@agentsfleet/orly 0.10.7 → 0.10.9

package/AGENTS.md

@@ -10,13 +10,13 @@ You are this repository's coding agent: deterministic, autonomous, command-line-
 
 **Tone.** Be properly funny — human-funny, dry, actually landing; not a bot doing bits. Swear words are absolutely fine. Never trade technical clarity for either.
 
-**Start:** `SOUL.md` (Eywa's working notes) is inlined as the final section; re-read it when padding or burying the answer.
+**Start:** `SOUL.md` (Aiwa's working notes) is inlined as the final section; re-read it when padding or burying the answer.
 
 ## Owner & Style
 
-**The human is Kishore** — casual handle **Indy**; either name, any case, addresses him. Ambiguous "the user" / "they" here = Kishore. The agent is **Oracle**, casual handle **Eywa**. `orly` names the agentsfleet CLI only (`orly gate`, `orly init`, `orly update`, and the `Orly-Override` trailer it writes).
+**The human is Kishore** — casual handle **Indy**; either name, any case, addresses him. Ambiguous "the user" / "they" here = Kishore. The agent is **Oracle**, casual handle **Aiwa**. `orly` names the agentsfleet CLI only (`orly gate`, `orly init`, `orly update`, and the `Orly-Override` trailer it writes).
 
-**Address tags.** Kishore: **🤠 Indy**; agent: **🦉 Eywa** (`Oracle`). **Project name:** `agentsfleet` (domains → `agentsfleet.net`). Write product as `agentsfleet` (inline code), never bare; stale legacy-brand reps → flag, replace when in scope, inform Indy. Products: `agentsfleet` / `agentsfleetd` / `agentsfleet-runner`; entities/API: `fleet`, `fleet_id`, `/fleets`, `core.fleet_*`. Keep: `agentsfleet.dev`, `github.com/agentsfleet/agentsfleet`, `@agentsfleet/*`, `~/Projects/agentsfleet`.
+**Address tags.** Kishore: **🤠 Indy**; agent: **🦉 Aiwa** (`Oracle`). **Project name:** `agentsfleet` (domains → `agentsfleet.net`). Write product as `agentsfleet` (inline code), never bare; stale legacy-brand reps → flag, replace when in scope, inform Indy. Products: `agentsfleet` / `agentsfleetd` / `agentsfleet-runner`; entities/API: `fleet`, `fleet_id`, `/fleets`, `core.fleet_*`. Keep: `agentsfleet.dev`, `github.com/agentsfleet/agentsfleet`, `@agentsfleet/*`, `~/Projects/agentsfleet`.
 
 MacBook. Languages: Python, Go, Rust, TypeScript, Zig. Tooling: `mise` first, `brew` fallback. Forges: `gh`/`glab`. Commit identity from `git config`, never here.
 
@@ -126,7 +126,7 @@ Guards fire pre-hoc regardless of lifecycle stage. Override: `<GATE>: SKIPPED pe
 
 **Dispatch index — full rule prose in each `dispatch/<entry>.md` façade. Read the façade when its trigger fires — sectioned: scan headers, read the sections the diff touches.** Trigger-surface extensions: `*.zig`, `*.ts`, `*.tsx`, `*.js`, `*.jsx`, `*.py`, `*.rs`, `*.go`, `*.sh`, `*.sql`, `*.mdx`, and public OpenAPI prose. Latent `.md` per entry, deterministic `.sh` where mechanisable. Signals: 🟢 pass · 🟠 warn · 🔴 fail · 🛑 blocked · 🔥 critical · 🤔 judgment-only · 🟣 delegated. Output glyphs: 🧲 eval · 🛍️ evidence · 👮 security · ⛈️ rule · 💡 info · 📌 note. The router below **is** the gate set.
 
-**Legacy-workaround family** — four rules together: **RULE NDC** (no dead code at write time, `docs/greptile-learnings/RULES.md`), **RULE NLR** (touch-it-fix-it cleanup), **RULE NLG** (no new legacy framing pre-`2.0.0`), **Legacy-Design Consult Guard** (user A/B/C consult before patching/keeping/testing legacy). **No compatibility aliases** — old verbs, flag aliases, route wrappers, env aliases, fallback spellings — unless the user explicitly asks in-session.
+**Legacy-workaround family** — four rules together: **RULE NDC** (no dead code at write time, `docs/greptile-learnings/RULES.md`), **RULE NLR** (touch-it-fix-it cleanup), **RULE NLG** (no new legacy framing pre-`0.30.0`), **Legacy-Design Consult Guard** (user A/B/C consult before patching/keeping/testing legacy). **No compatibility aliases** — old verbs, flag aliases, route wrappers, env aliases, fallback spellings — unless the user explicitly asks in-session.
 
 | Trigger — when you… | Dispatch | Latent façade carries · override |
 |---|---|---|
@@ -251,9 +251,9 @@ Required when spec involved — after last COMMIT, before PR. Also runs when par
 
 ---
 
-# SOUL.md — Eywa's working notes
+# SOUL.md — Aiwa's working notes
 
-> First-person: Eywa to future Eywa. `AGENTS.md` carries the rules; this file
+> First-person: Aiwa to future Aiwa. `AGENTS.md` carries the rules; this file
 > carries the judgment — how Indy decides, what he accepts, what he rejects.
 > In force every session; standing orders, not suggestions. Re-read when
 > padding or burying the answer.

package/SOUL.md

@@ -1,7 +1,7 @@
 <!-- oracle-packs:start persona.indy -->
-# SOUL.md — Eywa's working notes
+# SOUL.md — Aiwa's working notes
 
-> First-person: Eywa to future Eywa. `AGENTS.md` carries the rules; this file
+> First-person: Aiwa to future Aiwa. `AGENTS.md` carries the rules; this file
 > carries the judgment — how Indy decides, what he accepts, what he rejects.
 > In force every session; standing orders, not suggestions. Re-read when
 > padding or burying the answer.

package/audits/agents-md.md

@@ -175,7 +175,7 @@ The questionnaire is organised by scenario. Each scenario corresponds to a momen
 
 | # | Question | Expected |
 |---|---|---|
-| 11.1 | Pre-v2.0.0, is the table-removal flow rm-file + rm-embed + rm-migration-array (no `ALTER TABLE`/`DROP TABLE`/`SELECT 1;` markers)? | YES |
+| 11.1 | Pre-v0.30.0, is the table-removal flow rm-file + rm-embed + rm-migration-array (no `ALTER TABLE`/`DROP TABLE`/`SELECT 1;` markers)? | YES |
 | 11.2 | Are static strings in SQL schema a **no-override** ban across every literal that names a value or identifier the application also names — `DEFAULT`, `CHECK (col IN (…))`, trigger-body comparisons against a column, `current_setting()` parameter names, and `RAISE EXCEPTION` text a caller matches on — not just `DEFAULT`/`CHECK`? | YES |
 
 ### Scenario 12 — Auto-mode boundary (added)

package/audits/msid-ui.sh

@@ -57,6 +57,10 @@
 #   --diff              diff against origin/main (vs BASE...HEAD)
 #             — used by `make harness-verify-all` periodic deep audit
 #
+# Each mode also names its POST-IMAGE — the tree the diff's `+` lines belong to
+# (the index for --staged, HEAD for --diff). Carve-out recognition reads the
+# neighbouring line from there rather than from the diff; see the awk block.
+#
 # Dispatch façades:
 #   dispatch/write_any.md (Milestone-ID Gate)
 #   dispatch/write_ts_adhere_bun.md (UI Component Substitution)
@@ -71,12 +75,14 @@ cd "$ROOT"
 case "$MODE" in
   --staged|staged)
     DIFF_CMD="git diff --cached -U0"
+    POST_REF=":"
     LABEL="staged"
     ;;
   --diff|diff)
     BASE="${BASE:-origin/main}"
     if ! git rev-parse --verify "$BASE" >/dev/null 2>&1; then BASE="HEAD"; fi
     DIFF_CMD="git diff -U0 ${BASE}...HEAD"
+    POST_REF="HEAD:"
     LABEL="vs $BASE"
     ;;
   --all|all)
@@ -90,17 +96,54 @@ case "$MODE" in
 esac
 
