@codyswann/lisa 2.349.1 → 2.351.0

package/package.json

@@ -120,7 +120,7 @@
     }
   },
   "name": "@codyswann/lisa",
-  "version": "2.349.1",
+  "version": "2.351.0",
   "description": "Claude Code governance framework that applies guardrails, guidance, and automated enforcement to projects",
   "main": "dist/index.js",
   "exports": {

package/plugins/lisa/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "lisa",
-  "version": "2.349.1",
+  "version": "2.351.0",
   "description": "Universal governance — agents, skills, commands, hooks, and rules for all projects",
   "author": {
     "name": "Cody Swann"

package/plugins/lisa/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "lisa",
-  "version": "2.349.1",
+  "version": "2.351.0",
   "description": "Universal governance: agents, skills, commands, hooks, and rules for all projects.",
   "author": {
     "name": "Cody Swann"

package/plugins/lisa/.codex-plugin/skills/lisa-github-read-issue/SKILL.md

@@ -48,6 +48,7 @@ Walk the markdown body and capture each top-level `## ` section by name. Standar
 - `Acceptance Criteria` (preserve the Gherkin code-fence verbatim)
 - `Out of Scope`
 - `Target Backend Environment`
+- `Branch Plan` (derived output per `derived-branch-plan` — parsed so callers can *compare* it against a recomputation, never so they can use it as the base branch)
 - `Sign-in Required`
 - `Repository`
 - `Source Artifacts`

package/plugins/lisa/.codex-plugin/skills/lisa-github-validate-issue/SKILL.md

@@ -98,6 +98,7 @@ Per-type content requirements are defined once in the vendor-neutral `work-item-
 | S16 Source Requirement traceability | `product-clarity` | true |
 | S17 Improvement measurability | `acceptance-criteria` | true |
 | S18 Stateless-pickup dry-run | `product-clarity` | true |
+| S19 Branch Plan derivation | `technical` | false |
 | F1 Issue type label exists in repo | `structural` | false |
 | F2 Parent sub-issue exists and is the right type | `structural` | false |
 | F3 Linked issues exist | `structural` | false |
@@ -299,6 +300,26 @@ The autonomy gate, run last, on every build-ready leaf (use the S15 classificati
 
 Zero questions → PASS. Any question → FAIL, with each question listed verbatim as its own remediation line — these are exactly the clarifying comments the caller posts to the source. The structure gates are proxies; this gate checks the readiness property itself: `ready` means a stateless agent can drive this item to its terminal state with zero human clarification (see `work-item-definition-of-ready`).
 
+#### S19 — Branch Plan derivation
+
+Enforces the `derived-branch-plan` rule: the `## Branch Plan` section is **derived only** — recompute it from current config and compare; never trust the rendered branches or read them as input. Resolve the environment under the S8 grammar, map that exact configured key forward through `.lisa.config.json` `deploy.branches`, and require the mapped branch on the remote.
+
+| Case | Verdict |
+|---|---|
+| `runtime_behavior_change = false` (doc-only / config-only / type-only) or an Epic/container, with no plan | `N/A` — absence is correct; never demand one |
+| Exempt work **carrying** a `## Branch Plan` | **FAIL** — hand-authored branches on work that declared no runtime target |
+| Plan matches the recomputed plan | PASS |
+| Plan conflicts with the recomputed plan | **FAIL** |
+| Plan missing the `Derived from:` provenance line and disagreeing | **FAIL** — treated as hand-authored |
+| `Branch from` and `PR into` name two different branches | **FAIL** — malformed; they are the same branch by construction |
+| Derivation hits a stop condition (env absent from `deploy.branches`, ambiguous / non-unique mapping, branch missing on the remote) | **FAIL** — the same stop `lisa-implement` takes; never default to `main` or the remote default |
+| Proposed spec (pre-write) for applicable work with no plan | **FAIL** — `lisa-github-write-issue` must render it before the write |
+| Live **legacy** issue for applicable work with no plan | `N/A` with a repair note — routed to claim time, where `lisa-implement` writes the derived assumption as a comment and proceeds |
+
+Failing a proposed spec is free; failing every existing issue would turn a legacy queue red for a section no human had a way to add, so legacy absence is repaired at claim time instead.
+
+FAIL names both plans (rendered and recomputed) and points the remediation at **the environment, never the branch** — e.g. `"Branch Plan conflicts with the environment mapping. Rendered 'Branch from: release/staging'; recomputed from Target Backend Environment 'production' via deploy.branches → 'main'. Correct the environment, not the branch — the branches are derived. Then re-render."` Never silently choose between the two plans.
+
 ### Feasibility Gates (require GitHub lookups; skip in `--spec-only`)
 
