ticketlens 0.38.33 → 0.38.35

package/README.md

@@ -453,7 +453,7 @@ Every note is scanned before saving — anything shaped like a real secret (API
 
 **Removing a note:** `ticketlens note delete --id="..." [--ticket=KEY]` removes a note from your local vault. Local only — if it was already pushed to a team, teammates who pulled it keep their copy; deleting it there too is a manager action from the Console (Admin > Recall).
 
-**Any MCP-capable AI harness:** `ticketlens mcp` starts a stdio [MCP](https://modelcontextprotocol.io) server exposing `fetch`, `triage`, `compliance`, `review`, `standup`, `pr`, `stats`, `history`, `collisions`, `doctor`, `recall_add`, `recall_search`, `ticket_comment`, `ticket_transition`, `ticket_assign`, `ticket_duplicates`, `ticket_link`, `ticket_update`, and `ticket_create` as native tools — any MCP-compatible AI assistant, not just Claude Code, can call them directly instead of constructing a shell command. It's a thin adapter over the exact same code as the CLI commands above — same license gate per tool, same secret scan/local vault/tracker writes, same team sync — nothing is reimplemented. `fetch`, `doctor`, and `standup` are Free; `triage`'s base scan is Free with some options gated Pro/Team, same as the CLI (`ticketlens triage --help`); `compliance` and `pr` are Free, sharing a 3-checks/month cap on their requirements-coverage section, Pro unlimited; `review` is Free for branch/files/ticket context, with its coverage/focus section requiring Pro as a plain license check — it does not draw from that same monthly counter; `stats` is Free with a 7-day lookback cap, Pro extends it to 30 days, same split as the CLI (`ticketlens stats --help`); `history` reads local triage history only (zero network) and requires Pro; `collisions` requires `ticketlens login` (Console access) plus a Team license; every other tool needs Pro. Point your harness's MCP config at it: `{ "command": "ticketlens", "args": ["mcp"] }` — or run `ticketlens mcp install` in a project to write that entry into its `.mcp.json` for you (creates the file if it doesn't exist, merges in if it does — never touches any other entry already there; `--dry-run` to preview first).
+**Any MCP-capable AI harness:** `ticketlens mcp` starts a stdio [MCP](https://modelcontextprotocol.io) server exposing `fetch`, `triage`, `compliance`, `review`, `standup`, `pr`, `stats`, `history`, `collisions`, `ledger`, `doctor`, `recall_add`, `recall_update`, `recall_delete`, `recall_search`, `ticket_comment`, `ticket_transition`, `ticket_assign`, `ticket_duplicates`, `ticket_link`, `ticket_update`, and `ticket_create` as native tools — every CLI action now has an MCP tool, so any MCP-compatible AI assistant, not just Claude Code, can call them directly instead of constructing a shell command. It's a thin adapter over the exact same code as the CLI commands above — same license gate per tool, same secret scan/local vault/tracker writes, same team sync — nothing is reimplemented. `fetch`, `doctor`, and `standup` are Free; `triage`'s base scan is Free with some options gated Pro/Team, same as the CLI (`ticketlens triage --help`); `compliance` and `pr` are Free, sharing a 3-checks/month cap on their requirements-coverage section, Pro unlimited; `review` is Free for branch/files/ticket context, with its coverage/focus section requiring Pro as a plain license check — it does not draw from that same monthly counter; `stats` is Free with a 7-day lookback cap, Pro extends it to 30 days, same split as the CLI (`ticketlens stats --help`); `history` reads local triage history only (zero network) and requires Pro; `collisions` requires `ticketlens login` (Console access) plus a Team license; `ledger` exports the local, signed compliance audit trail (zero network) and requires Pro; every other tool needs Pro. `recall_update` overwrites an existing Recall note's body — internal plumbing for the note quality loop, not typically called directly. `recall_delete` is destructive and local-vault-only — requires `confirm: true` alongside `id` to actually execute; there is no interactive y/N prompt under MCP (no real terminal to prompt against), so omitting it always fails rather than silently blocking. Point your harness's MCP config at it: `{ "command": "ticketlens", "args": ["mcp"] }` — or run `ticketlens mcp install` in a project to write that entry into its `.mcp.json` for you (creates the file if it doesn't exist, merges in if it does — never touches any other entry already there; `--dry-run` to preview first).
 
 `note add`'s save confirmation and `recall`'s search results are styled by default in a terminal; add `--plain` to either for bare, pipe-safe output. `recall` always shows each note's file ID (e.g. `[1784135399545-fe01c4.md]`) so you can open it directly (`cat ~/.ticketlens/recall/<PREFIX>/<id>`), or pass `--full` to print the full body content inline instead. Each result shows a relative time (`2h ago`, `3d ago`) rather than a bare date — the full-precision timestamp is always in the note file's own frontmatter.
