Skip to content

ROADMAP.md

Latest commit

 

History

History
96 lines (88 loc) · 15.3 KB

ROADMAP.md

File metadata and controls

96 lines (88 loc) · 15.3 KB
Raw

Codex SDLC Wizard Roadmap

How to use this roadmap

  • GitHub issues are the source of truth for scope, acceptance criteria, milestone membership, and status.
  • ROADMAP.md is this repository's single ordered project backlog and cold-session resume contract.
  • The numbered order below is the project priority. The first open item is next unless the maintainer explicitly reorders the queue.
  • A GitHub milestone is a planned release; its open issues are the work required before that release can close.
  • An optional generated GOALS.md may scope execution inside a consumer repo, but it never overrides this project's priority queue.
  • Remove completed issues from the queue, add every new actionable issue once, and keep every open issue assigned to a milestone.
  • Pull requests reference their owning issue. Failing proof, review, or required CI never merges.

Cold-session checkpoint — resume here

  • The active engineering slice is #175 in /Users/stefanayala/codex-sdlc-issue175 on fix/issue-175-final-fable-response, based on current origin/main at b7c8d4590a483afa0be61f1bb9ed4408d3b520cd.
  • Tell tool-free Fable to return one immediate final schema-valid JSON review from the complete supplied evidence; the runtime guarantee remains strict rejection and a non-certifying receipt for deferral, narration, or any other invalid output.
  • #157 is integrated through PR #166, #158 through PR #165, #167 through PR #170, and #172 through PR #173. #158's one-time meta-review owner exception is consumed; do not reuse or broaden it. Required pre/post-merge Windows CI passed for the integrated slices.
  • #174 is frozen at candidate tree 2636c7da25fcd9a19735112f938e050981012a6c. Canonical proof passed 11/11, but Fable returned only repository-inspection deferral prose; the strict gate wrote a non-certifying invalid_execution receipt, cancelled incomplete Sol, and recommends no retry.
  • #86 is frozen at final candidate tree 7cfa28ec0add78c706cd24f28b15a9e9e99d023b. Canonical proof passed 11/11, but normal Sol High plus Fable High review converged on the #174 P1, so its receipt is non-certifying and the candidate must not be delivered or retried.
  • Do not edit, stage, clean, or merge the dirty root checkout. It contains unrelated historical work.
  • After #175 integrates, resume #174 from a fresh current-main worktree and fresh proof/review candidate. Then resume #86, followed by a fresh authorized plan for frozen #109 v4; do not reuse consumed gate attempts or obsolete plans.
  • Continue through the complete 1.0.0 block below. Do not start post-1.0 work while an earlier release blocker remains open.

Current State

  • Latest published GitHub release: v0.7.37.
  • Latest published npm release: codex-sdlc-wizard@0.7.37.
  • Development package metadata: 0.7.38; it is not a published release.
  • Next major release: 1.0.0 — Bounded autonomous delivery.
  • The synchronized 1.0 milestone contains ten open release issues and fifteen completed issues as of 2026-08-23.
  • The release objective is bounded autonomous delivery: exact-candidate proof, terminating independent review, honest failures, one canonical workflow entry, trustworthy Windows behavior, real Windows acceptance, and immutable publish verification.
  • Real Windows Codex Desktop and CLI acceptance remains the final hardware-dependent gate before publish verification.
  • 1.0.0 is the next public release. Do not divert the release train into another 0.7.x feature batch.