 #### F1 — Issue type label exists in repo
@@ -455,6 +476,7 @@ Output is a single fenced text block. Callers parse it; do not add free-form pro
 - [PASS|FAIL|N/A] S16 Source Requirement traceability — <one-line reason>
 - [PASS|FAIL|N/A] S17 Improvement measurability — <one-line reason>
 - [PASS|FAIL|N/A] S18 Stateless-pickup dry-run — <one-line reason>
+- [PASS|FAIL|N/A] S19 Branch Plan derivation — <one-line reason>
 
 ### Feasibility Gates  (omit this section when --spec-only)
 - [PASS|FAIL|N/A] F1 Issue type label exists in repo — <one-line reason>

package/plugins/lisa/.codex-plugin/skills/lisa-github-write-issue/SKILL.md

@@ -100,6 +100,21 @@ Scenario: <name>
  Human confirmation replaces the automated annotation with a bare key or
  `Confirmed: <env>`. Skip only for doc/config/type-only issues.]
 
+## Branch Plan
+[GENERATED, never hand-authored. Render only when the issue has a Target
+ Backend Environment; omit entirely when `runtime_behavior_change = false`
+ (doc-only / config-only / type-only) or for an Epic/container — absence is
+ correct there. Derive per the `derived-branch-plan` rule: resolve the
+ environment, map it forward through `.lisa.config.json` `deploy.branches`,
+ and prove the branch exists on the remote. Do not accept caller-supplied
+ branches; recompute them. Exactly three lines:
+   Branch from: <branch>
+   PR into: <branch>
+   Derived from: Target Backend Environment <env> via .lisa.config.json deploy.branches
+ Both fields name the same branch by construction. A missing, ambiguous, or
+ non-unique mapping, or a branch absent from the remote, STOPS the write —
+ never default to `main` or the remote default to keep the write alive.]
+
 ## Sign-in Required
 [Include this section ONLY if the work touches authenticated surfaces.
  Specify: the account/role, where to get the credentials (1Password item,

package/plugins/lisa/.codex-plugin/skills/lisa-implement/SKILL.md

@@ -107,6 +107,12 @@ Using the general-purpose agent in Team Lead session, **determine the base branc
    - The only normalization is built-in `prod` ↔ `production`, and only when exactly one of those keys exists in `deploy.branches`; normalize to that configured key. No other aliases exist.
    - Never infer from arbitrary branch text, URL paths or query strings, or substrings inside other words or hostname labels. Multiple conflicting signals after normalization **stop** the flow. If there are no signals, resolve the remote default branch (`gh repo view --json defaultBranchRef -q .defaultBranchRef.name`, or `git remote set-head origin -a` then `origin/HEAD`). When it reverse-maps uniquely, write the env-bearing `Assumption:` form; when the reverse-map is not unique, write the branch-only form and continue on the remote default without inventing an environment or blocking solely for that ambiguity. Record the fallback assumption in the plan/tracker artifact.
 2. **Map the resolved environment to a base branch** through `.lisa.config.json` `deploy.branches` — the forward direction of the env-keyed `done` resolution. The selected exact configured key must map uniquely, and the mapped branch must exist on the remote. A missing/ambiguous mapping or remote branch **stops** the flow; never guess or silently fall back.
+   - **Reconcile the ticket's `## Branch Plan` (`derived-branch-plan` rule).** The rendered plan is derived output, never input: **recompute** it here from current config and the remote — the mapping you just resolved — and compare. Four arms, no fifth:
+     - **Match** → proceed.
+     - **Legacy (no plan)** → derive it, **write the assumption onto the ticket as a comment**, then proceed. Visible prose plus a dedupe marker, e.g. ``Branch plan derived for this item: branch from `main`, PR into `main` (Target Backend Environment: production via .lisa.config.json deploy.branches).`` followed by `<!-- [lisa-branch-plan] key=<work-item-ref>::<branch> -->` (marker-dedupe on `<work-item-ref>::<branch>`, so a re-claim adds no duplicate; where a vendor cannot host an HTML comment the visible line alone carries it). **No silent guess** — if the comment cannot be written, that is a stop, because proceeding would make the inference invisible.
+     - **Conflict with a human-confirmed environment, or with an existing open PR's base** → **stop under the existing confirmation rules** below. A branch plan never overrides the environment and never supplies the confirmation itself.
+     - **Stale (config changed since the plan was rendered)** → current config wins: re-render the section onto the ticket, record the change, and never follow the stale plan. Staleness is output falling behind its input, not a conflict to escalate.
+   - Exempt work carries no plan and needs none: `runtime_behavior_change = false` (doc-only / config-only / type-only) and containers have no `## Target Backend Environment` to derive from, so **absence is correct** — never demand or invent one.
 3. **Establish the feature branch off the latest base, conflict-free:**
    - `git fetch origin`.
    - Already on a feature branch with an **open PR** → reuse it. If the PR's base ≠ the resolved base branch, surface the mismatch and re-target only with confirmation — the ticket's environment is the source of truth.