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ticketlens",
-  "version": "0.38.33",
+  "version": "0.38.35",
   "description": "Jira CLI for developers — fetch ticket context, triage your queue, and stop tab-switching. Zero dependencies, all local.",
   "type": "module",
   "bin": {

package/skills/jtb/SKILL.md

@@ -1,4 +1,4 @@
-<!-- jtb-skill-version: 0.38.0 -->
+<!-- jtb-skill-version: 0.39.0 -->
 ---
 name: jtb
 description: Fetch a Jira ticket's full context (description, comments, linked issues, code references) and assemble a structured TicketBrief for implementation planning. Use when user types /jtb, mentions a Jira ticket key, or wants to plan work from a Jira ticket.
@@ -66,7 +66,7 @@ Fetches a Jira ticket and produces a structured brief with code references, then
 /jtb assign PROD-1234 --to=me          # assign the ticket to yourself (Pro)
 ```
 
-**Destructive commands** (`note delete`, `cloud-keys remove`, `delete <profile>`) prompt for interactive y/N confirmation and refuse outright in a non-interactive shell unless `--yes` is passed — there is no way to silently skip this. Only pass `--yes` when the user has explicitly asked for that specific deletion in this conversation (their message *is* the confirmation); never add it to route around the prompt for a deletion you decided to make on your own.
+**Destructive commands** (`note delete`, `cloud-keys remove`, `delete <profile>`) prompt for interactive y/N confirmation and refuse outright in a non-interactive shell unless `--yes` is passed — there is no way to silently skip this. Only pass `--yes` when the user has explicitly asked for that specific deletion in this conversation (their message *is* the confirmation); never add it to route around the prompt for a deletion you decided to make on your own. If this harness has TicketLens's MCP server configured, `note delete`'s equivalent is the `recall_delete` tool (`id`/`ticket`/`confirm`) — it has no interactive prompt at all (no real terminal exists under MCP to prompt against), so `confirm: true` is the only way it ever executes; the same rule applies — only pass it when the user's message is the confirmation for that specific note.
 
 ## Prerequisites
 
@@ -370,6 +370,8 @@ When it does apply, run up to 3 rounds:
      ticketlens note patch --id="THE-ID-PRINTED-ABOVE" --ticket=TICKET-KEY --expect-mtime="THE-MTIME-FROM-STEP-1"
    ```
    `--expect-mtime` is what keeps this safe: if the file changed since step 1 (the user hand-edited it while you were drafting), the patch silently no-ops and prints "not found or already changed" — the user's own edit always wins, never gets clobbered by a stale background draft.
+
+   If this harness has TicketLens's MCP server configured (a tool named `recall_update` — often shown as `mcp__ticketlens__recall_update` — visible in your tool list), prefer it over the bash form: same license gate, same structural/secret-scan checks, same optimistic-concurrency behavior via `expectMtime` — just no shell command or stdin piping to construct. It accepts `id`/`body`/`ticket`/`expectMtime`.
 5. Repeat from step 1 (re-capture mtime/body fresh each round) up to 3 total rounds. If no round ever produces a fully-passing draft, patch in whichever round scored highest across all attempts, and let the "not found or already changed" message stand if that patch itself loses a late race — don't retry past round 3.
 
 This never calls any external API or bills any tokens beyond the session you already have open — the generator and validator are subagents inside your own Claude Code session, not a TicketLens server call.
@@ -506,6 +508,9 @@ The same tier-gated check also runs as its own command — `ticketlens complianc
 
 If this harness has TicketLens's MCP server configured (a tool named `compliance` — often shown as `mcp__ticketlens__compliance` — visible in your tool list), prefer it over the bash form: same tier gate (Free: 3 checks/month, Pro: unlimited), same report — just no shell command to construct or stdout to parse. It accepts `ticket`/`profile`, matching the standalone command's arguments above.
 
+### Ledger export
+`ticketlens ledger [--format=json|csv]` exports the same local, signed compliance ledger these checks write to — entirely local, no network call. `[Pro]`. If this harness has TicketLens's MCP server configured (a tool named `ledger` — often shown as `mcp__ticketlens__ledger` — visible in your tool list), prefer it over the bash form: same tier gate, same export, just no shell command to construct or stdout to parse. It accepts `format` (`json`/`csv`, defaults to `json`).