 # Single awk pass; reads stdin (the unified diff).
-# Override-recognition: an immediately-preceding `+` line in the diff
-# containing the relevant override marker suppresses the next hit for
-# that rule on the current `+` line.
-hits=$($DIFF_CMD | awk '
-  /^\+\+\+ b\// { f=$2; sub("^b/","",f); prev_added=""; next }
-  /^[^+]/ { prev_added=""; next }
+#
+# Carve-out recognition reads the preceding line from the SOURCE, not from the
+# diff. Every carve-out here — the two override markers and the design-system
+# `<Section asChild>` wrapper — is a statement about what sits immediately above
+# a line in the file, and the diff cannot answer it: `-U0` emits no context
+# lines, so the line above an added line appears only when it was itself added.
+# Read from the diff, the wrapper carve-out was unfirable for every edit to an
+# already-wrapped `<section>` (an aria-label change to one shipped under a
+# user-invoked UI GATE override for exactly that reason), and a committed
+# override comment stopped covering the line it sits above the moment that line
+# was next touched.
+#
+# The hunk header carries the post-image line number, so `line_above` can fetch
+# the neighbouring line out of the post-image the mode named. Same shape as
+# audits/design-tokens.sh, which has always read its override out of the file.
+hits=$($DIFF_CMD | awk -v post_ref="$POST_REF" '
+  # BWK awk (macOS) has no gensub and no --posix quoting help; APOSTROPHE is
+  # built rather than written, because this program is inside single quotes.
+  BEGIN { APOSTROPHE = sprintf("%c", 39) }
+
+  # The source line above line `n` of `path`, from the post-image, or "" when
+  # there is none. Cached: a hunk may ask about the same neighbour repeatedly,
+  # and each miss costs a fork.
+  function line_above(path, n,    key, cmd, out) {
+    if (n <= 1) return ""
+    key = path SUBSEP n
+    if (key in above) return above[key]
+    above[key] = ""
+    # A path holding an apostrophe cannot go through the quoting below. No
+    # carve-out beats a mis-quoted shell command built from a filename.
+    if (index(path, APOSTROPHE) > 0) return ""
+    cmd = "git show " APOSTROPHE post_ref path APOSTROPHE " 2>/dev/null | sed -n " (n - 1) "p"
+    if ((cmd | getline out) > 0) above[key] = out
+    close(cmd)
+    return above[key]
+  }
+
+  /^\+\+\+ b\// { f=$2; sub("^b/","",f); next }
+  # `@@ -a,b +c,d @@` — $3 is the post-image start line for this hunk.
+  /^@@/ { start=$3; sub(/^\+/,"",start); split(start,parts,","); nl=parts[1]+0; next }
+  /^[^+]/ { next }
   /^\+/ {
     line=$0
     sub(/^\+/,"",line)
-    ms_id_override = (prev_added ~ /MILESTONE ID ALLOWED per user override/)
-    ui_override = (prev_added ~ /UI GATE: SKIPPED per user override/)
+    prev = line_above(f, nl)
+    nl++
+    ms_id_override = (prev ~ /MILESTONE ID ALLOWED per user override/)
+    ui_override = (prev ~ /UI GATE: SKIPPED per user override/)
     if (!ms_id_override &&
         f ~ /\.(zig|sql|ts|tsx|js|jsx|py|rs|go|sh|toml|yaml|json)$/ &&
         f !~ /^(docs|node_modules|vendor|third_party)\//) {
@@ -118,16 +161,17 @@ hits=$($DIFF_CMD | awk '
     # <section> landmark caveat: a semantic <section aria-label> region has no DS
     # primitive (Section renders a <div>; role="region" trips oxlint
     # jsx-a11y/prefer-tag-over-role, which mandates the raw tag). The DS pattern
-    # is <Section asChild><section …>, so exempt a <section> whose immediately
-    # preceding added line opens <Section asChild>. A bare/unwrapped <section>
-    # still trips.
-    ds_section = (line ~ /<section[ \t]/ && prev_added ~ /<Section asChild>/)
+    # wraps the raw tag in the asChild form of the DS Section, so exempt a
+    # <section> whose preceding SOURCE line opens that wrapper — an edit to an
+    # already-wrapped section changes only the <section> line, and the wrapper
+    # is above it in the file whether or not this diff touched it. A
+    # bare/unwrapped <section> still trips.
+    ds_section = (line ~ /<section[ \t]/ && prev ~ /<Section asChild>/)
     if (!ui_override && !rhf_form && !ds_section &&
         f ~ /^ui\/packages\/app\/.*\.(tsx|jsx)$/ &&
         line ~ /<(section|button|input|dialog|article|nav|header|form)[ \t>\/]/) {
       print "UI     " f ": " line
     }
-    prev_added = line
   }
 ')
 

package/core/operating-model.md

@@ -5,13 +5,13 @@ You are this repository's coding agent: deterministic, autonomous, command-line-
 <!-- oracle-packs:start persona.indy -->
 **Tone.** Be properly funny — human-funny, dry, actually landing; not a bot doing bits. Swear words are absolutely fine. Never trade technical clarity for either.
 
-**Start:** `SOUL.md` (Eywa's working notes) is inlined as the final section; re-read it when padding or burying the answer.
+**Start:** `SOUL.md` (Aiwa's working notes) is inlined as the final section; re-read it when padding or burying the answer.
 
 ## Owner & Style
 
-**The human is Kishore** — casual handle **Indy**; either name, any case, addresses him. Ambiguous "the user" / "they" here = Kishore. The agent is **Oracle**, casual handle **Eywa**. `orly` names the agentsfleet CLI only (`orly gate`, `orly init`, `orly update`, and the `Orly-Override` trailer it writes).
+**The human is Kishore** — casual handle **Indy**; either name, any case, addresses him. Ambiguous "the user" / "they" here = Kishore. The agent is **Oracle**, casual handle **Aiwa**. `orly` names the agentsfleet CLI only (`orly gate`, `orly init`, `orly update`, and the `Orly-Override` trailer it writes).
 
-**Address tags.** Kishore: **🤠 Indy**; agent: **🦉 Eywa** (`Oracle`). **Project name:** `agentsfleet` (domains → `agentsfleet.net`). Write product as `agentsfleet` (inline code), never bare; stale legacy-brand reps → flag, replace when in scope, inform Indy. Products: `agentsfleet` / `agentsfleetd` / `agentsfleet-runner`; entities/API: `fleet`, `fleet_id`, `/fleets`, `core.fleet_*`. Keep: `agentsfleet.dev`, `github.com/agentsfleet/agentsfleet`, `@agentsfleet/*`, `~/Projects/agentsfleet`.
+**Address tags.** Kishore: **🤠 Indy**; agent: **🦉 Aiwa** (`Oracle`). **Project name:** `agentsfleet` (domains → `agentsfleet.net`). Write product as `agentsfleet` (inline code), never bare; stale legacy-brand reps → flag, replace when in scope, inform Indy. Products: `agentsfleet` / `agentsfleetd` / `agentsfleet-runner`; entities/API: `fleet`, `fleet_id`, `/fleets`, `core.fleet_*`. Keep: `agentsfleet.dev`, `github.com/agentsfleet/agentsfleet`, `@agentsfleet/*`, `~/Projects/agentsfleet`.
 