package/plugins/lisa/.codex-plugin/skills/lisa-jira-validate-ticket/SKILL.md

@@ -95,6 +95,7 @@ Per-type content requirements are defined once in the vendor-neutral `work-item-
 | S16 Source Requirement traceability | `product-clarity` | true |
 | S17 Improvement measurability | `acceptance-criteria` | true |
 | S18 Stateless-pickup dry-run | `product-clarity` | true |
+| S19 Branch Plan derivation | `technical` | false |
 | F1 Issue type valid in project | `structural` | false |
 | F2 Epic parent exists and is an Epic | `structural` | false |
 | F3 Linked tickets exist | `structural` | false |
@@ -301,6 +302,26 @@ The autonomy gate, run last, on every build-ready leaf (use the S15 classificati
 
 Zero questions → PASS. Any question → FAIL, with each question listed verbatim as its own remediation line — these are exactly the clarifying comments the caller posts to the source. The structure gates are proxies; this gate checks the readiness property itself: `ready` means a stateless agent can drive this item to its terminal state with zero human clarification (see `work-item-definition-of-ready`).
 
+#### S19 — Branch Plan derivation
+
+Enforces the `derived-branch-plan` rule: the `Branch Plan` section (`h2.` / `##` in proposed text, or an ADF heading in live JIRA) is **derived only** — recompute it from current config and compare; never trust the rendered branches or read them as input. Resolve the environment under the S8 grammar, map that exact configured key forward through `.lisa.config.json` `deploy.branches`, and require the mapped branch on the remote.
+
+| Case | Verdict |
+|---|---|
+| `runtime_behavior_change = false` (doc-only / config-only / type-only) or an Epic/container, with no plan | `N/A` — absence is correct; never demand one |
+| Exempt work **carrying** a `Branch Plan` | **FAIL** — hand-authored branches on work that declared no runtime target |
+| Plan matches the recomputed plan | PASS |
+| Plan conflicts with the recomputed plan | **FAIL** |
+| Plan missing the `Derived from:` provenance line and disagreeing | **FAIL** — treated as hand-authored |
+| `Branch from` and `PR into` name two different branches | **FAIL** — malformed; they are the same branch by construction |
+| Derivation hits a stop condition (env absent from `deploy.branches`, ambiguous / non-unique mapping, branch missing on the remote) | **FAIL** — the same stop `lisa-implement` takes; never default to `main` or the remote default |
+| Proposed spec (pre-write) for applicable work with no plan | **FAIL** — `lisa-jira-write-ticket` must render it before the write |
+| Live **legacy** ticket for applicable work with no plan | `N/A` with a repair note — routed to claim time, where `lisa-implement` writes the derived assumption as a comment and proceeds |
+
+Failing a proposed spec is free; failing every existing ticket would turn a legacy queue red for a section no human had a way to add, so legacy absence is repaired at claim time instead.
+
+FAIL names both plans (rendered and recomputed) and points the remediation at **the environment, never the branch** — e.g. `"Branch Plan conflicts with the environment mapping. Rendered 'Branch from: release/staging'; recomputed from Target Backend Environment 'production' via deploy.branches → 'main'. Correct the environment, not the branch — the branches are derived. Then re-render."` Never silently choose between the two plans.
+
 ### Feasibility Gates (require JIRA lookups; skip in dry-run if requested)
 
 #### F1 — Issue type valid in project
@@ -388,6 +409,7 @@ Output is a single fenced text block. Callers parse it; do not add free-form pro
 - [PASS|FAIL|N/A] S16 Source Requirement traceability — <one-line reason>
 - [PASS|FAIL|N/A] S17 Improvement measurability — <one-line reason>
 - [PASS|FAIL|N/A] S18 Stateless-pickup dry-run — <one-line reason>
+- [PASS|FAIL|N/A] S19 Branch Plan derivation — <one-line reason>
 
 ### Feasibility Gates  (omit this section when --spec-only)
 - [PASS|FAIL|N/A] F1 Issue type valid in project — <one-line reason>

package/plugins/lisa/.codex-plugin/skills/lisa-jira-write-ticket/SKILL.md

@@ -92,6 +92,21 @@ h2. Target Backend Environment
  Human confirmation replaces the automated annotation with a bare key or
  `Confirmed: <env>`. Skip only for doc/config/type-only tickets.]
 