+
 ---
 
 ## Advanced Options

package/skills/jtb/hooks/recall-nudge-lib.mjs

@@ -18,6 +18,19 @@ export const NOTE_ADD_RE = /\bticketlens\s+note\s+add\b|\/jtb\s+note\b/;
 // (mcp__<server-alias>__<tool-name>) — the server alias is whatever the user
 // named it in their own .mcp.json, so only the tool-name suffix is fixed.
 export const NOTE_ADD_MCP_RE = /^mcp__.+__recall_add$/;
+// Matches jtb's fetch: the CLI's bare/default ticket-key form ("ticketlens
+// PROJ-123" or its "tl" bin alias), the "get" alias form, and the /jtb skill
+// wrapper. Deliberately excludes an explicit "fetch" subcommand — there is
+// no such subcommand; parseCommand() (cli.mjs) only special-cases "get" as
+// an alias, so "ticketlens fetch PROJ-123" falls through to the catch-all
+// with "fetch" itself still in argv and errors as an invalid ticket key
+// (live-verified during code review) — matching it here would silently
+// swallow a broken invocation as if a real fetch had happened. Also
+// deliberately does not match any other tracked subcommand (triage/
+// compliance/etc) — those are lowercase words and can never satisfy the
+// uppercase ticket-key class required immediately after the command name.
+export const FETCH_RE = /\bticketlens\s+(?:get\s+)?[A-Z][A-Z0-9]{1,9}-\d+\b|\btl\s+(?:get\s+)?[A-Z][A-Z0-9]{1,9}-\d+\b|\/jtb\s+(?:get\s+)?[A-Z][A-Z0-9]{1,9}-\d+\b/;
+export const FETCH_MCP_RE = /^mcp__.+__fetch$/;
 
 export function readStdinJson() {
   const raw = fs.readFileSync(0, 'utf8');
@@ -170,9 +183,19 @@ export function hasRecentNag(cwd, now = Date.now()) {
  * early, and picking one deterministic match keeps this function decoupled
  * from profiles.json (it stays a pure transcript reader; profile lookup
  * and its own fallback chain belong entirely to resolveProfile()).
+ *
+ * `sawFetch` (backlog #15) is deliberately narrower than `sawTicketKey`: it
+ * only goes true when jtb's fetch tool actually ran (CLI or MCP form, same
+ * dual-detection shape as sawNoteAdd below) — not merely when a ticket-key-
+ * shaped string appears anywhere in the transcript. `sawTicketKey` false-
+ * positived on any incidental match (a doc, a code comment, a test fixture
+ * name like BETA-42), which is fine for its own low-stakes use (picking
+ * which profile's settings apply) but was wrong as shouldNag()'s nag
+ * precondition — SKILL.md's own capture guidance is scoped to "whenever
+ * jtb's fetch was used," so the hook now checks the same thing it backstops.
  */
 export function scanTranscript(transcriptPath) {
-  const result = { sawTicketKey: false, sawRecallFlag: false, sawNoteAdd: false, ticketKey: null };
+  const result = { sawTicketKey: false, sawRecallFlag: false, sawNoteAdd: false, sawFetch: false, ticketKey: null };
   let lines;
   try {
     lines = fs.readFileSync(transcriptPath, 'utf8').split('\n').filter(Boolean);
@@ -217,6 +240,10 @@ export function scanTranscript(transcriptPath) {
         const isCliNoteAdd = block.name === 'Bash' && NOTE_ADD_RE.test(block.input?.command ?? '');
         const isMcpNoteAdd = NOTE_ADD_MCP_RE.test(block.name ?? '');
         if (isCliNoteAdd || isMcpNoteAdd) result.sawNoteAdd = true;
+
+        const isCliFetch = block.name === 'Bash' && FETCH_RE.test(block.input?.command ?? '');
+        const isMcpFetch = FETCH_MCP_RE.test(block.name ?? '');
+        if (isCliFetch || isMcpFetch) result.sawFetch = true;
       }
     }
   }
@@ -235,9 +262,17 @@ export function scanTranscript(transcriptPath) {
  * compaction/session_id rollover, or nagging more than once per session).
  * Strict's actual effect on capture volume comes from SKILL.md's lowered
  * in-session capture bar, not from this function.
+ *
+ * Gated on `sawFetch`, not `sawTicketKey` (backlog #15) — including the
+ * `sawRecallFlag` broken-promise case, which is why the gate is a blanket
+ * `!sawFetch` check rather than per-branch: a flag can't legitimately fire
+ * outside real ticket work, and this keeps the whole function's trigger
+ * matching SKILL.md's own capture-guidance scope exactly ("unconditionally
+ * whenever jtb's fetch was used"), instead of firing on any incidental
+ * ticket-key-shaped string.