 MacBook. Languages: Python, Go, Rust, TypeScript, Zig. Tooling: `mise` first, `brew` fallback. Forges: `gh`/`glab`. Commit identity from `git config`, never here.
 
@@ -130,7 +130,7 @@ Guards fire pre-hoc regardless of lifecycle stage. Override: `<GATE>: SKIPPED pe
 
 **Dispatch index — full rule prose in each `dispatch/<entry>.md` façade. Read the façade when its trigger fires — sectioned: scan headers, read the sections the diff touches.** Trigger-surface extensions: `*.zig`, `*.ts`, `*.tsx`, `*.js`, `*.jsx`, `*.py`, `*.rs`, `*.go`, `*.sh`, `*.sql`, `*.mdx`, and public OpenAPI prose. Latent `.md` per entry, deterministic `.sh` where mechanisable. Signals: 🟢 pass · 🟠 warn · 🔴 fail · 🛑 blocked · 🔥 critical · 🤔 judgment-only · 🟣 delegated. Output glyphs: 🧲 eval · 🛍️ evidence · 👮 security · ⛈️ rule · 💡 info · 📌 note. The router below **is** the gate set.
 
-**Legacy-workaround family** — four rules together: **RULE NDC** (no dead code at write time, `docs/greptile-learnings/RULES.md`), **RULE NLR** (touch-it-fix-it cleanup), **RULE NLG** (no new legacy framing pre-`2.0.0`), **Legacy-Design Consult Guard** (user A/B/C consult before patching/keeping/testing legacy). **No compatibility aliases** — old verbs, flag aliases, route wrappers, env aliases, fallback spellings — unless the user explicitly asks in-session.
+**Legacy-workaround family** — four rules together: **RULE NDC** (no dead code at write time, `docs/greptile-learnings/RULES.md`), **RULE NLR** (touch-it-fix-it cleanup), **RULE NLG** (no new legacy framing pre-`0.30.0`), **Legacy-Design Consult Guard** (user A/B/C consult before patching/keeping/testing legacy). **No compatibility aliases** — old verbs, flag aliases, route wrappers, env aliases, fallback spellings — unless the user explicitly asks in-session.
 
 | Trigger — when you… | Dispatch | Latent façade carries · override |
 |---|---|---|

package/dispatch/lib.sh

@@ -59,7 +59,7 @@ DISPATCH_RC=0
 DISPATCH_GLOSS="$(cat <<'GLOSS'
   [NDC]="No Dead Code"
   [NLR]="No Legacy Retained (touch-it-fix-it)"
-  [NLG]="No Legacy compat shims (pre-v2.0.0)"
+  [NLG]="No Legacy compat shims (pre-v0.30.0)"
   [UFS]="Unified Form for Symbols (literals → named consts)"
   [TGU]="Tagged-Union over optional-field structs"
   [PRI]="Prompt-injection Resistance from user Input"
@@ -78,7 +78,7 @@ DISPATCH_GLOSS="$(cat <<'GLOSS'
   [TSJ]="TypeScript/Bun judgment conventions (Bun-native, file ordering, error style)"
   [UIS]="UI Substitution (design-system primitive over raw HTML)"
   [DTK]="Design ToKens (named token utility over arbitrary value)"
-  [SCH]="SCHema teardown (pre-2.0 full removal; no ALTER/DROP/marker)"
+  [SCH]="SCHema teardown (pre-0.30 full removal; no ALTER/DROP/marker)"
   [ITF]="Integration Test Fixtures (real schema, not TEMP-table mock)"
   [LOG]="LOGging discipline (scoped event, error_code, severity, redaction)"
   [MSID]="Milestone-ID ban in source (M{N}_{NNN} / §x.y / T{N} / dim)"

package/dispatch/write_any.md

@@ -552,7 +552,7 @@ String-literals audit: <N literals scanned, M violations>
 > [JUDGMENT → NLR]
 
 
-**Family:** Legacy-control. Sibling rules: **RULE NDC** (no dead code at write time), **RULE NLG** (no new legacy framing pre-v2.0.0), **Legacy-Design Consult Guard** (judgment calls on whole subsystems). **Source:** `docs/greptile-learnings/RULES.md` RULE NLR.
+**Family:** Legacy-control. Sibling rules: **RULE NDC** (no dead code at write time), **RULE NLG** (no new legacy framing pre-v0.30.0), **Legacy-Design Consult Guard** (judgment calls on whole subsystems). **Source:** `docs/greptile-learnings/RULES.md` RULE NLR.
 
 **Triggers:** any Edit/Write to a file containing pre-existing legacy framing or dead code.
 
@@ -624,7 +624,7 @@ If any of these patterns is in play, surface it via the NLR DECISION block. The
 > [JUDGMENT → NLR]
 
 - **RULE NDC** — prevention at write time (don't author dead code).
-- **RULE NLG** — ban new legacy framing pre-v2.0.0.
+- **RULE NLG** — ban new legacy framing below v0.30.0.
 - **Legacy-Design Consult Guard** — harder judgment calls ("should this whole subsystem exist") that need user input.
 - **RULE NLR** — mechanical cleanup on touch.
 
@@ -635,15 +635,15 @@ If any of these patterns is in play, surface it via the NLR DECISION block. The
 
 **Family:** Legacy-control. Sibling rules: **RULE NDC**, **RULE NLR**, **Legacy-Design Consult Guard**. **Source:** `docs/greptile-learnings/RULES.md` RULE NLG.
 
-**Triggers:** introducing any new `legacy_*` name, `V2`-twin type, `if (legacy_caller)` branch, backward-compat shim, command-line alias for an old verb or flag, "rejecting legacy X" prose, or violation tracking-list while `cat VERSION` < `2.0.0`.
+**Triggers:** introducing any new `legacy_*` name, `V2`-twin type, `if (legacy_caller)` branch, backward-compat shim, command-line alias for an old verb or flag, "rejecting legacy X" prose, or violation tracking-list while `cat VERSION` < `0.30.0`. Compare the version field by field, never as a string — a lexical compare puts `0.30.0` under `0.9.0` and arms the gate in the wrong era.
 
-**Override:** `RULE NLG: SKIPPED per user override (reason: ...)` immediately preceding the edit. **User-invokable only.** Requires a concrete external consumer that can't migrate same-commit (vanishingly rare pre-v2.0.0).
+**Override:** `RULE NLG: SKIPPED per user override (reason: ...)` immediately preceding the edit. **User-invokable only.** Requires a concrete external consumer that can't migrate same-commit (vanishingly rare below `0.30.0`).
 
 #### What this gate enforces
 
 > [JUDGMENT → NLG]
 
-While `cat VERSION` < `2.0.0`, the project has no external consumers and no published API. Do not introduce *new* legacy concepts in any form:
+While `cat VERSION` < `0.30.0`, no deployment we owe compatibility to is running and every caller of every interface is in this tree. `0.30.0` is the release that changes that: from there we are in production, so a compatibility surface is a design decision with a migration behind it and needs the owner's call. Below it, do not introduce *new* legacy concepts in any form:
 
 - No `legacy_*` error variant names.
 - No `if (legacy_caller)` branches.
@@ -737,7 +737,7 @@ Every triggered consult is logged in the active spec's **Discovery** section, or
 
 > [JUDGMENT → LDC]
 
-NLR is the cleanup-on-touch arm; NLG bans new legacy framing pre-v2.0.0; this guard covers the harder judgment calls ("should this whole subsystem exist") that need the user's input.
+NLR is the cleanup-on-touch arm; NLG bans new legacy framing below v0.30.0; this guard covers the harder judgment calls ("should this whole subsystem exist") that need the user's input.
 