+h2. Branch Plan
+[GENERATED, never hand-authored. Render only when the ticket has a Target
+ Backend Environment; omit entirely when `runtime_behavior_change = false`
+ (doc-only / config-only / type-only) or for an Epic/container — absence is
+ correct there. Derive per the `derived-branch-plan` rule: resolve the
+ environment, map it forward through `.lisa.config.json` `deploy.branches`,
+ and prove the branch exists on the remote. Do not accept caller-supplied
+ branches; recompute them. Exactly three lines:
+   Branch from: <branch>
+   PR into: <branch>
+   Derived from: Target Backend Environment <env> via .lisa.config.json deploy.branches
+ Both fields name the same branch by construction. A missing, ambiguous, or
+ non-unique mapping, or a branch absent from the remote, STOPS the write —
+ never default to `main` or the remote default to keep the write alive.]
+
 h2. Sign-in Required
 [Include this section ONLY if the work touches authenticated surfaces.
  Specify: the account/role to sign in as, where to get the credentials

package/plugins/lisa/.codex-plugin/skills/lisa-linear-access/SKILL.md

@@ -65,18 +65,40 @@ Error: no Linear access substrate available. Authenticate the Linear MCP or set
 All GraphQL calls use:
 
 ```bash
+# Resolve the key through the chokepoint before giving up on the environment.
+# Without this rung a project that keeps its credentials in Bitwarden, Doppler,
+# or AWS has no path to the key at all — the ladder stopped at rung one, which
+# is also what left `/lisa:setup:linear` reading an OS keychain directly with
+# nowhere to migrate to. Mirrors `atlassian-access` and `notion-access`.
+read_linear_key() {
+  [ -n "${LINEAR_API_KEY:-}" ] && { echo "$LINEAR_API_KEY"; return; }
+  local resolver
+  for resolver in .claude/skills/lisa-secrets-access/scripts/resolve-secret.mjs \
+                  .agents/skills/lisa-secrets-access/scripts/resolve-secret.mjs; do
+    if [ -f "$resolver" ]; then
+      local via_lisa
+      via_lisa=$(node "$resolver" get LINEAR_API_KEY 2>/dev/null) \
+        && [ -n "$via_lisa" ] && { echo "$via_lisa"; return; }
+      break
+    fi
+  done
+  return 1
+}
+
 linear_graphql() {
   local query="$1"
   local variables="${2:-{}}"
-  [ -n "$LINEAR_API_KEY" ] || {
-    echo "Error: LINEAR_API_KEY is not set." >&2
+  local key
+  key=$(read_linear_key) || {
+    echo "Error: no Linear API key. Set LINEAR_API_KEY, or store it as" >&2
+    echo "LINEAR_API_KEY in this project's secrets provider." >&2
     return 1
   }
   jq -n --arg query "$query" --argjson variables "$variables" \
     '{query:$query, variables:$variables}' |
     curl -sS -X POST "https://api.linear.app/graphql" \
       -H "Content-Type: application/json" \
-      -H "Authorization: '"$LINEAR_API_KEY"'" \
+      -H "Authorization: $key" \
       --data-binary @-
 }
 ```

package/plugins/lisa/.codex-plugin/skills/lisa-linear-validate-issue/SKILL.md

@@ -96,6 +96,7 @@ Per-type content requirements are defined once in the vendor-neutral `work-item-
 | S16 Source Requirement traceability | `product-clarity` | true |
 | S17 Improvement measurability | `acceptance-criteria` | true |
 | S18 Stateless-pickup dry-run | `product-clarity` | true |
+| S19 Branch Plan derivation | `technical` | false |
 | F1 Issue type valid in team | `structural` | false |
 | F2 Project parent exists and is in same team | `structural` | false |
 | F3 Linked items exist | `structural` | false |
@@ -303,6 +304,26 @@ The autonomy gate, run last, on every build-ready leaf (use the S15 classificati
 
 Zero questions → PASS. Any question → FAIL, with each question listed verbatim as its own remediation line — these are exactly the clarifying comments the caller posts to the source. The structure gates are proxies; this gate checks the readiness property itself: `ready` means a stateless agent can drive this item to its terminal state with zero human clarification (see `work-item-definition-of-ready`).
 
