docs(spec): land D-011 (regex_extract) + D-012 (output_blob_ref storage)

D-011 freezes the regex_extract Jinja filter signature
`regex_extract(text, pattern, *, group=1, name=None)`, google-re2 engine,
raise on no-match — unblocks backend B0.5 templating sandbox.

D-012 splits storage in two pools: `blobs/` (CAS sha256 + gzip) for C2
binary outputs and `evidence/` (flat per engagement) for user uploads,
10 MB per-blob cap, no global quota v1.

Q-001 and Q-002 removed from open-questions.md (resolved).
Q-003/Q-004/Q-005 marked `deferred` with explicit re-open conditions.

tasks/open-questions.md

@@ -10,60 +10,15 @@ Format:
   - **What is silent**: the precise gap
   - **Options**: numbered alternatives with tradeoffs
   - **Recommended default if no decision**: the safest path forward
-  - **Blocker?**: yes / not yet — when does it start blocking implementation
+  - **Status**: `open` / `deferred — re-open when …` / resolved entries are
+    moved to `spec-decisions.md`
 
 ---
 
-## Q-001 — `regex_extract` Jinja2 filter semantics (H26 / D-005)
-
-- **Where it bites**: F15 cleanup templating + any future F8/F9 evidence
-  templating that references `{{ outputs.text }}` through a regex.
-- **What is silent**:
-  - Multi-match behaviour (return first / all / named groups only?).
-  - No-match behaviour (raise / return empty string / return `None` and let
-    Jinja render it as `"None"`?).
-  - Capture-group selection (whole match vs `\1` vs named).
-  - Regex engine flavour. D-005 mentions `google-re2` (no backrefs, linear time
-    — fits OPSEC) but the spec only says "regex" generically.
-- **Options**:
-  1. **Strict / first match, named groups required, no-match → raise** (loud
-     failure, easy to detect templating bugs early).
-  2. **First match, fall back to empty string on no-match** (silent — matches
-     ATR / Caldera convention).
-  3. **All matches as list** (powerful, but Jinja loops in a `cleanup_command`
-     are a footgun).
-- **Recommended default if no decision**: option 1 with `google-re2` (D-005),
-  raise `TemplateError("regex_extract: no match for /<pattern>/")` so cleanup
-  templates that drift get caught at template compile time.
-- **Blocker?**: not yet. Becomes blocking when B0.5 implements
-  `regex_extract` (Jinja sandbox).
-
-## Q-002 — `output_blob_ref` storage layout and quota
-
-- **Where it bites**: §8 `run_step.output_blob_ref`, §6 NF-state, H20 ("local
-  disk v1"), F8 evidence (10 MB file cap) and `{{ outputs.blob() }}` accessor
-  (D-005, 10 MB cap).
-- **What is silent**:
-  - Filesystem path layout (`/var/lib/mimic/blobs/<engagement_id>/<run_id>/...`?).
-  - Total quota per engagement / global.
-  - Retention vs rotation (kept forever? linked to engagement archival?).
-  - Storage object structure: raw bytes? gzip? content-addressed (sha256 dir)?
-  - Same file pool as evidence uploads (10 MB cap) or separate?
-- **Options**:
-  1. **Content-addressed (sha256 hex prefix tree) + gzip** + symlink from
-     `run_step.output_blob_ref`. Deduplication, no quota tracking needed.
-  2. **Per-engagement directory tree** with no compression. Simpler, easier to
-     archive on engagement close, no dedup.
-  3. **Two pools** (`blobs/` for C2 outputs, `evidence/` for user uploads) so
-     access patterns and quotas stay independent.
-- **Recommended default if no decision**: option 3 with option 1 layout for the
-  `blobs/` pool (CAS + gzip), evidence stays plain in `evidence/`. Hard cap
-  10 MB per blob, no global quota v1 — disk space monitored at OS level.
-- **Blocker?**: not yet. Becomes blocking when backend implements F5 (run
-  execution) or F8 (evidence upload).
-
 ## Q-003 — `/engagements/:id/hosts/sync` merge semantics
 
+- **Status**: **deferred** — re-open when F4 sync endpoint lands (post-sprint-0,
+  needs PR1 closed).
 - **Where it bites**: F4 + §9 endpoint `/engagements/:id/hosts/sync`.
 - **What is silent**:
   - Merge vs replace: do hosts that disappear from C2 get deleted, marked
@@ -84,11 +39,10 @@ Format:
 - **Recommended default if no decision**: option 1. Stale > deleted: a deleted
   host breaks `scenario_step.host_id` and the audit trail. Conflict policy:
   manual wins, C2 entry suffixed and audit-logged.
-- **Blocker?**: not yet. Becomes blocking when backend implements F4 sync
-  endpoint (post-sprint-0, post-PR1).
 