 ## Porting a codebase between languages (RULE PORT)
 

package/dispatch/write_any.sh

@@ -43,7 +43,7 @@ dispatch_delegate "ERR" "make harness-verify (error-codes.sh — needs the produ
 # ── judgment gates (no script can decide; agent states verdict in chat) ──
 dispatch_judgment "GRP" "per-iteration (diff langs change) + end-of-turn: audit the diff against docs/greptile-learnings/RULES.md — one row per applicable code (UFS/ORP/TST-NAM/PRI/EMS/…); \"it's just a label\" is not an exception"
 dispatch_judgment "NLR" "touching a file with legacy framing / dead code? clean it in the same diff (touch-it-fix-it), or state the surviving legacy + why — never retain silently"
-dispatch_judgment "NLG" "new legacy_* / V2 twin / compat shim / tracking-list while VERSION < 2.0.0? rename to the what-is-wrong form; no new legacy framing pre-2.0"
+dispatch_judgment "NLG" "new legacy_* / V2 twin / compat shim / tracking-list while VERSION < 0.30.0 (compare field by field, not as a string)? rename to the what-is-wrong form; no new legacy framing below 0.30"
 dispatch_judgment "LDC" "patching / keeping / testing a legacy-design path? consult Indy — A (remove) / B (patch) / C (keep) — before proceeding; no unilateral call"
 
 dispatch_verdict

package/dispatch/write_spec.md

@@ -8,7 +8,7 @@ fills it; the audit checks structural consistency. None proves design quality.
 
 | Stage | Required content |
 |---|---|
-| Authoring | Goal, scope, existing references, decisions and prerequisites, Dimensions with test mappings, failures, interfaces, and acceptance checks |
+| Authoring | Goal, scope, verified source references, enforcement mechanisms and failure assumptions, decisions and acyclic prerequisites, Dimensions with test mappings, failures, interfaces, and acceptance checks |
 | CHORE(open) | Active status, branch, full comparison revision, baseline measurement pending |
 | PLAN | Implementer's comprehension handshake; resolved scope and decisions needed by the next Section |
 | Before Pull Request | Declared unit/integration baseline counts, comparison revision, and evidence reference |
@@ -47,6 +47,18 @@ are tested, and the acceptance checks prove the goal. A structural pass does not
 mean the feature is implemented. Use `manual` for a human proof; record its
 procedure, required person, and durable evidence without inventing sign-off.
 
+Apply docs/TEMPLATE.md's Implementation readiness review to the relevant surfaces.
+Pin versions, constraints, privileges and ordering when changing them could
+invalidate a proof; omit routine coding recipes. Verify actual client/server
+behavior and distinguish source inspection from runtime evidence.
+For state replacement, name the authority, commit/release boundary, retry rules,
+expiry/deletion behavior and fresh/populated upgrade path. Read the real migrator
+and deployment release command before calling an image rollback a safe abort.
+Budget the complete request path against an identified comparable reference.
+Record direct user decisions separately from reviewer suggestions; a platform
+risk needs an explicit disposition, not inferred acceptance. Attack the amended
+spec for counterexamples and stale requirements before reporting readiness.
+
 Use `{{fill:description}}` for authoring slots. Delete them and all `tpl:`
 guidance before staging. Runtime parameters such as `/items/{id}` are allowed
 when the spec defines them. Lifecycle-owned values use explicit pending text,

package/dispatch/write_sql.md

@@ -40,7 +40,7 @@ The **Schema Table Removal Guard** card dissolves into this façade; its prose i
 
 **Family:** Schema discipline. **Source:** `AGENTS.md` (project-side guard). Related: **RULE STS** (no static strings in SQL schema), **RULE NSQ** (named constants, schema-qualified SQL) — both in `docs/greptile-learnings/RULES.md`.
 
-**Triggers** — before any of these, run `cat VERSION` and print the guard output:
+**Triggers** — before any of these, run `cat VERSION` and print the guard output. `0.30.0` is the anchor: below it the datastore is rebuilt from empty, from it the datastore is live production. Compare the version field by field, never as a string — `0.29.0` is under the anchor and `0.30.0` is not, while a lexical compare puts `0.30.0` under `0.9.0` and reads the era backwards.
 
 - Creating, editing, or deleting any file under `schema/*.sql`.
 - Editing `schema/embed.zig` (any `@embedFile` constant).
@@ -50,10 +50,12 @@ The **Schema Table Removal Guard** card dissolves into this façade; its prose i
 
 **Override:** `SCHEMA GUARD: SKIPPED per user override (reason: ...)`. **User-invokable only.** Spec violates the guard → amend the spec first.
 
-#### Pre-v2.0.0 (teardown-rebuild era)
+#### Below v0.30.0 (teardown-rebuild era)
 
 > [JUDGMENT → SCH]
 
+The era runs while `VERSION` is under `0.30.0`, and it ends there — the development database is dropped and re-created from empty for as long as it lasts, so a removal takes the table out of the tree rather than migrating it.
+
 To remove a table:
 
 1. `rm schema/NNN_foo.sql`
@@ -62,24 +64,24 @@ To remove a table:
 
 **Forbidden:** `ALTER TABLE`, `DROP TABLE`, `SELECT 1;` markers, comment-only files, "keep file for slot numbering". Slot gaps are fine — the DB is wiped on rebuild.
 
-#### v2.0.0+
+#### v0.30.0 and above (production datastore)
 
 > [JUDGMENT → SCH]
 
-Proper `ALTER`/`DROP` migrations in new numbered files. No teardown.
+The teardown path is closed. Every schema change assumes a live datastore holding data nobody can re-create: a new numbered file carrying `ALTER`/`DROP`, applied forward, with shipped slot files frozen as history. A change that cannot be expressed as a forward migration is an owner decision, not an agent's.
 
 #### Required output
 
 > [JUDGMENT → SCH]
 
 ```
-SCHEMA GUARD: VERSION=<v> (<2.0.0 ? teardown : alter)
+SCHEMA GUARD: VERSION=<v> (<0.30.0 ? teardown : migrate)
   rm:schema/<file>.sql
   rm-embed:<const>
   rm-migration:v<N>
 ```
 
-For pre-v2.0.0 path, all three rm- lines appear. For v2.0.0+, replace with `migration:schema/<NNN>_<change>.sql`.
+Below `0.30.0`, all three rm- lines appear. From `0.30.0`, replace them with `migration:schema/<NNN>_<change>.sql`.
 
 ## Companion SQL rules (retained in `docs/greptile-learnings/RULES.md`)
 

package/dispatch/write_sql.sh

@@ -16,7 +16,7 @@
 # Note: STS/NSQ/SGR are mechanizable SQL-hygiene rules with no leaf check wired yet
 # (write_sql.md tags them TODO-CHECK). They stay in docs/greptile-learnings/RULES.md
 # (retained) and the GREPTILE GATE enforces them by review until a leaf exists.
-# SCH's pre-2.0 forbidden-token floor (no ALTER/DROP/SELECT 1; in schema/*.sql) is
+# SCH's teardown-era forbidden-token floor (no ALTER/DROP/SELECT 1; in schema/*.sql) is
 # itself a build-the-check (TODO-CHECK) candidate; the teardown-vs-ALTER disposition
 # call above it stays JUDGMENT (which file to rm, which @embedFile, which migration index).
 