+#### S19 — Branch Plan derivation
+
+Enforces the `derived-branch-plan` rule: the `## Branch Plan` section is **derived only** — recompute it from current config and compare; never trust the rendered branches or read them as input. Resolve the environment under the S8 grammar, map that exact configured key forward through `.lisa.config.json` `deploy.branches`, and require the mapped branch on the remote.
+
+| Case | Verdict |
+|---|---|
+| `runtime_behavior_change = false` (doc-only / config-only / type-only) or a Project/container, with no plan | `N/A` — absence is correct; never demand one |
+| Exempt work **carrying** a `## Branch Plan` | **FAIL** — hand-authored branches on work that declared no runtime target |
+| Plan matches the recomputed plan | PASS |
+| Plan conflicts with the recomputed plan | **FAIL** |
+| Plan missing the `Derived from:` provenance line and disagreeing | **FAIL** — treated as hand-authored |
+| `Branch from` and `PR into` name two different branches | **FAIL** — malformed; they are the same branch by construction |
+| Derivation hits a stop condition (env absent from `deploy.branches`, ambiguous / non-unique mapping, branch missing on the remote) | **FAIL** — the same stop `lisa-implement` takes; never default to `main` or the remote default |
+| Proposed spec (pre-write) for applicable work with no plan | **FAIL** — `lisa-linear-write-issue` must render it before the write |
+| Live **legacy** Issue for applicable work with no plan | `N/A` with a repair note — routed to claim time, where `lisa-implement` writes the derived assumption as a comment and proceeds |
+
+Failing a proposed spec is free; failing every existing Issue would turn a legacy queue red for a section no human had a way to add, so legacy absence is repaired at claim time instead.
+
+FAIL names both plans (rendered and recomputed) and points the remediation at **the environment, never the branch** — e.g. `"Branch Plan conflicts with the environment mapping. Rendered 'Branch from: release/staging'; recomputed from Target Backend Environment 'production' via deploy.branches → 'main'. Correct the environment, not the branch — the branches are derived. Then re-render."` Never silently choose between the two plans.
+
 ### Feasibility Gates (require Linear lookups; skip in dry-run if requested)
 
 #### F1 — Issue type valid in team
@@ -394,6 +415,7 @@ Output is a single fenced text block. Callers parse it; do not add free-form pro
 - [PASS|FAIL|N/A] S16 Source Requirement traceability — <one-line reason>
 - [PASS|FAIL|N/A] S17 Improvement measurability — <one-line reason>
 - [PASS|FAIL|N/A] S18 Stateless-pickup dry-run — <one-line reason>
+- [PASS|FAIL|N/A] S19 Branch Plan derivation — <one-line reason>
 
 ### Feasibility Gates  (omit when --spec-only)
 - [PASS|FAIL|N/A] F1 Issue type valid in team — <one-line reason>

package/plugins/lisa/.codex-plugin/skills/lisa-linear-write-issue/SKILL.md

@@ -112,6 +112,21 @@ Linear descriptions are markdown (NOT Jira wiki markup — no `h2.` headings, us
  Human confirmation replaces the automated annotation with a bare key or
  `Confirmed: <env>`. Skip only for doc/config/type-only items.]
 
+## Branch Plan
+[GENERATED, never hand-authored. Render only when the item has a Target
+ Backend Environment; omit entirely when `runtime_behavior_change = false`
+ (doc-only / config-only / type-only) or for a Project/container — absence is
+ correct there. Derive per the `derived-branch-plan` rule: resolve the
+ environment, map it forward through `.lisa.config.json` `deploy.branches`,
+ and prove the branch exists on the remote. Do not accept caller-supplied
+ branches; recompute them. Exactly three lines:
+   Branch from: <branch>
+   PR into: <branch>
+   Derived from: Target Backend Environment <env> via .lisa.config.json deploy.branches
+ Both fields name the same branch by construction. A missing, ambiguous, or
+ non-unique mapping, or a branch absent from the remote, STOPS the write —
+ never default to `main` or the remote default to keep the write alive.]
+
 ## Sign-in Required
 [Include this section ONLY if the work touches authenticated surfaces.
  Specify: the account/role to sign in as, where to get the credentials

package/plugins/lisa/.codex-plugin/skills/lisa-secrets-access/SKILL.md

@@ -116,9 +116,11 @@ Every provider has a description field — Bitwarden `note`, 1Password notes, AW
 
 These are two different rules and should not be conflated:
 
-- **Rule A — the note must exist and be well-formed.** Universal: every secret, every provider, every surface. Enforced **statically** by `verify` and by `doctor`. Nothing to do with agents.
+- **Rule A — the note must exist and be well-formed.** Universal: every secret, every provider, every surface. Enforced **statically** by `verify` and by `doctor`, and it **blocks** — a malformed note is an error, not a warning. Nothing to do with agents.
 - **Rule B — an agent must read the note before first use.** Runtime, and **only in lanes where a consumer has latitude**. An agent could do anything with a write-scoped token, and the note is what bounds it. A reviewed workflow step resolving one credential by exact name has no latitude, so gating it there is ceremony that can only fail-closed and never inform.
 