Priority queue

  1. Reduce tool-free Fable deferral and fail closed — #175: instruct immediate review JSON and reject every invalid response (1.0.0, P0, active).
  2. Keep model-policy migration claims truthful — #174: leave migration pending while preserved guidance still authorizes autonomous xhigh (1.0.0, P0, frozen until #175 lands).
  3. Enforce High-only autonomous Sol work — #86: remove autonomous xhigh escalation and the former maintainer exception (1.0.0, P0, frozen until #174 lands).
  4. Complete the measured bounded-correction pilot — #109: incremental corrective commits with a final whole-PR gate (1.0.0, P0, frozen until #174 and #86 land).
  5. Block executable Git configuration at delivery — #141: reject exec-capable Git config in exact-candidate commands (1.0.0).
  6. Restore one canonical workflow entry — #127: remove duplicate global and repo-local $sdlc exposure (1.0.0).
  7. Eliminate nondeterministic Windows proof cleanup — #92: fix access-denied proof-stamp teardown (1.0.0).
  8. Keep Windows review evidence honest — #93: detect WindowsApps PowerShell before trusting review validation (1.0.0).
  9. Run the final real-Windows acceptance gate — #79: validate the exact 1.0 candidate in Codex Desktop and CLI, including observed config/picker precedence (1.0.0, human-required).
  10. Publish and verify the major release — #88: publish and verify codex-sdlc-wizard 1.0.0 across GitHub, npm, plugin, Codex Desktop, and Codex CLI surfaces (1.0.0, human-required where credentials require it).
  11. Audit host-managed memory before premium rollout — #129: compare host memory with checked-in repository truth (1.1.0).
  12. Implement the premium Codex host lane — #128: Fable High plans/reviews and Sol High builds with one reconciliation (1.1.0).
  13. Make Fable review output evidence-first — #123: tighten cross-model review output (1.1.0).
  14. Preserve known-good behavior during upgrades — #84: adopt a stability contract for model and harness changes (1.1.0).
  15. Classify self-targeting maintenance as SDLC work — #130: treat mutating setup, update, and repair commands honestly (1.1.0).
  16. Validate PowerShell policy before mutation — #153: reject invalid reviewer policy before updater writes (1.1.0).
  17. Align skill metadata with evidence lanes — #154: make headings and descriptions honest (1.1.0).
  18. Add one read-only health entrypoint — #82: implement a Codex-native doctor flow with explicit repair (1.1.0).
  19. Audit generated instruction ownership — #95: reconcile AGENTS.md with current Codex guidance (1.1.0).
  20. Make sibling harnesses coexist safely — #106: define shared-root ownership with claude-sdlc-harness (1.1.0).
  21. Teach the issue-to-PR-to-release lifecycle — #100: enforce bounded issue, pull-request, milestone, and merge state (1.1.0).
  22. Support ordered multi-milestone goals — #151: continue across ordered milestones and hand off cleanup safely (1.1.0).
  23. Make terminal lifecycle state declarative — #136: reconcile exact status after merge (1.1.0).
  24. Add an opt-in progress dashboard — #135: show issue and milestone progress without claiming readiness (1.1.0).
  25. Preserve privacy in feedback attribution — #126: make source-repository attribution opt-in (1.1.0).
  26. Soak the released contract in real consumer repos — #112: collect daily post-1.0 feedback and make only proven fixes (1.1.0).
  27. Measure adversarial refutation against bounded review — #138: compare cross-vendor review strategies (1.1.0, research).
  28. Research convergence-based termination — #132: compare convergence with fixed correction caps (1.1.0, research).
  29. Retain concise failed-proof evidence — #124: preserve failed-test summaries when broad proof output truncates (1.1.0).
  30. Separate confidence from defect severity — #131: require actionable reviewer remediation without conflating confidence (1.1.0).
  31. Make explicit updates one-step and agent-driven — #147: honor requests to update to the latest wizard (1.1.0).
  32. Make after-hours Windows pushes guard-compatible — #162: provide an explicit safe push override path (1.1.0).
  33. Emit bounded provider progress during reviews — #164: expose long-running dual-review progress (1.1.0).
  34. Harden standalone Fable supervision — #171: bound standalone Fable process trees and bind review-run identity (1.1.0).
  35. Track third-party release provenance — #168: add third-party asset provenance to release preflight (1.1.0).
  36. Research a fast incremental model lane — #114: evaluate GPT-5.3-Codex-Spark (1.2.0).
  37. Benchmark optional worker orchestration — #72: compare direct Sol driving with bounded Luna workers (1.2.0).
  38. Research context-pressure policy — #77: measure compaction thresholds and worker effects (1.2.0).
  39. Submit the proven release to the Plugins Directory — #66: complete universal directory submission (Distribution, human-required).
  40. Research recurring post-1.0 distribution — #140: evaluate agent marketplaces and directories (Distribution, research).
  41. Audit CI/CD for independent value — #108: retain only release-enforcing evidence (1.3.0).
  42. Prove fast-first without weakening full-final — #139: evaluate a bounded CI tier (1.3.0, research).
  43. Detect release drift deterministically — #99: reconcile package, tag, release, and issue state (1.3.0).
  44. Timebox the cumulative upstream audit — #65: evaluate upstream v1.74.0 through v1.91.0 (1.3.0).
  45. Reconcile upstream v1.96.0 — #113: evaluate the newer SDLC Wizard release (1.3.0).
  46. Reconcile upstream v1.98.0 — #160: translate applicable upstream changes (1.3.0).
  47. Keep release documentation impact-based — #105: add stabilized demos only when they materially help (1.4.0).
  48. Consolidate redundant proof documentation — #104: fold PROVE-IT.md into TESTING.md by default (1.4.0).
  49. Rename wizard to harness with a controlled migration — #101: apply the sibling-repo migration checklist safely (1.4.0).
  50. Clean contributor attribution only through an explicit history operation — #107: remove bot attribution without risking repository history (1.4.0, human-required).
  51. Research persistent Codex output style — #144: identify and measure the nearest supported mechanism (Research, non-blocking).
  52. Research DeepSeek Harness after 1.0 — #142: evaluate portable plugin ideas (Research, non-blocking).
  53. Track the in-app Browser range-slider defect — #134: require observable native slider interaction (Research, external-tooling, non-blocking).
  54. Map the SDLC experience across Copilot tiers — #125: compare every Copilot host (Research, non-blocking).
  55. Identify the Microsoft-enterprise coding host — #122: prove the supported enterprise surface (Research, human-required, non-blocking).
  56. Evaluate another CLI host — #97: research a GitHub Copilot CLI adapter (Research, non-blocking).
  57. Revisit portable enforcement after host adapters stabilize — #58: compare a portable SDLC MCP server with per-host skills (Research, non-blocking).
  58. Decide the Copilot Studio adoption path — #96: research a front end only after portable enforcement is justified (Research, human-required, non-blocking).
  59. Keep Copilot Studio implementation contingent — #163: build and E2E-test the agent only after #96 selects adoption or experiment (Research, contingent, non-blocking).
  60. Release the post-1.0 Copilot adapter — #169: build and release the GitHub Copilot SDLC harness after 1.0 (Research, post-1.0, non-blocking).