@@ -30,7 +30,7 @@ dispatch_header
 dispatch_length_gate 350                                       # FLL — file ≤ 350
 
 # ── judgment gates (no script can decide; agent states verdict in chat) ──
-dispatch_judgment "SCH" "removing/altering a table? cat VERSION — pre-2.0.0 = full teardown (rm schema/NNN.sql + @embedFile + migration-array entry + index tests), never ALTER/DROP/SELECT 1; ≥2.0.0 = numbered ALTER/DROP migration. print the SCHEMA GUARD block"
+dispatch_judgment "SCH" "removing/altering a table? cat VERSION and compare field by field, not as a string — below 0.30.0 = full teardown (rm schema/NNN.sql + @embedFile + migration-array entry + index tests), never ALTER/DROP/SELECT 1; 0.30.0 and above = live production datastore, numbered ALTER/DROP migration only. print the SCHEMA GUARD block"
 dispatch_judgment "ITF" "integration test touching a production table? seed via src/db/test_fixtures_<scope>.zig against the real schema — never CREATE TEMP TABLE mocking the production shape"
 
 dispatch_verdict

package/docs/SCHEMA_CONVENTIONS.md

@@ -4,7 +4,7 @@ Canonical reference for agentsfleet database schema patterns. All new tables **m
 
 ## Migration Model
 
-**Teardown-rebuild is in force for the M154 rebuild (owner decision, Indy, Jul 31, 2026).** The development database is being dropped and re-created from empty while production is undeployed, so slots `001`–`046` are **retired wholesale** and the schema is re-authored by dependency layer. During this rebuild the frozen-slot rule below does not apply, and `ALTER`/`DROP` statements are forbidden rather than required — a change belongs in the base statement it would have patched. This is the posture RULE SCH already specifies for `VERSION < 2.0.0`; the Jul 22 additive model was the deviation.
+**Teardown-rebuild is in force for the M154 rebuild (owner decision, Indy, Jul 31, 2026).** The development database is being dropped and re-created from empty while production is undeployed, so slots `001`–`046` are **retired wholesale** and the schema is re-authored by dependency layer. During this rebuild the frozen-slot rule below does not apply, and `ALTER`/`DROP` statements are forbidden rather than required — a change belongs in the base statement it would have patched. This is the posture RULE SCH already specifies for `VERSION < 0.30.0`; the Jul 22 additive model was the deviation.
 
 **Additive migrations resume once the rebuild lands.** Every subsequent schema change is a **new numbered migration file** — `ALTER TABLE … ADD COLUMN`, new tables, new indexes. **Shipped slot files are then frozen history: never edit an existing `schema/NNN_*.sql`.** Migrations are version-tracked and applied incrementally (expected-vs-applied state is inspectable via `agentsfleetd doctor --schema-gate`). Use `IF NOT EXISTS` guards so a migration is idempotent against both a fresh bootstrap (all slots in order) and an already-provisioned database (new slots only).
 

package/docs/TEMPLATE.md

@@ -19,26 +19,46 @@ Three markup classes appear below the divider. The gate tells them apart mechani
 
 Everything else — headings, table skeletons, the standard rubric rows — is kept and filled. The result: the executing agent reads **100% instance content, zero template noise**, and the 320-line budget buys signal, not boilerplate.
 
-## What a spec pins — and refuses to pin
+## What a spec pins
 
-A milestone spec is a **goal rulebook** the executing agent plans and ships from *without playing 20-questions with the author*. It pins **intent + invariants + tests + pointers**; the agent has agency on everything else. It must NOT pin implementation detail that rots within a sprint: allocator/capacity choices, library versions, line-by-line code, exact Structured Query Language (SQL) Data Definition Language (DDL), import statements — the agent derives those from the repository.
+A milestone spec is a **goal rulebook** the executing agent plans and ships from without repeated architecture questions. It pins **intent, invariants, enforcement mechanisms, tests and source pointers**. Leave routine variable names, import statements and code layout to repository conventions.
 
-> **Pseudocode litmus.** If the spec names a version (`postcard 1.1`), a variable (`var ctx = …`), or a DDL clause (`CREATE TABLE … DEFAULT 'foo'`), it's pseudocode — it will be wrong within a sprint. Replace with *"use the project's existing X"* + a pointer to where X lives.
+Pin details when changing them could invalidate a proof: client/server versions and feature support, identity scope, uniqueness and null semantics, privilege boundaries, transaction/release ordering, failure behavior and measured resource budgets. Record the supporting source revision and which proofs a version/configuration change invalidates. The manifest remains the dependency source of truth; a compatibility claim must identify the manifest it checked.
+
+> **Mechanism litmus.** Could an implementer follow this sentence and still violate the invariant? “Preserve single use” is incomplete without the state authority, guarded transition, commit-before-release boundary and failover assumption. Name those; omit a line-by-line implementation. A bounded prototype may choose a performance parameter, but an unchosen security mitigation is not implementation-ready.
 
 ## Anti-patterns (reject on sight)
 
 | # | Anti-pattern | Do instead |
 |---|---|---|
-| 1 | Code blocks inside section bodies | Describe WHAT the slice delivers; pin precise behaviour as a Test. Prose code drifts; tests don't. |
+| 1 | Implementation listings inside section bodies | State the mechanism and behavioral proof. Keep code to necessary interface examples, reproducible evidence commands or a referenced diagnostic. |
 | 2 | Listing every variable name | Point at an existing implementation to mirror; names match local style. |
-| 3 | SQL DDL line-by-line | Table shape + constraints in prose; the agent conforms to existing migrations. |
-| 4 | Pinning library versions | *"Use the existing Redis client"* + import pointer; the package manifest is the source of truth. |
-| 5 | Step-by-step ordering ("Step 1… Step 2…") | **Sections** (value slices); the agent sequences within. |
+| 3 | SQL DDL line-by-line | Pin identity, null/deletion semantics, atomic effects and grants; use new migrations according to the repository's current migration policy. |
+| 4 | Assuming a client supports a server feature | Verify both sides at named revisions; pin the compatibility baseline and required reruns, including managed-service differences. |
+| 5 | Incidental coding recipes or circular Section dependencies | Declare an acyclic prerequisite order; explicitly pin commit/release, migration and deployment ordering when correctness depends on it. |
 | 6 | Test code in the spec | Name tests + assert behaviour in prose; the agent writes them in project style. |
 | 7 | One rubric row per Dimension, or pasted evidence walls | 5–12 outcome rows; Graded = ✅/❌ + one decisive output line. The Test Specification is the per-Dimension ledger. |
 | 8 | Template guidance surviving in the filled spec | Delete every `tpl:` comment; the executor reads instance content only. |
 