  */
-export function shouldNag({ sawTicketKey, sawRecallFlag, sawNoteAdd, recallStrictness = 'balanced' }) {
-  if (!sawTicketKey || sawNoteAdd) return false;
+export function shouldNag({ sawFetch, sawRecallFlag, sawNoteAdd, recallStrictness = 'balanced' }) {
+  if (!sawFetch || sawNoteAdd) return false;
   if (recallStrictness === 'loose') return sawRecallFlag; // only the broken-promise case
-  return true; // balanced and strict: ticket work with no note is enough
+  return true; // balanced and strict: a fetch with no note is enough
 }

package/skills/jtb/hooks/recall-nudge-stop.mjs

@@ -3,12 +3,16 @@
  * Stop hook — end-of-session Recall check.
  *
  * Blocks (exit 2) at most ONCE per session — never traps the user in a
- * loop regardless of how Claude responds. Two cases force a check:
+ * loop regardless of how Claude responds. Both cases below require jtb's
+ * fetch to have actually run this session (backlog #15) — a ticket-key-
+ * shaped string appearing incidentally (a doc, a code comment, a test
+ * fixture) is not enough, matching SKILL.md's own capture-guidance scope.
+ * Given that:
  *   1. Claude flagged something (🔖 Recall-flag:) but never called note add
  *      — a broken promise, the strongest signal something was missed.
  *   2. Ticket work happened all session with zero flags and zero notes
  *      — the weaker "did anything ever get considered?" catch.
- * Anything else (no ticket work at all, or a note was already added) exits
+ * Anything else (no fetch this session, or a note was already added) exits
  * clean — this must never be the reason a session can't end.
  *
  * The per-session_id "asked once" state (readState/writeState) cannot
@@ -45,7 +49,7 @@ const cwd = input?.cwd ?? process.cwd();
 
 if (!sessionId || !transcriptPath) process.exit(0);
 
-const { sawTicketKey, sawRecallFlag, sawNoteAdd, ticketKey } = scanTranscript(transcriptPath);
+const { sawFetch, sawRecallFlag, sawNoteAdd, ticketKey } = scanTranscript(transcriptPath);
 
 // Refreshed on every check, independent of the once-per-session gate below —
 // a capture that happens AFTER this session already nagged once must still
@@ -58,7 +62,7 @@ if (state.stopChecked) process.exit(0); // already asked once this session — r
 const profile = resolveProfile(ticketKey, { cwd });
 const recallStrictness = normalizeRecallStrictness(profile?.recallStrictness);
 
-if (!shouldNag({ sawTicketKey, sawRecallFlag, sawNoteAdd, recallStrictness })) {
+if (!shouldNag({ sawFetch, sawRecallFlag, sawNoteAdd, recallStrictness })) {
   process.exit(0); // nothing this strictness level requires a capture for
 }
 

package/skills/jtb/scripts/fetch-ticket.mjs

@@ -488,10 +488,9 @@ export async function run(args, envOrOpts = process.env, fetcher = globalThis.fe
   }
 
   const printFn = opts.print ?? ((chunk) => process.stdout.write(chunk));
-  // Mirrors printFn above — the bare ticket-fetch, compliance, pr, review, and standup
-  // paths all use this now (each has an MCP tool). `ledger` above still writes directly
-  // to the real process.stderr (no MCP tool yet); `install-hooks` is CLI-only and never
-  // gets one.
+  // Mirrors printFn above — the bare ticket-fetch, compliance, pr, review, standup, and
+  // ledger paths all use this now (each has an MCP tool). `install-hooks` is CLI-only
+  // and never gets one.
   const printErrFn = opts.printErr ?? ((chunk) => process.stderr.write(chunk));
   // Reused on every recursive self-call in the bare-fetch path (profile-prompt retries)
   // so an injected print/printErr survives the retry instead of silently reverting to
@@ -610,7 +609,7 @@ export async function run(args, envOrOpts = process.env, fetcher = globalThis.fe
     const { isLicensed: isLic, showUpgradePrompt: showUpgrade } = await import('./lib/license.mjs');
     const resolvedConfigDir = configDir ?? (await import('./lib/config.mjs')).DEFAULT_CONFIG_DIR;
     if (!isLic('pro', resolvedConfigDir)) {
-      showUpgrade('pro', 'ledger', { stream: process.stderr });
+      showUpgrade('pro', 'ledger', { stream: errStream });
       process.exitCode = 1;
       return;
     }
@@ -621,7 +620,7 @@ export async function run(args, envOrOpts = process.env, fetcher = globalThis.fe
       printFn(result + '\n');
     } else {
       printFn(JSON.stringify(result, null, 2) + '\n');
-      process.stderr.write('  Verify signature: HMAC-SHA256 over {records, exportedAt} with key at ledger-key\n');
+      printErrFn('  Verify signature: HMAC-SHA256 over {records, exportedAt} with key at ledger-key\n');
     }
     return;
   }

package/skills/jtb/scripts/lib/api-utils.mjs

@@ -3,7 +3,7 @@
  * Centralised here to avoid triplicating the regex and warning logic.