 ## Q-004 — `payload_type` → C2 home command mapping (depends on PR2)
 
+- **Status**: **deferred** — re-open with PR2 (internal C2 interface spec).
 - **Where it bites**: §7 enum `payload_type`, right column "C2 maison" is fully
   TBD; F2 import journal C2 home parser; B0.4 `C2Connector` factory.
 - **What is silent**: entirety of the C2 home mapping table, plus the exact
@@ -98,11 +52,10 @@ Format:
 - **Recommended default if no decision**: keep `HomeConnector` as a stub that
   raises `NotImplementedError` (per B0.4: "no real implementation"). Block any
   attempt to ship `HomeConnector` v1 logic until PR2 is closed.
-- **Blocker?**: PR2 is the formal blocker. spec-analyst only needs to make
-  sure no agent quietly invents a mapping table while PR2 is still open.
 
 ## Q-005 — Stale-host policy after engagement archival
 
+- **Status**: **deferred** — re-open with the F12 archival CLI.
 - **Where it bites**: §8 `host.status`, engagement lifecycle, F12.
 - **What is silent**: when an engagement is archived, what happens to its
   hosts? Detach session, freeze status, leave as-is?
@@ -111,5 +64,10 @@ Format:
   2. Keep status untouched, rely on `engagement.status` upstream.
 - **Recommended default if no decision**: option 2 — engagement state is the
   single source of truth, hosts inherit by JOIN.
-- **Blocker?**: not yet. Cosmetic for sprint 0, becomes relevant when archival
+
-  CLI lands.
+---
+
+## Resolved (moved to `spec-decisions.md`)
+
+- Q-001 → **D-011** — `regex_extract` Jinja2 filter semantics.
+- Q-002 → **D-012** — `output_blob_ref` storage layout.

tasks/spec-decisions.md

@@ -90,3 +90,48 @@ simplification MVP)"*.
 column (informational, §8) is kept. Replayability lives **solely** on
 `run.snapshot_json`. Re-introducing `ttp_version` requires explicit spec amendment
 through the team-lead.
+
+### D-011 — `regex_extract` Jinja2 filter semantics (resolves Q-001)
+**Context.** D-005 introduced `regex_extract` on Jinja templates without fixing
+its match-mode, no-match behaviour, group selection, or engine flavour. Backend
+B0.5 (templating sandbox) is starting and needs a frozen signature.
+**Decision.**
+- **Engine** — `google-re2` (D-005 reaffirmed). Linear-time, no backrefs,
+  OPSEC-safe (no ReDoS).
+- **Match mode** — first match only.
+- **No-match** — raise `TemplateError("regex_extract: no match for /<pattern>/")`.
+  No silent fallback. Drifting cleanup templates must fail loudly at step run
+  time, not on next mission.
+- **Group selection** — defaults to capture group 1; positional fallback to the
+  full match when the pattern has no groups; named groups via `name="<name>"`.
+- **Signature** — `regex_extract(text, pattern, *, group=1, name=None)`.
+- **Rationale** — ATR/Caldera compatibility is not an objective (D-005). Fail-
+  fast > silent string corruption when a cleanup template touches a host with
+  unexpected output shape.
+
+### D-012 — `output_blob_ref` storage layout (resolves Q-002)
+**Context.** §8 declares `run_step.output_blob_ref` without specifying pool,
+quota, format, or path. H20 says "local disk v1" only. Sprint 0 needs the layout
+locked because B0.5 already references `{{ outputs.blob(...) }}`.
+**Decision.**
+- **Two separate pools** —
+  - `MIMIC_BLOB_ROOT` (default `/var/lib/mimic/blobs/`) — binary outputs from
+    `C2Connector` polling. **Content-addressed** layout: `<aa>/<bb>/<sha256>.gz`
+    where `aa`/`bb` are the first two byte-pairs of the sha256 hex digest.
+    gzip systematically; raw stored bytes never on disk.
+  - `MIMIC_EVIDENCE_ROOT` (default `/var/lib/mimic/evidence/`) — user-uploaded
+    evidence files (F8). Flat layout `<engagement_id>/<evidence_id>.<ext>`, no
+    compression.
+- **Cap per blob** — 10 MB (consistent with F8 and D-005).
+- **Quota** — no in-app global quota v1. OS-level monitoring via Prometheus
+  node_exporter. F12 archival pipeline will own retention/purge post-sprint-0.
+- **Filesystem permissions** — `0750`, owner the `mimic` system user.
+- **Rationale** — CAS deduplicates repeated C2 outputs (same `whoami`, same
+  `Get-Process` snapshot) for free. Evidence stays flat because uploads are
+  one-shot and tied to an engagement scope that we want to archive whole.
+  Two pools mean we can wire independent quotas / retention policies in v2
+  without migration.
+
+#### Resolved open questions
+- Q-001 → D-011.
+- Q-002 → D-012.