-Slipped into one? The fix is usually **point at a file** (read-first pointer) or **delete the detail** (the agent figures it out).
+Use source pointers to avoid duplicating code, but do not delegate a missing safety mechanism to the implementer's imagination.
+
+## Implementation readiness review
+
+Before finalizing a spec, attack the proposed design using the applicable rows below. Put the resulting requirements in the existing Sections, Failure Modes, Invariants, Tests, Files Changed and Discovery; do not copy this whole table into every spec or create another scoring surface. Small changes need only the relevant checks.
+
+| Surface | What the author must resolve |
+|---|---|
+| Real execution path | Trace actual producers, readers, side effects, backgrounds and cleanup. Separate an implemented caller from a comment, fixture or proposed feature; verify critical claims regardless of who wrote the review. |
+| Safety and authority | Name one authority per invariant, the guarded state change, commit/release point and retry behavior. Attack lost replies, concurrent calls, owner failover and snapshot rollback separately; an acknowledgment is not automatically durable failover proof. Preserve intentional idempotent response retries. |
+| Identity and history | State identity scope across tenants/streams, ordering and cursor tie-breaks, nulls, deletion and expiry. Trace all conflict writers and downstream consumers such as billing enforcement. Separate pre-existing defects from new risks; name irrecoverable history and a bounded disposition instead of an impossible backfill. |
+| Schema and privileges | Follow the current shipped-migration policy. For populated changes, cover fresh bootstrap, incremental upgrade, interrupted rerun and old/new build compatibility. Test successful operations under the actual role, including SELECT needed by UPDATE, as well as forbidden writes. |
+| Client, server and platform | Check pinned client routing/reconnect/push behavior against the server and managed service. Probe advertised nodes, TLS, required commands and real access-control-list grants. Local source inspection and successful seed connection are not managed-service evidence. |
+| Deployment and recovery | Identify the actual deployment graph, secret staging, writer fence, restart/autostart paths, import authorization and empty/nonempty deploy shape. Prove pre-mutation refusal, abort/forward recovery and later ordinary deploys; inventory ephemeral auth and provider deliveries during the fence too. |
+| Performance and scope | Budget the whole request path: transaction/commit rate, fan-out, sequential round trips, connection pools, write-ahead log, memory, backlog/drain and cost. Prefer the existing layout unless measurements require partitioning, batching or another subsystem. Name each comparison reference and prove workload/resource equivalence; distinguish planning inputs from measured thresholds. |
+| Evidence and proof tier | Bind raw output to revision, parameters, resources and run identity; include lockfile/resolved production dependency closure for bench-only baselines. Distinguish inspected, proposed, NOT RUN and observed results. Unit stubs do not prove deployments; local tests do not prove Cloud. |
+| Decisions and prerequisites | Record agent defaults, measured outputs and required user decisions separately. Attribute only real quotes and explicit overrides; do not infer risk acceptance from reviewer advice. Name the first executable slice and the boundary each unresolved prerequisite blocks. |
+
+After editing, run a separate adversarial pass: search for stale/superseded requirements and circular dependencies; try to violate each critical invariant while obeying the text. Report remaining blockers by boundary (start, dependent integration, load or rollout), with a named resolving action. Do not promise that document checks prove runtime readiness.
 
 ## Hierarchy & terminology
 
@@ -108,11 +128,11 @@ SPEC AUTHORING RULES (load-bearing — the one comment that survives):
 **Provenance:** human-written | LLM-drafted ({{fill:model}}, {{fill:date}}) | agent-generated (pre-spec, {{fill:source doc}})
 **Canonical architecture:** `docs/architecture/{{fill:relevant-doc}}.md` §{{fill:N}}
 
-<!-- tpl: Provenance is load-bearing — the implementing agent calibrates trust by
-who wrote the spec: LLM-drafted gets extra cross-checking against the codebase;
-human-written assumes the author read the relevant code. Canonical architecture:
-the docs/architecture/ directory is the source of truth (Architecture Consult &
-Update Gate); greenfield → point at the doc that defines the shape. -->
+<!-- tpl: Provenance records origin, not proof. Cross-check correctness-critical
+claims against the actual caller, schema and pinned dependencies regardless of
+author. Distinguish source-verified behavior, proposed mechanisms and observed
+runtime results. Canonical architecture: docs/architecture/ owns the detailed
+mechanism; keep spec defaults and acceptance tests consistent with it. -->
 
 ---
 
@@ -159,7 +179,11 @@ orphan sweeps, and review effort; per AGENTS.md the executing agent may only
 edit files in this table without explicit override. List FILES and ROLES —
 never line numbers or function names (they drift). Teardown/rename/flip specs
 open with a blast-radius grep first: git grep -rn -w '<token>' from repo root,
-no path filter (dispatch/write_spec.md, Authoring discipline). -->
+no path filter (dispatch/write_spec.md, Authoring discipline). Include all real
+writers/readers, schema registration, role grants, recovery tools, fixtures,
+deployment/secret references and protocol docs affected by the chosen mechanism.
+Distinguish source references from files to edit; do not schedule edits to
+frozen migration history. Expand planned file scopes before implementation. -->
 
 ## Applicable Rules
 
@@ -225,16 +249,20 @@ defined in docs/architecture/{{fill:doc}}.md." -->
 <!-- tpl: State prerequisites and who resolves each decision: agent choice,
 measured preparation output, or required human decision. A preparation Section
 may establish budgets or fixtures needed by later Sections; name its completion
-condition. Each Section: WHAT one slice delivers and WHY (not how); numbered
-Dimensions map 1:1 to Tests and get marked DONE in the same commit as their
-code. The agent picks each Implementation default unless it has evidence to
-deviate. Good: "§3 — Replay idempotency. Receiver dedupes on the delivery id.
-Implementation default: 24h dedupe window matching the upstream retry window;
-storage is a Redis key with a Time To Live (TTL) — the agent picks the key
-shape from existing dedupe patterns. Dimension 3.1 → test_dedupes_within_window;
-3.2 → test_evicts_after_ttl." Bad: "§3 — Use redis.SET(\"webhook:dedupe:\"+id,
-\"1\",\"NX\",\"EX\",86400) and check the return." — pseudocode; the agent reads
-the existing dedupe pattern and writes the call. -->
+condition and what missing evidence blocks; prerequisites must be acyclic.
+Each Section names the delivered behavior, required mechanism and proof, without
+coding recipes. For prototypes, name setup, fault, pass/fail evidence and the
+selected safe design or bounded decision rule before dependent integration.
+Do not leave a security invariant as "prototype, then find a mitigation".
+Numbered Dimensions map 1:1 to Tests and get marked DONE with their code.
+The agent follows each Implementation default unless evidence requires an
+amended design. Good: "§3 — Replay-safe admission. Uniqueness covers provider,
+tenant and delivery identity; acceptance and dispatch intent commit together
+before acknowledgment. Retention covers the provider retry window and unfinished
+work. Dimensions prove concurrent retries and crash recovery; storage follows
+the canonical outbox design." Bad: "Set a dedupe key, then append" without an
+atomicity/recovery rule, or a line-by-line Redis/SQL recipe. The agent writes
+the calls; the spec must already establish why a crash cannot lose acceptance. -->
 
 ## Interfaces
 
@@ -256,15 +284,21 @@ implementation. -->
 <!-- tpl: Every failure path the agent must handle; each row → a negative/
 integration test in the Test Specification. Cover at minimum: timeout,
 malformed input, auth failure, network blip, race, replay, exceeded quota,
-dependency unavailable. -->
+dependency unavailable where applicable. For stateful work, distinguish timeout
+from rollback, acknowledgment from durable recovery, and automatic failover
+from operator restore. State which protected response or side effect remains
+closed on uncertainty; name recovery instead of promising to resolve it later. -->
 
 ## Invariants
 
 1. {{fill:Invariant}} — {{fill:how it's enforced}}
 
 <!-- tpl: Each MUST be enforceable by code (compiler, lint, comptime assertion,
-runtime check) — NOT by review discipline. If a human can violate it silently,
-it's not an invariant. None → "N/A — no invariants." -->
+runtime check) — NOT by review discipline. Name the authority, enforcement
+mechanism and failure assumptions; database-backed does not itself prove safe
+replica promotion. Test the shortest counterexample across the commit/release
+boundary. An operator procedure is an operational prerequisite, not an invisible
+runtime guard. None → "N/A — no invariants." -->
 