  */
 
-export const DEFAULT_API_BASE = 'https://api.ticketlens.app';
+export const DEFAULT_API_BASE = 'http://api.ticketlens.test';
 export const DEFAULT_SITE_BASE = 'https://ticketlens.app';
 
 // Matches localhost, 127.0.0.1, and any hostname ending in .test or .local,

package/skills/jtb/scripts/lib/help.mjs

@@ -646,11 +646,14 @@ export function printMcpHelp({ stream = process.stdout } = {}) {
     '',
     `  Start an MCP (Model Context Protocol) stdio server exposing CLI actions as`,
     `  native tools — ${s.cyan('fetch')}, ${s.cyan('triage')}, ${s.cyan('compliance')}, ${s.cyan('review')}, ${s.cyan('standup')}, ${s.cyan('pr')}, ${s.cyan('stats')}, ${s.cyan('history')},`,
-    `  ${s.cyan('collisions')}, ${s.cyan('doctor')}, ${s.cyan('recall_add')}, ${s.cyan('recall_search')}, ${s.cyan('ticket_comment')}, ${s.cyan('ticket_transition')},`,
-    `  ${s.cyan('ticket_assign')}, ${s.cyan('ticket_duplicates')}, ${s.cyan('ticket_link')}, ${s.cyan('ticket_update')}, ${s.cyan('ticket_create')} — for any`,
-    `  MCP-compatible AI harness, not just Claude Code. Thin adapter over the same`,
-    `  code as ${s.cyan('TICKET-KEY')}/${s.cyan('doctor')}/${s.cyan('triage')}/${s.cyan('compliance')}/${s.cyan('review')}/${s.cyan('standup')}/${s.cyan('pr')}/${s.cyan('stats')}/${s.cyan('history')}/${s.cyan('collisions')}/`,
-    `  ${s.cyan('note add')}/${s.cyan('recall')}/${s.cyan('comment')}/${s.cyan('transition')}/${s.cyan('assign')}/${s.cyan('duplicates')}/${s.cyan('link')}/${s.cyan('update')}/${s.cyan('create')} above.`,
+    `  ${s.cyan('collisions')}, ${s.cyan('ledger')}, ${s.cyan('doctor')}, ${s.cyan('recall_add')}, ${s.cyan('recall_update')}, ${s.cyan('recall_delete')},`,
+    `  ${s.cyan('recall_search')}, ${s.cyan('ticket_comment')}, ${s.cyan('ticket_transition')}, ${s.cyan('ticket_assign')},`,
+    `  ${s.cyan('ticket_duplicates')}, ${s.cyan('ticket_link')}, ${s.cyan('ticket_update')}, ${s.cyan('ticket_create')} — every CLI action`,
+    `  now has an MCP tool, for any MCP-compatible AI harness, not just Claude Code.`,
+    `  Thin adapter over the same code as ${s.cyan('TICKET-KEY')}/${s.cyan('doctor')}/${s.cyan('triage')}/${s.cyan('compliance')}/${s.cyan('review')}/`,
+    `  ${s.cyan('standup')}/${s.cyan('pr')}/${s.cyan('stats')}/${s.cyan('history')}/${s.cyan('collisions')}/${s.cyan('ledger')}/${s.cyan('note add')}/${s.cyan('note patch')}/`,
+    `  ${s.cyan('note delete')}/${s.cyan('recall')}/${s.cyan('comment')}/${s.cyan('transition')}/${s.cyan('assign')}/${s.cyan('duplicates')}/${s.cyan('link')}/`,
+    `  ${s.cyan('update')}/${s.cyan('create')} above.`,
     `  ${s.cyan('fetch')}, ${s.cyan('doctor')}, and ${s.cyan('standup')} are Free; ${s.cyan('triage')} is Free with some Pro/Team-gated`,
     `  options (see ${s.cyan('ticketlens triage --help')}); ${s.cyan('compliance')} and ${s.cyan('pr')} are Free, sharing a`,
     `  3-checks/month cap on their requirements-coverage section, Pro unlimited;`,
@@ -666,6 +669,12 @@ export function printMcpHelp({ stream = process.stdout } = {}) {
     `  ${s.cyan('ticket_update')} has no priority field on GitHub and can partially succeed;`,
     `  ${s.cyan('ticket_create')} has no ticket key to target — --profile/the default profile picks the`,
     `  tracker, and it fabricates a real item, the highest blast radius of this family.`,
+    `  ${s.cyan('ledger')} exports the local, signed compliance audit trail — entirely local, no`,
+    `  network call. ${s.cyan('recall_update')} overwrites an existing Recall note's body — internal`,
+    `  plumbing for the note quality loop, not typically called directly.`,
+    `  ${s.cyan('recall_delete')} is destructive and local-vault-only — requires \`confirm: true\`,`,
+    `  alongside \`id\`, to actually execute; there is no interactive y/N prompt under`,
+    `  MCP, so omitting it always fails rather than silently blocking.`,
     `  Long-running — exits when the client closes stdin.`,
