + self-enforcing gate | labtalk/registries/ai_report_index.yaml |
AIF-072 |
Phase 7 manual web-ascent pick-up (DOCFLUSH full-stack push) | PROMOTION_CHECKLIST.md |
AIF-073 |
External Agent Memory & Continuity -- GPTbase (ChatGPT) design proposal, 2026-07-28 | doc missing AGENT_MEMORY_MODEL_V1.md · claim |
AIF-075 |
Interactive BBS auth/provenance fix: attribute + RBAC-gate shell POST/REPLY | doc missing PLAN_bbs_shell_provenance_fix.md · claim |
AIF-076 |
De-fuse BBS persistence substrate from pseudo-chat / CHAT lanes (doc-only) | doc missing DESIGN_bbs_pseudochat_two_lanes.md · claim |
AIF-077 |
Housekeeping: land Codex's abandoned uncommitted engine WIP (Codex out of credit) | doc missing cmd_evaldiff.cpp · claim |
AIF-079 |
Declared-but-unreferenced capability detection (validator tier); opened from a house-index investigation | docs/maintenance/DECLARED_CAPABILITY_VALIDATOR_LANE_V1.md |
AIF-080 |
Index container/engine orthogonality: separate container FORMAT from maintenance ENGINE so both are selectable | docs/maintenance/INDEX_BACKEND_ORTHOGONALITY_LANE_V1.md |
AIF-081 |
Output capture completeness: `DOTSCRIPT ... OUT` discards the entire user-facing command surface, found by dogfooding the engine's own transcript facility | docs/maintenance/OUTPUT_CAPTURE_COMPLETENESS_LANE_V1.md |
AIF-082 |
Onboarding cost and acceptance: measure what a cold start costs, give the portal a stopping rule and a falsifiable acceptance test | docs/maintenance/ONBOARDING_COST_AND_ACCEPTANCE_LANE_V1.md |
AIF-083 |
BBS agency legs: finish the AIF-075 sweep on the verbs it did not name | docs/maintenance/BBS_AGENCY_LEGS_LANE_V1.md |
AIF-084 |
Worktree lane isolation: stop every session sharing one working tree | docs/maintenance/AI_WORKTREE_LANE_ISOLATION_LANE_V1.md |
AIF-085 |
Cross-platform tooling rule + ported workflow tools, tested; Cowork 2026-08-02 | docs/maintenance/SESSION_CLOSEOUT_CROSS_PLATFORM_TOOLING_2026-08-02.md |
AIF-087 |
Triggers PDLC feature; hosted partner Grok/xAI (remote, hosted_proposal) | docs/maintenance/TRIGGERS_PHASE0_DECISIONS_SIGNOFF_V1.md |
AIF-088 |
Command catalog/runtime drift (R-APPEND-BLANK), caught by CRUD `--emit --ram` fsram dry run; Cowork 2026-08-04 | docs/maintenance/COMMAND_CATALOG_RUNTIME_DRIFT_PDLC_LANE_V1.md |
current_by_lane[LANE] -> runs[run_id] -> chat_handle.
Every run here is MAINTAINER_ATTESTED: the platform stamps the session id
not_exposed, so the closeout is the recovery path, not the chat link.
That is by design -- the record lives in the repo, not in a vendor's session store.AIPR-20260722-004
closed
Cowork
member.derald · committer member.derald
· planned by member.ai.chatgptproject.x64base.identity
· started 2026-07-22
· handle MAINTAINER_ATTESTEDAIPR-20260722-005
closed
Cowork
member.derald · committer member.derald
· planned by member.ai.chatgptproject.x64base.identity
· branch homegrown-cnx-20251112-branch
· started 2026-07-22
· handle MAINTAINER_ATTESTEDAIPR-20260722-007
active
Cowork
member.derald · committer member.derald
project.x64base.runtime
· branch homegrown-cnx-20251112-branch
· started 2026-07-22
· handle MAINTAINER_ATTESTEDAIPR-20260725-001
active
Cowork
member.derald · committer member.derald
project.ai_friendly
· branch development
· started 2026-07-25
· handle MAINTAINER_ATTESTEDAIPR-20260727-001
closed
Codex
member.derald · committer member.derald
· planned by member.deraldproject.x64base.runtime
· branch development
· started 2026-07-27
· handle NOT_RESOLVABLEAIPR-20260729-001
active
Cowork
member.derald · committer member.derald
· planned by member.deraldproject.x64base.runtime
· started 2026-07-29
· handle MAINTAINER_ATTESTEDAIPR-20260730-001
active
Cowork
member.derald · committer member.derald
· planned by member.deraldproject.x64base.runtime
· branch development
· started 2026-07-30
· handle MAINTAINER_ATTESTEDAIPR-20260803-003
closed
Codex
member.derald · committer member.derald
· planned by member.deraldproject.ai_systems.integration
· branch development
· started 2026-08-03
· handle SELF_REPORTED| Task | Lane | Channel | Status | Assignee |
|---|---|---|---|---|
task.command_catalog.append_blank_driftCommand catalog vs runtime drift reconciliation (R-APPEND-BLANK) |
AIF-088 |
ai_portal | open | member.derald |
