# CLARIFY — v1.27 PO State Catalog & Ciagent Compression > **Autonomy:** full. Auto-resolution with assumption logging per > `config.autonomy.level: "full"`. No human escalation unless > confidence < 0.60. The prior conversation resolved all material > ambiguities (4 user-answered questions). This file records the > assumptions for the v1.27 record. --- ## Method The clarify stage identifies ambiguities in the v1.27 specification and resolves them at full autonomy. The v1.27 spec is the user-approved plan from the prior conversation + the STATE.md design locked by 4 question answers. Each ambiguity gets a decision ID (D-214+; continuing from the v1.26 decisions D-200..D-213), a resolution, a confidence score, and a rationale. --- ## Prior-conversation resolutions (already locked, restated for the record) These were resolved by user-answered questions in the conversation that spawned v1.27. They are load-bearing for v1.27 execution and cited here so the v1.27 record is self-contained. ### Q-P1 — What should the new PO-reference file catalog? **Resolution:** Capability catalog (what the system can do today). **Confidence:** 1.0 (user-confirmed). **Decision:** D-214. ### Q-P2 — How should the new file relate to CAPABILITY_INVENTORY.md? **Resolution:** Call it `STATE.md`. PO-owned, ciagent-updated after milestone implementation. CAPABILITY_INVENTORY.md is archived. **Confidence:** 1.0 (user-confirmed). **Decision:** D-215. ### Q-P3 — Where should the file live, and who owns it? **Resolution:** Owned by the PO, updated by ciagent after the milestone is implemented with additives. **Confidence:** 1.0 (user-confirmed). **Decision:** D-216. ### Q-P4 — How should "additive when new features are implemented" be enforced? **Resolution:** On the last phase / milestone ship (the P-final Wave 3 "milestone ship" step). No regression-gate check in this pass. **Confidence:** 1.0 (user-confirmed). **Decision:** D-217. ### Q-P5 — Should the initial STATE.md backfill all shipped capabilities through v1.26? **Resolution:** Backfill all shipped capabilities through v1.26 (compressed one-liners for v1.1–v1.24; full entries for v1.25 + v1.26). **Confidence:** 1.0 (user-confirmed). **Decision:** D-218. ### Q-P6 — Should the v1.26 pre-execution artifacts (CLARIFY, GRILL, IDEATE, RESEARCH) be archived? **Resolution:** Archive all 4 to `.ciagent/archive/` with `-v1.26` suffixes. The next milestone's P0 writes fresh versions. Decisions are already folded into PROJECT.md load-bearing decisions + PLAN.md binding revisions. **Confidence:** 1.0 (user-confirmed). **Decision:** D-219. --- ## Ambiguities + Resolutions (this CLARIFY pass) ### Q1 — Is v1.27 a feature milestone or an NFR milestone? **Ambiguity:** v1.27 authors `STATE.md` (a new file/capability for the PO) and archives 11 files. Does the new-file authoring count as `feat:` (making this a feature milestone, tags on v1.26.x with progressive patches) or `docs:`/`chore:` (NFR milestone, same tag behavior but subject to the NFR purity gate)? **Resolution:** NFR milestone. `STATE.md` is documentation (a catalog of existing capabilities), not a new platform capability. The archive moves are `chore:` (file relocation, lossless). No code, no schema, no platform behavior change. Tags run on the v1.26.x patch line: `v1.26.0` (P0) → `v1.26.1..v1.26.3` (P1..P3). The final phase's patch (`v1.26.3`) IS the milestone release. **Confidence:** 0.95. **Decision:** D-220. ### Q2 — Where does the consumer-side archive (nova-blockchain-exchange/ROADMAP.md) land? **Ambiguity:** The platform archive convention is `.ciagent/archive/-.md`. The consumer subproject (`nova-blockchain-exchange/`) has no `archive/` subdirectory. Does the consumer ROADMAP archive at `.ciagent/archive/` (platform-side, mixed) or `.ciagent/nova-blockchain-exchange/archive/` (consumer-side, new subdir)? **Resolution:** Consumer-side. Create `.ciagent/nova-blockchain-exchange/archive/` and relocate to `ROADMAP-v1.26.md`. This preserves the per-project path convention (multi-project mode: `.ciagent//` paths). The platform archive directory is not mixed with consumer archives. **Confidence:** 0.92. **Decision:** D-221. ### Q3 — Does archiving CLARIFY/GRILL/IDEATE/RESEARCH lose the "how v1.26 was specified" traceability? **Ambiguity:** The pre-execution artifacts document the v1.26 decision path. Archiving them moves them out of active context. Is the traceability preserved? **Resolution:** Yes. Three layers preserve it: (1) the archive files are byte-identical relocations inside `.ciagent/archive/` (reachable by agents + git history); (2) the decisions D-200..D-213 are folded into `PROJECT.md` load-bearing decisions (the durable record); (3) git history at the v1.26 commits preserves the authoritative state. The active-context reduction is the point — v1.26 is shipped; the next P0 writes fresh CLARIFY/GRILL/IDEATE/RESEARCH. **Confidence:** 0.95. **Decision:** D-222. ### Q4 — Should IAM_POLICY.md be archived (it predates v1.26 and is dated v1.11)? **Ambiguity:** `IAM_POLICY.md` is dated v1.11 (2026-07-28). It predates v1.26 by 5 milestones. The D-207 future key-split (P1+ R-3 in REVIEW-AUDIT-P05) is pending. Archive or keep? **Resolution:** Keep active. `IAM_POLICY.md` is a live baseline — referenced by the regression gate (`tests/test_iam_policy_baseline.py`), enforced by a managed policy on account `581513795199`, and the D-207 key-split is a pending future- hardening item. It is not stale; it is a baseline that grows when grants change. The v1.11 date reflects the last grant addition, not staleness. **Confidence:** 0.90. **Decision:** D-223. ### Q5 — Should REGRESSION_REPORT.{json,md} be refreshed as part of v1.27? **Ambiguity:** Both files are dated 2026-08-01 (v1.10 Phase 52), show CAP-025 absent, and mark live-aws CAPs "Skipped" (state bucket absent pre-v1.26 re-bootstrap). They are stale. Should v1.27 refresh them? **Resolution:** No. Both files are machine-managed — written by `core/regression_verify.py:704-705` on every `run_regression.sh` run. They regenerate on the next regression run. v1.27 is docs/chore only (no code); touching machine-managed files by hand creates a drift source. The stale state is honest (the last gate run was v1.10; the next run regenerates). The STATE.md Domain 7 row "Regression gate" notes the current CAP range (CAP-001..025). **Confidence:** 0.88. **Decision:** D-224. ### Q6 — Does PROJECT.md get the v1.26 phase-status fix in v1.27 P1 or P2? **Ambiguity:** The plan splits work into P1 (author + archive) and P2 (fix stale + wire). The PROJECT.md phase-status fix (P3/P4/P5 pending → complete) is a "fix stale" item. P1 or P2? **Resolution:** P2. P1 is the additive authoring + lossless archive moves. P2 is the corrections to kept files + the ship-discipline wiring. This keeps P1 a pure-additive, no-edit phase (easier review + audit) and P2 the correction phase. The PROJECT.md fix is a correction; P2. **Confidence:** 0.85. **Decision:** D-225. --- ## Summary 6 prior-conversation resolutions (D-214..D-219, all user-confirmed) + 6 new ambiguities (D-220..D-225, all auto-resolved at full autonomy, confidence ≥ 0.60). 0 escalations. **Key decisions:** - D-220: v1.27 is an NFR milestone (tags on v1.26.x; final patch is the milestone release). - D-221: Consumer archives land in `.ciagent/nova-blockchain-exchange/archive/`. - D-222: Archiving pre-execution artifacts preserves traceability (archive files + PROJECT.md load-bearing decisions + git history). - D-223: IAM_POLICY.md stays active (live baseline, test-enforced, D-207 pending). - D-224: REGRESSION_REPORT.{json,md} regenerate on next `run_regression.sh` (machine-managed; v1.27 is docs/chore only). - D-225: PROJECT.md phase-status fix is P2 (correction phase), not P1 (additive phase).