+Rule A blocks because a warning enforced nothing. Both checks previously tested only whether a note was *present*, and reported a warning either way, so a vault of empty notes reported clean — which is precisely why the notes stayed empty. A check that cannot fail is a check nobody acts on.
+
 Format — first line prose, then `key: value` lines:
 
 ```text
@@ -129,6 +131,22 @@ ci: yes - injected by <mechanism>
 docs: <path>
 ```
 
+### What "well-formed" means, exactly
+
+`scripts/note-format.mjs` is the one validator; `verify` and `doctor` both call it, so they cannot reach different verdicts about the same note. It **blocks** on:
+
+| Defect | Why it blocks |
+| --- | --- |
+| `missing-note` | No note at all, empty, or only whitespace. |
+| `no-prose` | Every line is a `key: value` line, so nothing states what the credential *is*. |
+| `empty-field` | A field with nothing after the colon — it promises a fact the reader cannot find. |
+| `empty-tool-line` | A `tool:` line naming no tool: it asks for an install and supplies nothing. |
+| `bad-tool-name` | A `tool:` entry that is not a bare name, so the reader drops it silently. |
+
+It **warns**, without blocking, on `stray-separator` — a trailing or doubled comma in a tool list. The reader handles it; failing a vault over punctuation would only teach operators that the check is noise.
+
+Deliberately **not** checked: which fields a note carries, and how informative the prose is. The documented format mandates neither, and enforcing beyond the prose is the same defect as prose beyond the enforcement, pointing the other way. A line is treated as prose unless the text before its colon is a single bare lowercase token — so "Attio CRM: system of record" is a sentence, not a field.
+
 ### `tool:` — the CLI a secret implies
 
 One key is read by more than a human. A credential and the CLI that consumes it belong together — `SONARQUBE_CLI_TOKEN` is only useful with `sonar` — so a `tool:` (or `tools: a, b`) line declares that pairing where it cannot drift from the secret:
@@ -229,7 +247,7 @@ Cache **in-process only**. Never write a resolved value to disk except through t
 
 - Every name in `require` resolves.
 - Every key matches `^[A-Z][A-Z0-9_]*$`.
-- No secret has an empty note.
+- Every secret's note exists and is well-formed, per the table above. This is an **error**, so a vault that was passing on warnings will newly fail until its notes are written.
 - Every name in `rotating` has a resolvable bootstrap, so its replacement could be persisted.
 - **No secret is readable from two stores.** A value present in both the provider and a local cache is not a duplicate — it is **two live credentials**, one of which is untracked. This is the check most worth having: it catches drift before a deletion turns the forgotten copy into an orphan nobody can revoke.
 

package/plugins/lisa/.codex-plugin/skills/lisa-secrets-access/scripts/doctor-secrets.mjs

@@ -19,6 +19,7 @@
 
 import { createHash } from "node:crypto";
 
+import { validateNote } from "./note-format.mjs";
 import { fetchAll, fetchRotatable } from "./providers.mjs";
 import { readMaterialized } from "./resolve-secret.mjs";
 import { readConfig } from "./surfaces.mjs";
@@ -118,20 +119,23 @@ export function checkNaming(provider, report) {
 }
 
 /**
- * Assert every secret carries a usage note.
+ * Assert every secret carries a usage note in the documented format.
+ *
+ * This blocks rather than warns, and the promotion is the point. The contract
+ * has always said the note "must exist and be well-formed", enforced statically
+ * by `doctor` — but a warning enforces nothing, so a vault of empty notes
+ * reported clean and stayed empty. A check that never fails is a check nobody
+ * acts on.
+ *
+ * The grammar lives in `note-format.mjs` so this and `resolve-secret.mjs verify`
+ * cannot disagree about what a well-formed note is.
  * @param {Map<string, object>} provider Provider view.
  * @param {Function} report Finding collector.
  */
 export function checkNotes(provider, report) {
   for (const [name, entry] of provider) {
-    if (!entry.note?.trim()) {
-      report(
-        "warn",
-        name,
-        "has no usage note. An agent cannot learn this credential's scope " +
-          "without one, and inferring it from the name is exactly the guess " +
-          "that writes to the wrong system"
-      );
+    for (const defect of validateNote(entry?.note)) {
+      report(defect.level, name, defect.message);
     }
   }
 }

package/plugins/lisa/.codex-plugin/skills/lisa-secrets-access/scripts/note-format.mjs