task.docs.fullstack.flush.20260722Full-stack documentation flush II |
AIF-048 |
ai_portal | open | member.derald |
task.metadata.metacollect.238METACOLLECT 238-finding metadata mission |
METACOLLECT-238 |
ai_portal | open | member.derald |
task.retro.launchpad_laneRetro VM/emulator launchpad lane |
AIF-064 |
ai_portal | open | member.derald |
task.scan.bulk_record_ioBulk record-I/O scan follow-on |
AIF-046 |
ai_portal | open | member.derald |
task.script.header.contractScript header and usage-contract lane |
AIF-042 |
ai_portal | open | member.derald |
task.x32.traditional_xbase_supportx32 traditional xBase support feasibility lane |
AIF-068 |
ai_portal | open | member.derald |
task.arctictalk.workbenchArcticTalk retro TUI workbench |
AIF-049 |
ai_portal | in progress | member.derald |
task.beta1.stabilizationBETA-1 stabilization and regression validation |
AIF-041 |
ai_portal | in progress | member.derald |
task.ddl.schema_pdlcDDL schema PDLC and JSON/DBF parity lane |
AIF-063 |
ai_portal | in progress | member.derald |
task.historical_database_migration.empirical_proHistorical database migration empirical progress |
AIF-058 |
ai_portal | in progress | member.derald |
task.identity.rbac.m4_m6Identity, authentication, RBAC, and authorization follow-through |
AIF-045 |
ai_portal | in progress | member.derald |
task.export.sdf_fixed_widthEXPORT SDF fixed-width output |
AIF-069 |
ai_portal | done | member.derald |
task.build.palette_stub_parking.v1Palette generated-stub disposition |
PSEUDO-Q4 |
pseudo_chat | returned | member.derald |
| Proof | State | Evidence note |
|---|---|---|
proof.agency.modelAgency model -- capacity to act plus accountability for having acted |
source_defined | Names a model the system had already IMPLEMENTED but never stated, so the mechanisms stop reading as unrelated engineering choices. Four legs: identity (SYSMEMBER), authority (SYSMEMROLE->SYSROLEPERM->SYSPERM), authentication (Arg... |
proof.ai.roles_taxonomyAI roles taxonomy - agents vs Ollama vs GPTbase |
source_defined | Names three non-interchangeable roles - DOERS (agent-members that authenticate + change source + leave handoffs), LOCAL BRAIN (Ollama, isolated inference, not a member, needs a harness), HOSTED ADVISOR (GPTbase, advisory not autho... |
proof.ai_friendly.assimilation_docs_seedAI Friendly assimilation portal and book seeded |
source_defined | Provides repo-local onboarding for future AI or second-opinion AI if hosted chat/provider context is unavailable. |
proof.ai_friendly.maint_ai_assimilate.first_runMAINT AI ASSIMILATE first portal/runtime transcript |
runtime_observed | Proves LabTalk portal can run the DotTalk++ MAINT AI ASSIMILATE read-only command and capture the onboarding readback. |
proof.ai_portal.discoverability_caught_its_own_authorThe discoverability fix caught its own maker -- Phase 7 was the next dropped thread |
source_defined | IRONY, RECORDED. On 2026-07-28 the AI Portal's external-intake discoverability gap (AIF-071) was diagnosed and fixed: work that lives only in ephemeral memory -- a hosted chat, an assistant task list -- cannot be found later, so i... |
proof.ai_portal.dotscript_startup_readiness.v1DotTalk++ / DotScript startup readiness probe v1 |
runtime_observed | Proves the current compiled runtime exposes the mandatory read-only learning surfaces and that the portal's transcript acceptance gate accepts the green path. ORIENTED permits drafting and review only; it does not authorize a muta... |
proof.ai_portal.engineering_standards_seedAI Portal engineering-standards / definition-of-done seed on the mandatory path |
source_defined | Front-loads @dottalk.usage contracts, regression doctrine, the lane close-out checklist, and house conventions; wired into AI_README.md step 6 + AI_PORTAL.md + ROOT entry + fast-start closeout. Source_defined by nature; VERIFICATI... |
proof.ai_portal.lane_registeredAI Portal Alpha/Experimental lane registered |
source_defined | Records the governed Alpha/Experimental status, APH-0 through APH-6 gates, authority boundary, and requirement that the portal not be presented as production autonomous memory. |
proof.ai_portal.root_entry_and_external_package.v1Root AI Portal entry and external change-package contract defined |
