Skip to content

Latest commit

 

History

History
51 lines (43 loc) · 10.7 KB

File metadata and controls

51 lines (43 loc) · 10.7 KB

Feature register

Canonical feature status as of 2026-08-23. Status is evidence-based. A proposal saying “implemented” is insufficient without source/test evidence; conflicting claims use VERIFY.

Active and future features

Feature Status Milestone Primary specification Depends on Current truth / next outcome
ACCOUNT-001 Multi-account core VERIFY M1 Roadmap 01 Account/group and targeted generation foundations plus local snapshots ACCOUNT-101/ACCOUNT-102 exist; real two-same-network browser/limits/queue/WorldState acceptance remains open.
ACCOUNT-002 Per-account settings VERIFY M1 Roadmap 02 ACCOUNT-001 Resolver provenance, active-switch synchronization, /accounts editor and local backend/UI tests are present; staging/manual operator acceptance remains open.
PERSONA-001 Editorial personas and account voice IN_PROGRESS M1–M2 Roadmap 08, supporting account-voice spec, ADR-008 ACCOUNT-001, ACCOUNT-002, EVAL-001 ADR-008 is accepted; PERSONA-101/102 and PERSONA-103 portfolio dispatch/provenance plus persona assignment UI are locally verified, while held-out evaluation and external acceptance remain.
ENGAGE-001 Reviewed conversational engagement IN_PROGRESS M2–M3 Roadmap 08, ADR-008 PERSONA-001, EVAL-001 ADR-008 is accepted; ENGAGE-101 scorer/suggestion/policy seams, loop wiring and operator review queue are locally verified, while manual Threads/X review, native execution and EVAL-505-calibrated pilot remain.
MEDIA-001 AI image generation VERIFY M4 Roadmap 03 ACCOUNT-002, PERSONA-001 MEDIA-101 quota-safe IImagePort/Post.media foundation, official REST adapter and local poster/UI fallback are verified; provider credentials, HITL preview and native/staging evidence remain.
COST-001 LLM cost optimization IN_PROGRESS M5 Roadmap 05 EVAL-001 Exact/shared cache, scoped token budgets, atomic rejected-reservation handling, structured telemetry, durable LlmUsageEvent accounting, optional per-account daily reservation, prompt compression, opt-in price ordering and local cost dashboard are source/unit verified; hook-cache Redis runtime, semantic cache, degradation actions and external billing evidence remain.
REL-001 Self-healing and degradation VERIFY M1–M3 Roadmap 06 ACCOUNT-001 Resilience integrations, probes, health endpoint and Monitor panel are locally verified in REL-102-local-evidence-2026-08-23.md; real failure/recovery and staging evidence remain open.
EVAL-001 AI change release gate and agent evaluation VERIFY M0–M1 Roadmap 09, evaluation hub, ADR-009 Trace roots/propagation, evaluation contracts, side-effect-free candidate execution, local dataset manifest/stratification validation, provider-attempt telemetry, native prompt linkage, telemetry/redaction self-test, deterministic/online evaluators, static CI, durable review truth, calibration metrics and local feedback reconciliation are source/test verified; reviewed curation, hosted dataset/judge, real sync, human calibration and hosted experiments remain open.
INTEL-001 Conversation intelligence and demand radar IN_PROGRESS M2–M5 Roadmap 10, ADR-010 ENGAGE-001, EVAL-001 ADR-010 is accepted; INTEL-101 privacy/signal/deterministic-extractor/cluster/aggregate-insight foundation and operator UI are locally verified, while richer clustering, ambiguity evaluation, live source and EVAL-203 gates remain.
ATTR-001 Direct link attribution funnel VERIFY M0–M2 ADR-007, planning/ROADMAP.md §Z4 M2.1–M2.4 backend contract and conversion dashboard are locally source/test/build verified in archived ATTR-102/ATTR-103 and ATTR-104-local-evidence-2026-08-23.md; deployment, live non-zero funnel and revenue evidence remain.
ATTR-002 Reply-to-revenue assisted attribution SPEC_READY M2–M5 Roadmap 11, ADR-011 ATTR-001, ENGAGE-001 ADR-011 is accepted for aggregate direct/assisted reporting; incremental estimates remain HOLD until sufficient evidence.
POLICY-001 Platform policy and reputation control plane IN_PROGRESS M0–M3 Roadmap 12, ADR-012 REL-001, EVAL-001 ADR-012 is accepted; POLICY-101 registry/compiler/UI and POLICY-102 reputation/scoped recovery/safety UI are locally verified, while primary-source review, calibration, staging soak and live evidence remain.