@@ -0,0 +1,179 @@
+/**
+ * Validate a credential's usage note against the format the contract documents.
+ *
+ * The contract states Rule A — "the note must exist and be well-formed",
+ * enforced statically by `verify` and by `doctor`. Only half of that was ever
+ * true. Both consumers tested the note for emptiness and nothing tested its
+ * shape, so a note that promised a scope and delivered a bare colon, or a
+ * `tool:` line naming a URL instead of a CLI, passed every check Lisa had. The
+ * prose described an enforcement that did not exist, which is worse than no
+ * rule: readers stopped checking because they believed something else was.
+ *
+ * This module is that missing half, and it lives in one place on purpose. The
+ * same argument the values file makes for keeping its writer and its parser
+ * together applies here: `tools-from-notes.mjs` decides what a `tool:` line
+ * *is*, and if this validator held a second copy of that pattern, a note could
+ * pass doctor and then be silently ignored at install time with nothing to
+ * reveal the disagreement. So the matcher is defined here once and imported
+ * there.
+ *
+ * Messages are written for the person who has to fix the vault, who is often
+ * not an engineer. Each one names the secret's defect and the edit that clears
+ * it, never only the rule that was broken.
+ * @module note-format
+ */
+
+/**
+ * A `key: value` metadata line.
+ *
+ * The key must be a single bare lowercase token. That is what separates
+ * `scope: read-write` from prose that happens to contain a colon, such as
+ * "Attio CRM: system of record for sales" — a sentence has spaces before its
+ * colon, and a capitalised word opening a sentence is prose, not a field. The
+ * discriminator is deliberately generous: mistaking a field for prose only
+ * loosens a check, while mistaking prose for a field would fail a good note.
+ */
+const FIELD_LINE = /^[ \t]*([a-z][a-z0-9_-]*)[ \t]*:[ \t]*(.*)$/;
+
+/** Matches `tool: name` or `tools: a, b`, case-insensitively, on one line. */
+const TOOL_LINE_SINGLE = /^[ \t]*tools?[ \t]*:[ \t]*(.*)$/i;
+
+/**
+ * The same matcher, global and multiline, for scanning a whole note.
+ *
+ * Derived from the single-line pattern rather than written out again, so the
+ * two can never disagree about what a tool line is.
+ */
+export const TOOL_LINE = new RegExp(TOOL_LINE_SINGLE.source, "gim");
+
+/**
+ * A bare CLI name: what a `tool:` entry is permitted to be.
+ *
+ * Names are matched against Lisa's catalogue and never executed, so this is not
+ * the security boundary — the catalogue is. It is the check that tells an
+ * operator their entry will be silently dropped, which is the failure mode
+ * worth catching: a note asking for `sonar (for CI)` installs nothing, and the
+ * session fails much later for a reason that looks unrelated.
+ */
+const TOOL_NAME = /^[a-z0-9][a-z0-9._+-]*$/;
+
+/**
+ * Whether a line is structured metadata rather than a sentence.
+ * @param {string} line One line of a note.
+ * @returns {boolean} True when the line is a field or a tool line.
+ */
+function isMetadataLine(line) {
+  return FIELD_LINE.test(line) || TOOL_LINE_SINGLE.test(line);
+}
+
+/**
+ * Check the entries on one `tool:` / `tools:` line.
+ * @param {string} value The text after the colon.
+ * @param {object[]} defects Collector to append to.
+ */
+function checkToolLine(value, defects) {
+  if (!value.trim()) {
+    defects.push({
+      level: "error",
+      code: "empty-tool-line",
+      message:
+        "has a `tool:` line that names no tool. Either name the command-line " +
+        "tool this credential is for (for example `tool: sonar`) or delete " +
+        "the line — as written it asks for a tool and supplies none, so " +
+        "nothing gets installed",
+    });
+    return;
+  }
+
+  for (const raw of value.split(",")) {
+    const entry = raw.trim();
+    if (!entry) {
+      // A trailing or doubled comma is sloppy punctuation, not a broken note:
+      // the reader drops the empty entry and installs the rest correctly.
+      // Blocking a vault over a comma would teach operators the check is noise.
+      defects.push({
+        level: "warn",
+        code: "stray-separator",
+        message:
+          "has a `tool:` line with a stray comma. Nothing breaks, but tidy it " +
+          "so the list reads as the tools it actually names",
+      });
+      continue;
+    }
+    if (!TOOL_NAME.test(entry.toLowerCase())) {
+      defects.push({
+        level: "error",
+        code: "bad-tool-name",
+        message:
+          `has a \`tool:\` entry that is not a tool name: "${entry}". Each ` +
+          "entry must be one plain name such as `sonar` or `gh`, separated by " +
+          "commas — no descriptions, links, versions, or punctuation. As " +
+          "written this entry is ignored, so the tool never gets installed",
+      });
+    }
+  }
+}
+
+/**
+ * Report every way a usage note departs from the documented format.
+ *
+ * Returns defects rather than throwing so a caller can list a whole vault's
+ * problems in one pass. An operator fixing a vault wants every edit at once,
+ * not one error per run.
+ * @param {unknown} note The note text stored beside the secret.
+ * @returns {{level: string, code: string, message: string}[]} Defects found.
+ */
+export function validateNote(note) {
+  if (typeof note !== "string" || !note.trim()) {
+    // One cause, one instruction. Also accusing an absent note of lacking a
+    // first line would leave the operator guessing which fix comes first.
+    return [
+      {
+        level: "error",
+        code: "missing-note",
+        message:
+          "has no usage note. Write one in the secret's own description field " +
+          "in the vault: a first line saying what this credential is for and " +
+          "what it can reach, then `key: value` lines such as `scope:`, " +
+          "`owner:`, and `tool:`. Guessing a credential's purpose from its " +
+          "name is exactly the guess that writes to the wrong system",
+      },
+    ];
+  }
+
+  const defects = [];
+  const lines = note.split(/\r?\n/);
+  const populated = lines.filter(line => line.trim());
+
+  if (!populated.some(line => !isMetadataLine(line))) {
+    defects.push({
+      level: "error",
+      code: "no-prose",
+      message:
+        "has no plain-language first line. Add a sentence at the top saying " +
+        "what this credential is for and what it can reach — the `key: value` " +
+        "lines below it list details, but they never say what the thing is",
+    });
+  }
+
+  for (const line of populated) {
+    const tool = TOOL_LINE_SINGLE.exec(line);
+    if (tool) {
+      checkToolLine(tool[1], defects);
+      continue;
+    }
+    const field = FIELD_LINE.exec(line);
+    if (field && !field[2].trim()) {
+      defects.push({
+        level: "error",
+        code: "empty-field",
+        message:
+          `has an empty \`${field[1]}:\` line. Either fill in the value or ` +
+          "delete the line — a field with nothing after the colon promises a " +
+          "fact the reader then cannot find anywhere",
+      });
+    }
+  }
+
+  return defects;
+}