source_defined | Provides one repository-root fast-start link for outside AI engines, mandatory SDLC entry, and a required patch-package format tied to an exact public baseline commit. Source-defined only until promoted and verified in the public ... |
proof.ai_portal.sdlc_fast_start.v1AI Portal SDLC fast-start seed registered |
source_defined | Makes owning lifecycle, SDLC lane, truth state, proof state, risk, next gate, and closeout mandatory across AI Portal tasks and packages. PLDC and publication cannot bypass underlying SDLC evidence. |
proof.ai_portal.seed_connection_prototype_direction.v1AI fast-start seed connection prototype direction registered |
design_intended | Records that the AI Portal is a machine-facing fast-start system for AI development partners, not a student AI-access portal, and that the first connector prototype remains read-only and deterministic. |
proof.ai_portal.source_mutation_contract_gate.v1Source mutation contract preflight registered |
source_defined | Records the maintainer rule that AI source-code mutations require contract-shelf, lifecycle, subsystem-contract, and source-annotation review before a patch is applied. This record does not grant mutation authority. |
proof.aif078.slot_cost_measured_per_toolchainMAX_AREA per-slot cost measured on both supported toolchains; the two axes bind on different platforms |
runtime_observed | AIF-078 gate G0. MSVC (_MSC_VER=1944): DbArea 840, AreaState 144, fstream 280, 1016 B/slot. g++/libstdc++ x86-64: DbArea 1088, AreaState 176, fstream 528, 1296 B/slot. sizeof(XBaseEngine) is 4104 on both because it is only MAX_ARE... |
proof.aif078.workspace_path_preserves_depth1DataAddress workspace scalar widened to a path with no depth-1 behavior change |
runtime_observed | AIF-078 Q7. WorkspaceIdentity workspace_ became WorkspacePath workspace_path_ (outermost first); the scalar constructor kept its signature, delegates, and yields depth 1. Asserted rather than inspected: depth-1 renders exactly "MC... |
proof.bbs.guestAI-BBS guest leave-a-message (member.guest + board.guestbook) |
runtime_observed | As member.guest -> POST board.guestbook OK posted |
proof.bbs.m1_boardAI-BBS M1 board tables + BBS command |
runtime_observed | BBS BOARDS created SYSBOARD/THREAD/POST.dbf; POST/READ round-trip; board.governance projects SYSGRANT. |
proof.bbs.m2_net_egressAI-BBS M2 NET EGRESS permissioned toggle |
runtime_observed | AI member refused NET EGRESS OPEN ("no in-scope role permission"); owner STATUS = Block. host.network.egress owner-only. |
proof.bbs.m3_argon2AI-BBS M3 Argon2id token crypto (libsodium) |
runtime_observed | USER TOKEN -> 43-char base64url CSPRNG; correct token logs in, wrong token rejected. Gates M4. |
proof.bbs.m4_serveAI-BBS M4 BBS SERVE loopback listener + Ollama bridge |
runtime_observed | 127.0.0.1:8765 LISTENING (loopback only); AUTH OK; CHAT haiku from isolated Ollama while egress=Block; OK posted. |
proof.bbs.m6_daemonAI-BBS M6 standalone dottalk_bbsd daemon |
runtime_observed | Own binary; headless via DotTalkBBSD task (SYSTEM/session 0, no window); AUTH/CHAT/POST across restarts; SO_EXCLUSIVEADDRUSE; SIGPIPE guard. |
proof.bbs.worklogAI-BBS worklog/handoff board + simplex cascade guard |
runtime_observed | 2026-07-25 built + observed. board.worklog listed by BBS BOARDS (kind=2, post=bbs.post); as member.ai.grok.xai, POST board.worklog SESSION-20260725 :: RUN=... -> OK posted #6, READ back OK. Handoff-post convention (RUN/STATE/DID/O... |
proof.build.gcc_linux_warning_baselineGCC/Linux build of x64base under WSL -- warning baseline |
runtime_observed | Captured 2026-07-30 building the tree under WSL with GCC 13. 94 warning lines across 16 classes, of which MSVC reports none -- so this is the only non-MSVC build evidence the repository holds, for a project whose stated position i... |
proof.case.registry_existsCase registry exists |
case_registered | |
proof.codev.system_corrects_its_extenderThe tree's own artifacts and gates corrected an outside AI extender, repeatedly |
source_defined | Eleven recorded corrections in one lane. Owner redirects (five) converted construction proposals into consumption of existing machinery -- locks, tuple carrier, buffering, unique registry -- with catch latency falling to zero by t... |