     '',
     `  ${s.dim('If a tool call rejects a parameter you expect right after upgrading ticketlens,')}`,

package/skills/jtb/scripts/lib/mcp-server.mjs

@@ -22,7 +22,7 @@
 import readline from 'node:readline';
 import { DEFAULT_CONFIG_DIR, getVersion } from './config.mjs';
 import { runDoctor } from './doctor-command.mjs';
-import { runNoteAdd } from './note-command.mjs';
+import { runNoteAdd, runNotePatch, runNoteDelete } from './note-command.mjs';
 import { runRecall } from './recall-command.mjs';
 import { runTicketComment, runTicketTransitionList, runTicketTransition, runTicketAssign, runTicketDuplicates, runTicketLinkList, runTicketLink, runTicketUpdate, runTicketCreate } from './ticket-command.mjs';
 import { run as runFetchTicket } from '../fetch-ticket.mjs';
@@ -239,6 +239,25 @@ async function callPr(args, deps) {
   return callFetchTicketRun(buildPrArgs, args, deps, 'pr failed');
 }
 
+/**
+ * `ledger` is a subcommand of the same fetch-ticket.mjs `run()` that
+ * `callFetch`/`callCompliance`/`callPr` already wrap — reuses `runFetchTicketFn`,
+ * no new dependency. Unlike those, it has no `ticket`/`profile` argument — the
+ * ledger is local and config-dir scoped, not per-ticket. Its two direct
+ * `process.stderr` writes (the license-gate upgrade prompt and the
+ * verify-signature note printed alongside a successful json export) were
+ * threaded through opts.printErr as part of adding this tool.
+ */
+function buildLedgerArgs({ format }) {
+  const args = ['ledger'];
+  if (format) args.push(`--format=${format}`);
+  return args;
+}
+
+async function callLedger(args, deps) {
+  return callFetchTicketRun(buildLedgerArgs, args, deps, 'ledger export failed');
+}
+
 function buildDoctorArgs({ fix, profile }) {
   const args = ['--format=json'];
   if (fix === true) args.push('--fix');
@@ -355,6 +374,68 @@ async function callRecallAdd(args, { configDir, runNoteAddFn }) {
   return written ? { content } : { isError: true, content };
 }
 
+/**
+ * Builds runNotePatch's cmdArgs array — same single-opaque-element reasoning
+ * as buildNoteAddArgs above. `expectMtime` is optimistic-concurrency: a
+ * caller that fetched a note via recall_search and wants to refine it
+ * without racing a concurrent edit passes back the mtime it observed.
+ */
+function buildNotePatchArgs({ id, ticket, expectMtime }) {
+  const args = [`--id=${id}`];
+  if (ticket) args.push(`--ticket=${ticket}`);
+  if (expectMtime !== undefined) args.push(`--expect-mtime=${expectMtime}`);
+  return args;
+}
+
+async function callRecallUpdate(args, { configDir, runNotePatchFn }) {
+  if (!args.id) {
+    return { isError: true, content: [{ type: 'text', text: 'Missing required argument: id' }] };
+  }
+  if (!args.body) {
+    return { isError: true, content: [{ type: 'text', text: 'Missing required argument: body' }] };
+  }
+  const capture = capturingStream();
+  const { patched } = await runNotePatchFn(buildNotePatchArgs(args), {
+    configDir,
+    stream: capture,
+    readStdin: async () => args.body,
+  });
+  const content = [{ type: 'text', text: capture.text }];
+  return patched ? { content } : { isError: true, content };
+}
+
+/**
+ * Builds runNoteDelete's cmdArgs array — same single-opaque-element reasoning
+ * as buildNoteAddArgs above. Always passes `--yes`: runNoteDelete's own
+ * confirmDestructive gate refuses outright in non-interactive mode (no real
+ * TTY exists under the MCP transport to prompt against), so without it every
+ * call would fail. The caller's `confirm: true` is enforced in callRecallDelete
+ * below instead, before runNoteDeleteFn is ever reached — same nudge-and-audit
+ * -trail spirit as ticket_transition/ticket_link, but deliberately a different
+ * code path: those two defer the refusal to the wrapped CLI function, which
+ * here would surface as that generic non-interactive error instead of a
+ * dedicated one naming `confirm`.