CRM-001 Creator relationship and collaboration CRM IN_PROGRESS M4–M5 Roadmap 13, ADR-013 ENGAGE-001, INTEL-001, POLICY-001 ADR-013 is accepted for public-data/manual human-controlled CRM; CRM-101 public identity, relationship, evidence, cooldown-aware recommendation, human-only proposal foundation, explicit cross-network identity link/unlink, safe interaction wiring and operator signal board/timeline are locally verified, while privacy evaluation, real DB and manual outreach evidence remain.
BRIDGE-001 Soulwise editorial data bridge SPEC_READY M2–M4 Roadmap 14, ADR-014 EVAL-001, GROUND-001 ADR-014 is accepted for read-only PUBLIC_FACT v1; cross-repository owner, contract, deployment and rollback remain.
SYND-001 End-to-end article syndication VERIFY M1 planning/ROADMAP.md M1.4 SYND-100 real canonical verifier plus locally verified article draft→poster→visibility→canonical fail-closed slice are in HEAD/worktree evidence (packages/backend/artifacts/evaluation/SYND-101-local-evidence-2026-08-23.md); live Dev.to account, external canonical HTML, deployment, approval and rollback evidence remain.
ORCH-001 Orchestrator GA IN_PROGRESS M3 .forge master plan REL-001, EVAL-001 Archived ORCH-101 reconciles frozen boxes as historical; current GA/dual-path/watchdog/soak evidence remains ORCH-102/ORCH-103 after REL-102 and EVAL-505.
NETWORK-001 Additional network validation IN_PROGRESS M4 README.md platform matrix, Roadmap 15 REL-001 X/Threads are current validated focus; 2026-08-23 transport decision applies the standing rule "free official API → API, else stealth browser": Bluesky/Mastodon go API-first (AT Protocol / Mastodon API), Facebook/LinkedIn stay browser-only.
HARDEN-001 Codebase hardening and unification IN_PROGRESS M1 (cross-cutting) Execution roadmap — hardening track CI coverage/UI gates, prod Redis eviction fix, god-class decomposition (generation.service, posting.service, x.poster), unified .js import convention, domain/network-profiles single source of platform knowledge, dead multilingual-config removal, design-system primitive completion. Tracked as CI-1xx/REFACTOR-1xx/DESIGN-1xx/DOCS-1xx.
CONTROL-001 Operator control bot (Telegram) IN_PROGRESS M4–M5 Roadmap 16 REFACTOR-105 Local HITL loop from chat is implemented and source/unit verified: status, push alerts, approve/reject drafts, pause/resume flow-control and chat_id allowlist. Real Telegram round-trip and alert-drill evidence remain.
DIST-001 Multi-instance distribution VERIFY Later Feature dossier REL-001 Read-only reconciliation artifact packages/backend/artifacts/evaluation/DIST-101-reconciliation-2026-08-23.md confirms limited shared-single-Redis support but watchdog ownership, fencing, two-process/real-Redis and browser-affinity evidence remain open behind REL-102.
BROWSER-001 Browser E2E replay harness VERIFY Cross-cutting Feature dossier Implementation tasks BROWSER-101/BROWSER-102 are archived DONE; redacted fixture, selector drift and evidence classification are source/test verified. Browser launch, live dry-run/replay and platform acceptance remain unverified, hence VERIFY (not PLANNED).
GROUND-001 Fact-checking and reviewed memory IN_PROGRESS M4–M5 Roadmap 08 PERSONA-001, EVAL-001 GROUND-101 evidence/memory lifecycle, lexical ports, admin-guarded review queues, claim-gate integration and grounding UI are locally verified; contradiction/held-out eval, privacy deletion and hosted/vector gates remain.

Completed platform features

Completed task evidence lives in archive/2026-Q3.md.

Feature Status Evidence dossier Remaining work belongs to
PLATFORM-001 A/B variant infrastructure DONE A/B testing Judge calibration → EVAL-001
PLATFORM-002 Content adapters DONE Content adapters New adapters get new tasks
PLATFORM-003 Operator dashboard foundation DONE Operator dashboard Evaluation panels → EVAL-001
PLATFORM-004 Langfuse prompt versioning/diff DONE Prompt versioning Native prompt trace linkage → EVAL-001

Feature status rules

  • IN_PROGRESS reflects an active owned workstream, not percentage complete.
  • VERIFY is deliberately used where docs and source disagree or a referenced decision/evidence artifact is missing.
  • Optional expansion does not reopen a completed feature; create a new feature/task.
  • Roadmap milestones derive from this table, not the other way around.