proof.contract.dottalk_file_fulltree@dottalk.file universal contract - full-tree backfill at 100% coverage |
runtime_observed | census 1034/1034 (100.0%) observed from source_census.py after the sweep; M3 --strict now exits 0 (gate passable, promotion is a maintainer decision). Schema v1 revised: path:/provenance: dropped (duplicated git tracking; path: go... |
proof.edref.topic_existsEDREF topic exists |
source_defined | |
proof.engine.append_blank_catalog_driftRuntime APPEND rejects the BLANK token the command catalog documents (silent clobber) |
runtime_observed | AIF-086, repair R-APPEND-BLANK. The command catalog (src/cli/shell_commands.cpp -> content/docs/dottalk/command-catalog.mdx) documents APPEND_BLANK with syntax "APPEND BLANK", but the runtime APPEND parser rejects the BLANK token ... |
proof.engine.key_metadata_survives_workspace_roundtripPRIMARY/UNIQUE key declarations persist through WORKSPACE SAVE and LOAD |
runtime_observed | Phase 2 of a registry whose own header asked for it ("Phase 1: session-scoped ... storage/backfill policy belongs elsewhere"). The registry was emptied between SAVE and LOAD, so restoration is proven, not assumed: LOAD echoed KEY ... |
proof.engine.scan_limit_reports_truncationRelation scans announce when the scan limit stopped them |
runtime_observed | AIF-073 finding RDB-06 closed. A scan that stopped at the cap previously returned a plausible-looking partial answer with no indication -- a wrong result indistinguishable from a right one. At limit 1 the run reports "REL: scan li... |
proof.engine.two_read_families_buffer_visibilityTwo coherent read families; buffer-visibility defects occur only at the hybrid seam |
runtime_observed | Under TABLE BUFFER ON a REPLACE is buffered, not written to the DBF, so every reader must choose a source. Three probes established that x64base has TWO internally coherent families and that defects appear ONLY where they are mixe... |
proof.engine.typed_equality_crosses_declared_typesRelation equality compares numerically whenever both sides are numeric literals |
runtime_observed | AIF-073 finding RDB-03 converted from divergence to conformance. Previously only a numerically DECLARED child field got numeric comparison, so a character child holding "1" failed against a numeric parent's "1.00" -- an asymmetry ... |
proof.evidence.layer_versionedProof transcripts un-ignored and committed -- registry claims now verifiable from a clone |
runtime_observed | MEASURED 2026-07-25: .gitignore line 36 carried a blanket *.log written for runtime noise; it also swallowed labtalk/proofs/runs/*.log, the transcripts proofs.yaml rows cite AS THEIR EVIDENCE. 71 proof artifacts on disk, 0 tracked... |
proof.export.sdf.fixed_width_exportEXPORT SDF writes fixed-width records using the shared TUPTALK row formatter |
runtime_observed | AIF-069 closed as a runtime-observed development slice. Debug build passed and REGRESSION RUN EXPORT_SDF exported two schema-width rows with no header: "AB 7 12.3T" and "WXYZ123 -4.5F", each length 14. The generated temp SDF ... |
proof.help.cmdhelp_source_existsCMDHELP source exists |
source_defined | |
proof.lab.database_literacy_starter.first_runDatabase Literacy Starter first portal/runtime transcript |
runtime_observed | Proves LabTalk portal registry item can run a DotTalk++ DTS script and capture EDREF output. |
proof.lab.selfdoc.comments_to_contracts.first_runSelfDoc comments-to-contracts first CMDHELP/CMDHELPCHK run |
validated | Captures CMDHELP/CMDHELPCHK runtime output plus read-only comments/contracts crosswalk for CMDHELP and CMDHELPCHK. |
proof.lmdb.mapsize_ladder_honoured_after_fixBUILDLMDB TINY survives an index attach -- the size ladder is honoured |
runtime_observed | AIF-065 FIX VERIFIED. mdb_env_set_mapsize(env_, 0) replaces the hardcoded 1 GiB at cdx_backend.cpp and lmdb_backend.cpp, so the attach path ADOPTS the persisted mapsize instead of asserting one. SYSARGS rebuilt with BUILDLMDB CLEA... |
proof.lmdb.mapsize_override_on_attachLMDB env resizes 128 MiB -> 1 GiB on index auto-attach, ignoring the BUILDLMDB ladder |
runtime_observed | AIF-065 promoted from source_defined to runtime_observed. SYSSUBCMD.cdx.d/data.mdb measured at 134,217,728 bytes immediately before the run and 1,073,741,824 bytes immediately after -- exactly 8x, on a 31-ROW TABLE, from a single ... |