+ */
+function buildNoteDeleteArgs({ id, ticket }) {
+  const args = [`--id=${id}`];
+  if (ticket) args.push(`--ticket=${ticket}`);
+  args.push('--yes');
+  return args;
+}
+
+async function callRecallDelete(args, { configDir, runNoteDeleteFn }) {
+  if (!args.id) {
+    return { isError: true, content: [{ type: 'text', text: 'Missing required argument: id' }] };
+  }
+  if (args.confirm !== true) {
+    return { isError: true, content: [{ type: 'text', text: 'Deletion requires confirm: true — this cannot be restored.' }] };
+  }
+  const capture = capturingStream();
+  const { deleted } = await runNoteDeleteFn(buildNoteDeleteArgs(args), { configDir, stream: capture });
+  const content = [{ type: 'text', text: capture.text }];
+  return deleted ? { content } : { isError: true, content };
+}
+
 async function callRecallSearch(args, { configDir, runRecallFn }) {
   const capture = capturingStream();
   const { ok } = await runRecallFn([args.query ?? ''], {
@@ -532,11 +613,14 @@ async function handleToolsCall(params, deps) {
   if (name === 'review') return callReview(args, deps);
   if (name === 'standup') return callStandup(args, deps);
   if (name === 'pr') return callPr(args, deps);
+  if (name === 'ledger') return callLedger(args, deps);
   if (name === 'doctor') return callDoctor(args, deps);
   if (name === 'stats') return callStats(args, deps);
   if (name === 'history') return callHistory(args, deps);
   if (name === 'collisions') return callCollisions(args, deps);
   if (name === 'recall_add') return callRecallAdd(args, deps);
+  if (name === 'recall_update') return callRecallUpdate(args, deps);
+  if (name === 'recall_delete') return callRecallDelete(args, deps);
   if (name === 'recall_search') return callRecallSearch(args, deps);
   if (name === 'ticket_comment') return callTicketComment(args, deps);
   if (name === 'ticket_transition') return callTicketTransition(args, deps);
@@ -548,7 +632,7 @@ async function handleToolsCall(params, deps) {
   return { isError: true, content: [{ type: 'text', text: `Unknown tool: ${name}` }] };
 }
 
-async function handleMessage(raw, { configDir, runFetchTicketFn, runTriageFn, runDoctorFn, runStatsFn, runHistoryFn, runCollisionsFn, runNoteAddFn, runRecallFn, runTicketCommentFn, runTicketTransitionListFn, runTicketTransitionFn, runTicketAssignFn, runTicketDuplicatesFn, runTicketLinkListFn, runTicketLinkFn, runTicketUpdateFn, runTicketCreateFn }) {
+async function handleMessage(raw, deps) {
   let msg;
   try {
     msg = JSON.parse(raw);
@@ -578,7 +662,7 @@ async function handleMessage(raw, { configDir, runFetchTicketFn, runTriageFn, ru
 
   if (method === 'tools/call') {
     try {
-      const result = await handleToolsCall(params, { configDir, runFetchTicketFn, runTriageFn, runDoctorFn, runStatsFn, runHistoryFn, runCollisionsFn, runNoteAddFn, runRecallFn, runTicketCommentFn, runTicketTransitionListFn, runTicketTransitionFn, runTicketAssignFn, runTicketDuplicatesFn, runTicketLinkListFn, runTicketLinkFn, runTicketUpdateFn, runTicketCreateFn });
+      const result = await handleToolsCall(params, deps);
       return jsonRpcResult(id, result);
     } catch (err) {
       return jsonRpcError(id ?? null, -32603, `Internal error: ${err.message}`);
@@ -606,6 +690,8 @@ export function runMcpServer({
   runHistoryFn = runHistory,
   runCollisionsFn = runCollisions,
   runNoteAddFn = runNoteAdd,
+  runNotePatchFn = runNotePatch,
+  runNoteDeleteFn = runNoteDelete,
   runRecallFn = runRecall,
   runTicketCommentFn = runTicketComment,
   runTicketTransitionListFn = runTicketTransitionList,
@@ -624,6 +710,11 @@ export function runMcpServer({
   stdin.on('error', () => {});
   stdout.on('error', () => {});
 
+  // Assembled once and passed straight through handleMessage to handleToolsCall,
+  // which is the only place the individual functions are read — so a new tool
+  // needs its dependency named here and in the parameter list above, nowhere else.