package/plugins/lisa/.codex-plugin/skills/lisa-secrets-access/scripts/resolve-secret.mjs

@@ -30,6 +30,7 @@
 import { existsSync, readFileSync } from "node:fs";
 
 import { parseEnv } from "./envfile.mjs";
+import { validateNote } from "./note-format.mjs";
 import { ENV_KEY, fetchAll } from "./providers.mjs";
 import { materializedPaths, readConfig } from "./surfaces.mjs";
 
@@ -109,6 +110,19 @@ function sourceOf(name, provider, file) {
   return "         ";
 }
 
+/**
+ * Say what is wrong with a note, not merely that something is.
+ *
+ * "NO NOTE" against a note that exists but is malformed sends the operator
+ * looking for a missing field they already wrote.
+ * @param {object} [fault] The blocking defect, if any.
+ * @returns {string} A status label for the verify table.
+ */
+function noteStatus(fault) {
+  if (!fault) return "noted";
+  return fault.code === "missing-note" ? "NO NOTE" : `NOTE ${fault.code}`;
+}
+
 /**
  * Verify every declared secret, mirroring the ladder's own order.
  *
@@ -119,12 +133,20 @@ function sourceOf(name, provider, file) {
  * This command is read-only. Proving that a rotating credential can actually be
  * written back requires a write, so that check lives in the rotation path
  * rather than here.
+ *
+ * The two views are parameters so a test can supply them. Reaching for the
+ * provider CLI is the one thing in here that cannot run in a test, and a check
+ * nobody can test is how the note rule stayed unenforced for so long.
  * @param {object} cfg Resolved configuration.
+ * @param {Map<string, object>} [provider] Provider view.
+ * @param {Map<string, string>} [file] Materialized view.
  * @returns {number} Count of secrets that failed.
  */
-function verify(cfg) {
-  const provider = fetchAll(cfg);
-  const file = readMaterialized(cfg);
+export function verify(
+  cfg,
+  provider = fetchAll(cfg),
+  file = readMaterialized(cfg)
+) {
   const names = cfg.require ?? [
     ...new Set([...provider.keys(), ...file.keys()]),
   ];
@@ -139,13 +161,17 @@ function verify(cfg) {
       provider.get(name)?.value
     );
     const named = ENV_KEY.test(name);
-    const noted = Boolean(provider.get(name)?.note);
-    if (!resolves || !named || !noted) bad += 1;
+    // Presence is not well-formedness. Testing `Boolean(note)` passed a note
+    // that was a single stray character, which is exactly the note an agent
+    // learns nothing from. Same validator doctor uses, so the two agree.
+    const defects = validateNote(provider.get(name)?.note);
+    const noteFault = defects.find(d => d.level === "error");
+    if (!resolves || !named || noteFault) bad += 1;
     console.log(
       `  ${name.padEnd(30)} ${resolves ? "resolves" : "MISSING "} ` +
         `${sourceOf(name, provider, file)} ` +
         `${named ? "name ok" : "NAME NOT UPPER_SNAKE"}  ` +
-        `${noted ? "noted" : "NO NOTE"}` +
+        `${noteStatus(noteFault)}` +
         `${rotating.has(name) ? "  rotating" : ""}`
     );
   }