proof.lmdb.mapsize_override_replication_sysfuncLMDB mapsize override replicates on a second table (SYSFUNC, 69 rows) |
runtime_observed | Independent replication of proof.lmdb.mapsize_override_on_attach on a different table. SYSFUNC.cdx.d/data.mdb measured 134,217,728 before and 1,073,741,824 after a single index auto-attach at USE -- delta 939,524,096 on a 69-row t... |
proof.pdlc.cobol_fixed_record.hop_closedDBF to fixed-record to GnuCOBOL read, 200 records |
runtime_observed | COBOL EXPORT wrote 200 records; the file is 22,200 bytes = 200 x 111 EXACTLY. That division is the fidelity proof -- a fixed-record file carries no header count and no delimiters, so a short write is invisible to anything that mer... |
proof.pdlc.csv_to_x64base.type_fidelityCSV to x64base preserves types, NULLs and long names |
runtime_observed | 18 rows, 13 fields from the sealed Cascade Items.csv. All 13 type letters matched a derivation taken from src/import/import_profile.cpp BEFORE the run: N C C C C C N N N N N C N. Two behaviours predicted from the code and confirme... |
proof.pdlc.dbf_csv_dbf.roundtrip_losslessDBF to CSV to DBF round trip is lossless |
runtime_observed | Origin and return legs both 18 rows / 13 fields, with IDENTICAL structure -- same type letters, same widths, same decimal counts. EXPORT rendered every value such that re-inference reproduced the origin schema exactly, which is st... |
proof.pdlc.dialect_ladder.dbase3_to_x64dBASE III to FoxPro 2.6 to VFP to x64 to x64 VECTOR, lossless |
runtime_observed | The historical migration demonstration proper, walked inside one engine with no external tool, from a genuine legacy artifact: dbf/og/STUDENTS.DBF, version byte 0x03, 200 records, 9 fields, every field name at or under the classic... |
proof.portal.truth_audit.latestPortal truth audit latest report |
validated | Captures the portal collection-of-truths audit across LabTalk registries, local proof records, and x64base.com SelfDoc/SDLC publication pages. AUDIT SCOPE INCLUDED THREE SITE PAGES, cited here until 2026-08-02: content/docs/labtal... |
proof.product.manualgen.board_preservedManualgen product board and map preservation |
validated | Preserves the first Manualgen product board and records initial hardening evidence using repo-local vcpkg Python 3.12.9. |
proof.selfdoc.comments_workflow_existsComments to HELP workflow exists |
source_defined | |
proof.sqlsel.select_statement_matches_sqliteSQLSEL SELECT ... FROM returns row sets equal to SQLite over the same data |
runtime_observed | The lane's one genuinely new component. Projection, star, WHERE and LIMIT each returned the same rows as the in-process SQLite oracle over identical data (1|ADAMS|CSCI, 2|BAKER|MATH, 3|CARTER|CSCI; WHERE MAJOR='CSCI' -> ADAMS + CA... |
proof.tooling.cross_platformWorkflow tooling ported to cross-platform Python, each tested |
runtime_observed | Rule adopted 2026-08-02 at maintainer direction: no PowerShell-only or bash-only solutions; Python 3 + stdlib or DotScript, ASCII, shell only as a thin wrapper. The project is cross-platform C++ on cross-compatible libraries and t... |
proof.wal.dbf_recordTABLE BUFFER write-ahead log -- DBF record atomicity + crash recovery |
runtime_observed | Built and PROVEN 2026-07-19 in three teed phases. Phase A: COMMIT/ROLLBACK over the .tbj redo log with retained edits preserved (three edits to one field kept as seq 1/2/3, COMMIT landing the highest-priority Edit3). Phase B: CRAS... |
proof.wal.memo_atomicityWhole-row atomicity including the memo store |
design-intended | GAP, not yet built. An x64 memo REPLACE converts text to a stored object-id and the DBF field holds the id, so the WAL journals the id and never the content. A crash between the durable WAL commit and the memo store's own write le... |
proof.worktree.lane_isolationWorktree-per-lane isolation -- sessions stop sharing one working tree |
design-intended | MEASURED 2026-08-02: a Cowork session worked several hours against development and found 191 commits had landed from parallel work while it held a stale picture; three of four queued operations were already done by someone else, a... |