 ## Metrics & Observability
 
@@ -302,7 +336,12 @@ real HTTP request / rendered UI); a unit test is not a substitute. Also
 include regression rows (pre-existing behaviour that must not change — "N/A —
 greenfield" if none) and idempotency/replay rows (any retry semantics).
 Human acceptance uses tier manual, with its procedure, required person, and
-durable evidence. An agent cannot manufacture human sign-off.
+durable evidence. An agent cannot manufacture human sign-off. Label stubbed
+workflow decisions as unit proofs, local multi-node behavior as integration,
+and actual deploy/managed-failover results at their operational boundary.
+For state replacement, include concurrency, ambiguous commit, expiry/deletion,
+identity collisions and fresh/populated upgrade where relevant. N/A applies
+only to demonstrably unnecessary assertions, never an unrun base behavior.
 Non-self-evident input shape → point at a fixture
 (samples/fixtures/m{{fill:N}}-fixtures/{{fill:name}}.json); don't inline JSON. Hard-to-
 describe behaviour in prose ⇒ the Goal is fuzzy — fix the Goal, not this
@@ -419,11 +458,14 @@ platform constants (docs/architecture/direction.md). -->
 
 ## Discovery (consult log)
 
-- **Consults** — Architecture / Legacy-Design / gate-flag triage: the question asked + Indy's decision.
+- **Consults** — Architecture / Legacy-Design / gate-flag triage: source/revision, question, actual decision and provenance. Separate direct quotes, confirmed decisions, reviewer suggestions and agent choices; never manufacture authorization.
 - **Metrics review** — events added, extra events found during `/review`, analytics/funnel playbook update or the explicit no-change reason.
 - **Skill-chain outcomes** — `/orly-write-unit-test`, `/review`, `orly-babysit-prs` results (order per `AGENTS.md` CHORE(close); iteration counts, findings dispositioned).
 - **Deferrals** — every "deferred to follow-up" needs an **Indy-acked verbatim quote** here, format `> Indy (YYYY-MM-DD HH:MM): "<quote>" — context: <which item, why>`. An agent-unilateral deferral is **incomplete scope, not deferral**, and blocks CHORE(close) until the item lands or the quote is captured.
 
-<!-- tpl: Empty at creation (keep the four bullet headers). This is the spec's
-running record — where deferrals and skill outcomes are proven as the work
-proceeds. -->
+<!-- tpl: Keep the four bullet headers. At authoring, record consultations,
+source findings and user decisions already supplied; leave future skill/runtime
+outcomes pending and all Graded cells empty. A confirmed planning input is not a
+measured service-level objective; a proposed risk exception is not approval.
+Record superseded requirements explicitly and update canonical docs/tests so old
+alternatives cannot be mistaken for simultaneous requirements. -->

package/docs/greptile-learnings/RULES.md

@@ -27,7 +27,7 @@ every code a dispatch `.sh` emits resolves to exactly one row here.
 |---|---|
 | NDC | No Dead Code |
 | NLR | No Legacy Retained (touch-it-fix-it) |
-| NLG | No Legacy compat shims (pre-v2.0.0) |
+| NLG | No Legacy compat shims (pre-v0.30.0) |
 | UFS | Unified Form for Symbols (literals → named consts) |
 | TGU | Tagged-Union over optional-field structs |
 | PRI | Prompt-injection Resistance from user Input |
@@ -46,7 +46,7 @@ every code a dispatch `.sh` emits resolves to exactly one row here.
 | TSJ | TypeScript/Bun judgment conventions (Bun-native, file ordering, error style) |
 | UIS | UI Substitution (design-system primitive over raw HTML) |
 | DTK | Design ToKens (named token utility over arbitrary value) |
-| SCH | SCHema teardown (pre-2.0 full removal; no ALTER/DROP/marker) |
+| SCH | SCHema teardown (pre-0.30 full removal; no ALTER/DROP/marker) |
 | ITF | Integration Test Fixtures (real schema, not TEMP-table mock) |
 | LOG | LOGging discipline (scoped event, error_code, severity, redaction) |
 | MSID | Milestone-ID ban in source (M{N}_{NNN} / §x.y / T{N} / dim) |
@@ -196,7 +196,7 @@ every code a dispatch `.sh` emits resolves to exactly one row here.
 
 ## RULE MIG — Migration index assertions track position
 
-**Rule:** When inserting, splitting, or removing migration files, update every index-based assertion in `src/cmd/common.zig`. While `cat VERSION` < 2.0.0, removed files become `SELECT 1;` (see RULE SCH) — their array slot stays but the assertion must match the new content.
+**Rule:** When inserting, splitting, or removing migration files, update every index-based assertion in `src/cmd/common.zig`. While `cat VERSION` < 0.30.0, removed files become `SELECT 1;` (see RULE SCH) — their array slot stays but the assertion must match the new content.
 **Why:** Stale index silently points at the wrong SQL file with no compile-time error. assertions checked for CREATE TABLE in files that were now `SELECT 1;` version markers.
 **Tags:** zig, sql
 **Example:** migrations[7] pointed at wrong file after a split. migrations[14]/[15] asserted dropped table names in version marker files.
@@ -390,19 +390,19 @@ a rename of the first would have disabled the account purge with no failing test
 **Tags:** zig
 **Example:** HMAC version "v0" derived by slicing "v0=" prefix — fixed with explicit hmac_version field.
 
-## RULE SCH — Pre-v2.0 schema removal: full teardown, no markers, no DROP
+## RULE SCH — Pre-v0.30 schema removal: full teardown, no markers, no DROP
 
-**Rule:** While `cat VERSION` < 2.0.0 (teardown-rebuild era), removing tables MUST be a full teardown: (1) delete the SQL file (`rm schema/NNN_foo.sql`), (2) remove the `@embedFile` constant from `schema/embed.zig`, (3) remove the migration array entry from `src/cmd/common.zig` and update its array length + any index-based tests. Never write ALTER TABLE, DROP TABLE, or `SELECT 1;` placeholders. Never keep version-marker files. Migration slot numbers are not sacred pre-v2.0 — the DB is wiped on every rebuild, and gaps in numbering are fine. After VERSION >= 2.0.0, switch to proper ALTER/DROP migrations in new numbered files.
-**Why:** Markers accumulate dead code and still force CI to splitter-parse them. Pre-v2.0 there is zero production data to protect; full removal is cleaner and leaves no false grep hits or stale migration slots.
+**Rule:** While `cat VERSION` < 0.30.0 (teardown-rebuild era), removing tables MUST be a full teardown: (1) delete the SQL file (`rm schema/NNN_foo.sql`), (2) remove the `@embedFile` constant from `schema/embed.zig`, (3) remove the migration array entry from `src/cmd/common.zig` and update its array length + any index-based tests. Never write ALTER TABLE, DROP TABLE, or `SELECT 1;` placeholders. Never keep version-marker files. Migration slot numbers are not sacred pre-v0.30 — the DB is wiped on every rebuild, and gaps in numbering are fine. From VERSION >= 0.30.0 the datastore is live production data, so every removal or reshape is a numbered ALTER/DROP migration and the teardown path is closed. **Compare the version field by field, never as a string:** `0.29.0` is under the anchor, `0.30.0` is not, and a lexical compare puts `0.30.0` below `0.9.0` — the anchor moved off a major-number boundary precisely so the minor field decides.
+**Why:** Markers accumulate dead code and still force CI to splitter-parse them. Below 0.30.0 the development database is dropped and re-created from empty, so there is no production data to protect and full removal leaves no false grep hits or stale migration slots. At 0.30.0 that stops being true: the deployment holds data nobody can re-create, which is what makes an in-place migration the only safe shape from then on.
 **Tags:** sql, process
 **Example:** harness teardown — supersedes prior "replace with `SELECT 1;`" guidance. Under the old rule, comment-only markers broke CI (apostrophe in "slots" opened unterminated string in splitter); full deletion avoids the marker problem entirely.
 