+  const deps = { configDir, runFetchTicketFn, runTriageFn, runDoctorFn, runStatsFn, runHistoryFn, runCollisionsFn, runNoteAddFn, runNotePatchFn, runNoteDeleteFn, runRecallFn, runTicketCommentFn, runTicketTransitionListFn, runTicketTransitionFn, runTicketAssignFn, runTicketDuplicatesFn, runTicketLinkListFn, runTicketLinkFn, runTicketUpdateFn, runTicketCreateFn };
+
   const rl = readline.createInterface({ input: stdin, terminal: false });
   let queue = Promise.resolve();
 
@@ -635,7 +726,7 @@ export function runMcpServer({
     // never resolving (a dropped rejection isn't a resolution) — the
     // server would hang on shutdown instead of exiting.
     queue = queue.then(async () => {
-      const response = await handleMessage(line, { configDir, runFetchTicketFn, runTriageFn, runDoctorFn, runStatsFn, runHistoryFn, runCollisionsFn, runNoteAddFn, runRecallFn, runTicketCommentFn, runTicketTransitionListFn, runTicketTransitionFn, runTicketAssignFn, runTicketDuplicatesFn, runTicketLinkListFn, runTicketLinkFn, runTicketUpdateFn, runTicketCreateFn });
+      const response = await handleMessage(line, deps);
       if (response) stdout.write(response);
     }).catch(() => {});
   });

package/skills/jtb/scripts/lib/mcp-tool-schemas.mjs

@@ -88,6 +88,16 @@ export const TOOLS = [
       required: ['ticket'],
     },
   },
+  {
+    name: 'ledger',
+    description: 'Export the local compliance ledger — a signed, tamper-evident record of every compliance check run (ticket key, commit SHA, author, timestamp, coverage %). Read-only, entirely local — no network call. Verifiable offline: the json format includes an HMAC-SHA256 signature over {records, exportedAt}, keyed at a local ledger-key file. Requires a TicketLens Pro license.',
+    inputSchema: {
+      type: 'object',
+      properties: {
+        format: { type: 'string', enum: ['json', 'csv'], description: 'Output shape: "json" (default) includes the HMAC signature; "csv" is a flat export with no signature.' },
+      },
+    },
+  },
   {
     name: 'stats',
     description: 'Show response-time and triage-cadence metrics from local triage history — average/median response time, clear rate, triage run count, current urgency breakdown. Read-only, entirely local — no network call. Free tier: 7-day lookback max; TicketLens Pro extends it to 30 days.',
@@ -147,6 +157,33 @@ export const TOOLS = [
       required: ['title', 'body'],
     },
   },
+  {
+    name: 'recall_update',
+    description: 'Overwrite an existing Recall note\'s body with a better draft — internal plumbing for the jtb skill\'s note quality loop, not typically called directly. The new body gets the same structural and secret-scan checks a user-typed body gets. Local vault only. Requires a TicketLens Pro license.',
+    inputSchema: {
+      type: 'object',
+      properties: {
+        id: { type: 'string', description: 'Note id to patch, as printed by recall_add or recall_search.' },
+        body: { type: 'string', description: 'The replacement note body.' },
+        ticket: { type: 'string', description: 'Ticket key the note is about, e.g. PROJ-123. Optional — narrows the search when omitted, the vault is searched across all ticket prefixes.' },
+        expectMtime: { type: 'number', description: 'The note file\'s mtime (ms) last observed by the caller. If the note changed since then, the patch is a no-op rather than an overwrite — optimistic concurrency, not a hard requirement.' },
+      },
+      required: ['id', 'body'],
+    },
+  },
+  {
+    name: 'recall_delete',
+    description: 'Delete a Recall note from the local vault. Irreversible locally, and destructive — requires `confirm: true` alongside `id` to actually execute; there is no interactive y/N prompt under MCP (no real terminal exists to prompt against), so omitting confirm always fails rather than silently blocking. Local vault only — if this note was already pushed to a team, teammates who pulled it keep their copy; deleting it there too is a manager action from the Console (Admin > Recall). Requires a TicketLens Pro license.',
+    inputSchema: {
+      type: 'object',
+      properties: {
+        id: { type: 'string', description: 'Note id to delete, as printed by recall_add or recall_search.' },
+        ticket: { type: 'string', description: 'Ticket key the note is about, e.g. PROJ-123. Optional — narrows the search when omitted, the vault is searched across all ticket prefixes.' },
+        confirm: { type: 'boolean', description: 'Must be true to actually execute the deletion — a nudge and audit trail, not just a formality. Omitted or false always fails.' },
+      },
+      required: ['id'],
+    },
+  },
   {
     name: 'recall_search',
     description: 'Search saved Recall notes by free-text query or ticket key. Requires a TicketLens Pro license.',