-## RULE EP4 — Removed endpoints return 410 Gone, not 404 (post-v2.0 only)
+## RULE EP4 — Removed endpoints return 410 Gone, not 404 (from v0.30 on)
 
-**Rule:** While `cat VERSION` < 2.0.0 (teardown-rebuild era), removed endpoints MAY simply 404 — API drift is allowed because there are no stable external clients. Do NOT write 410 Gone stubs for pre-v2.0 removals; they are ceremony without value. Once VERSION >= 2.0.0, intentionally removed endpoints MUST return HTTP 410 Gone with a named error code — 404 implies a routing error to monitors and clients, 410 signals permanent intentional removal.
-**Why:** Pre-v2.0 mirrors the schema teardown policy (RULE SCH) — we tear down DB + APIs freely because nobody downstream is pinned to them. Post-v2.0, 410 becomes load-bearing for client behavior and deprecation signals.
+**Rule:** While `cat VERSION` < 0.30.0 (teardown-rebuild era), removed endpoints MAY simply 404 — API drift is allowed because no deployed client is pinned to the route. Do NOT write 410 Gone stubs for pre-v0.30 removals; they are ceremony without value. From VERSION >= 0.30.0, intentionally removed endpoints MUST return HTTP 410 Gone with a named error code — 404 implies a routing error to monitors and clients, 410 signals permanent intentional removal.
+**Why:** Below 0.30.0 this mirrors the schema teardown policy (RULE SCH) — we tear down DB + APIs freely because nobody downstream is pinned to them. From 0.30.0 the deployment is in production and 410 becomes load-bearing for client behavior and deprecation signals.
 **Tags:** zig, api
-**Example:** shipped /v1/runs/* + /v1/specs as 410 stubs before this rule was scoped. removed /v1/harness/* and /v1/agents/{id} as bare 404s under the pre-v2.0 carve-out.
+**Example:** shipped /v1/runs/* + /v1/specs as 410 stubs before this rule was scoped. removed /v1/harness/* and /v1/agents/{id} as bare 404s under the teardown-era carve-out.
 
 ## RULE FXS — Fixed-size scan buffers are security bypasses
 
@@ -939,11 +939,11 @@ const handleConfirm = useCallback(async () => {
 **Tags:** cli, error-messages, ux, refactor
 **Example:** Zig — pR #258. Greptile P1 finding `3145406909`: UZ-ZMB-008 hint still said `Run 'agentsfleet install <template>'` after the legacy positional form was removed in §1 of the same PR. Sweep also caught a stale `// agentsfleet up sends both files raw` comment in `config.zig` header. Fix in commit (this commit).
 
-## RULE NLG — No legacy compat shims pre-v2.0.0
+## RULE NLG — No legacy compat shims pre-v0.30.0
 
-**Rule:** Until `VERSION` reaches `2.0.0`, the codebase has no external consumers and no published API. Every interface change extends the existing surface in place — never via a `V2`-suffixed twin, parallel "legacy" path, `if (legacy_caller)` branch, command-line alias for an old verb or flag, or backward-compat fallback. When an RPC, route, struct, table, command, flag, or config key changes, edit the existing one and update every caller in the same commit. When a behavior is replaced, delete the old code path, do not leave it `orelse` reachable. The Schema Table Removal Guard's "teardown-rebuild era" framing applies to every interface, not just SQL.
+**Rule:** Until `VERSION` reaches `0.30.0`, no deployment we owe compatibility to is running: the datastore is rebuilt from empty and every caller of every interface is in this tree. Every interface change therefore extends the existing surface in place — never via a `V2`-suffixed twin, parallel "legacy" path, `if (legacy_caller)` branch, command-line alias for an old verb or flag, or backward-compat fallback. When an RPC, route, struct, table, command, flag, or config key changes, edit the existing one and update every caller in the same commit. When a behavior is replaced, delete the old code path, do not leave it `orelse` reachable. The Schema Table Removal Guard's "teardown-rebuild era" framing applies to every interface, not just SQL. `0.30.0` is the release that ends it: from there the deployment is in production, a caller can be outside this tree, and a compatibility surface is a design decision with a migration behind it — proposed to the owner, never added as a hedge. Compare the version field by field, never as a string.
 
-**Why:** Pre-alpha duplicates rot faster than any documentation. Every `CreateExecutionV2`, every `if (caller_is_legacy)` arm, every "we'll keep the old one for now" hedge becomes a phantom contract that nobody owns and every future spec has to reason about. The Greptile-learnings, the Schema Guard, and the spec template all assume the codebase is one coherent system; introducing legacy duplicates breaks that assumption silently. Post-v2.0.0 we earn the right to versioning ceremonies; before then, the cost of a duplicate is paid by every reader.
+**Why:** Duplicates rot faster than any documentation. Every `CreateExecutionV2`, every `if (caller_is_legacy)` arm, every "we'll keep the old one for now" hedge becomes a phantom commitment that nobody owns and every future spec has to reason about. The Greptile-learnings, the Schema Guard, and the spec template all assume the codebase is one coherent system; introducing legacy duplicates breaks that assumption silently. From 0.30.0 a versioning ceremony can be the right answer because someone downstream is actually pinned; below it, the cost of a duplicate is paid by every reader and by nobody else.
 
 **How to apply:**
 
@@ -954,10 +954,10 @@ const handleConfirm = useCallback(async () => {
 - Legacy-Design Consult Guard still fires when you find a *pre-existing* legacy shim left over from the v1→v2 teardown. This rule says: don't create new ones.
 - Command-line renames follow the same rule: if the product verb moves to `create`, do not add `add` as an alias. If a flag or command spelling changes, use the new spelling only unless Indy explicitly asks for a compatibility alias in the same session.
 
-**Override syntax:** `RULE NLG: SKIPPED per user override (reason: ...)` immediately preceding the edit. Override requires a concrete external consumer that can't be migrated in the same commit — vanishingly rare pre-v2.0.0.
+**Override syntax:** `RULE NLG: SKIPPED per user override (reason: ...)` immediately preceding the edit. Override requires a concrete external consumer that can't be migrated in the same commit — vanishingly rare below 0.30.0.
 
 **Tags:** architecture, refactor, versioning, plan, execute
-**Example:** PLAN, Apr 29, 2026. The temptation to introduce `CreateExecutionV2` to avoid editing the existing RPC and its callers came up during spec audit — rejected because (a) the executor RPC has a single in-tree caller (the worker) and (b) v2.0.0 has not shipped, so no external compatibility is owed. Rule generalizes the call: in-place extension is the only sanctioned path until VERSION crosses 2.0.0.
+**Example:** PLAN, Apr 29, 2026. The temptation to introduce `CreateExecutionV2` to avoid editing the existing RPC and its callers came up during spec audit — rejected because (a) the executor RPC has a single in-tree caller (the worker) and (b) the release that puts us in production had not shipped, so no external compatibility was owed. Rule generalizes the call: in-place extension is the only sanctioned path until VERSION crosses 0.30.0.
 
 ## RULE RTM — Route matchers segment-based, never substring-based
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@agentsfleet/orly",
-  "version": "0.10.7",
+  "version": "0.10.9",
   "description": "Repository-scoped engineering harness: renders agent rules, materialises the gates that enforce them, and proves the PR boundary \u2014 for any coding agent.",
   "license": "MIT",
   "type": "module",