@llamaventures/cli 1.22.0 → 1.23.0

package/AGENT_BRIEFING.md

@@ -66,6 +66,7 @@ A teammate says "I just met them and heard…" or pastes a chunk of notes. Your
 2. **Split what they gave you into two kinds** — this is the whole data model:
    - **A source packet with 2+ facts, or facts + a note → ingest once.** Build a JSON object with `source`, `facts`, and optional `note`, then run `llama deal ingest <dealId> --file <packet.json>`. This is the preferred agent path: one atomic, deduplicated, retry-safe commit. Reuse an explicit `idempotencyKey` when retrying the same material. The matching MCP tool is `deal_ingest` and uses the same packet shape.
    - **Verifiable claims → facts.** `llama deal fact add <dealId> --category <cat> --claim "…" --source "<where it came from>" --source-url <url>`. A claim someone *relayed* ("their ARR is $3M", "raised from a16z") is a fact at **unverified** trust — it's hearsay until checked. Pass `--attested` ONLY if you actually verified it against a source yourself. In raw API terms, the fact text field is `claim` (`value` is only a compatibility alias), `source` is the human-readable provenance label, and `sourceUrl` is the canonical URL.
+     - **Trust and contest are separate axes.** A *contested* fact is excluded from what the Deal Agent treats as current, whatever its trust rung — so a fact can read as both `human-vouched` and suppressed. `fact verify --status confirmed` does **not** lift a contest; the response tells you when one is still standing. To lift one: `llama deal fact uncontest <dealId> <factId> --reason "…"`. The lifted contest is kept in the fact's history, not erased.
    - **Their judgment / impression → a note.** `llama post <dealId> "…"`. "Founder seemed evasive", "I'd lean pass", "worth a second meeting" — opinion, not fact. Attributed, never "verified".
    - A pasted blob → pull the verifiable claims out as facts, capture their take as a note.
 3. **Read it back before you claim it's saved.** A generic tool call returning `{ok:true}` is NOT proof the content is where the user will look for it. `deal ingest` is the exception because its response is built from the rows and note block read back after commit: confirm `createdFacts`, `skippedFacts`, `note`, and `summary` in that receipt. For every other write, run `llama deal feed <dealId>` and confirm your fact/note actually appears. Never say "记好了 / saved" from a request acknowledgment alone — the #1 failure is an agent writing to the wrong surface (e.g. the brief, which is the Memo and does NOT appear in the feed) and reporting success anyway. If the authoritative ingest receipt or the feed read-back does not contain the material, fix it before reporting success.
@@ -217,6 +218,7 @@ llama deal create "Company" --description "..." --source-direction Outbound --st
 llama deal create "Company" --description "..." --source-direction Outbound --status Outreached --deal-owner "owner@llamaventures.vc"
 llama deal create "Company" --description "..." --source-direction Inbound --status Sourced --deal-owner "Exact Name"
 llama deal update <dealId> <field> <value>
+llama deal fact uncontest <dealId> <factId> --reason "..."   # lift a wrong contest; verify does NOT
 #   writable: status theirStage stage notes dealOwner source sourceDirection description website
 #             location founders proposedAmount roundSize valuation sector subsector
 #             foundedYear leadInvestor investors   (each write logs a deal_events row)

package/CHANGELOG.md

@@ -6,6 +6,43 @@ this project adheres to [Semantic Versioning](https://semver.org).
 
 ## [Unreleased]
 
+## [1.23.0] - 2026-08-05
+
+### Added
+- `llama deal fact uncontest <dealId> <factId> --reason "..."` lifts a contest
+  that turned out to be wrong. Contesting a fact removes it from what the Deal
+  Agent treats as current, and until now nothing could put it back — so a true
+  claim contested on weak evidence stayed suppressed in every regenerated brief
+  and memo. The lifted contest is kept in the fact's history server-side, not
+  erased. Requires Llama Command >= 3.21.0.
+
+### Changed
+- `--help` and `AGENT_BRIEFING.md` now state that **trust and contest are
+  separate axes**: `fact verify --status confirmed` does NOT lift a contest, and
+  a contested fact stays excluded from synthesis whatever its trust rung. The
+  server now returns a warning when a verify leaves a contest standing.
+- `deal update` help says what `notes` is for — the one-line Summary headline at
+  the top of the deal page (~280 chars), not a place for meeting notes. It was
+  previously listed among 20 field names with no semantics, so the honest
+  reading was "notes" and long text landed in a headline slot.
+
+### Note
+This release also carries the unpublished 1.22.2 fix below.
+
+## [1.22.2] - 2026-08-04
+
+### Fixed
+- `llama claim` and `llama nominate` now require `--reason` and send it to
+  Llama Command's audited Owner-label endpoint. The CLI no longer emits a
+  request that Core must reject before recording the responsibility change.
+
+## [1.22.1] - 2026-08-04
+
+### Changed
+- Refresh the transitive `ip-address` dependency from 10.2.0 to 10.4.0 and
+  retain the same CLI and MCP behavior across the supported Node 18/20/22
+  matrix.
+
 ## [1.22.0] - 2026-07-28
 
 ### Added

package/bin/llama-mcp.mjs

@@ -457,8 +457,7 @@ server.registerTool(
   {
     description:
       "Create a new pipeline deal. Source defaults to the caller's user record. " +
-      "Owner assignment goes through the partner-approval queue (status: pending) " +
-      "until a partner approves it via /partner/approvals.",
+      "Owner is an audited responsibility label, not an operating-permission boundary.",
     inputSchema: {
       companyName: z.string(),
       description: z.string().optional().describe("one-liner: what they do"),

package/bin/llama.mjs

@@ -395,6 +395,9 @@ Deals:
       description, website, location, founders, founderInfo, proposedAmount,
       roundSize, valuation, deckLink, folderUrl, sector, subsector,
       foundedYear, leadInvestor, investors, agentActive.
+      'notes' is the ONE-LINE judgment shown as the Summary headline at the top of
+      the deal page (~280 chars). Meeting notes and narrative go in a comment
+      (llama post); verifiable claims go in facts (llama deal fact add).
       e.g.  llama deal update <dealId> website https://acme.ai
             llama deal update <dealId> status Interested
             llama deal update <dealId> sector "Developer Tools"
@@ -446,12 +449,12 @@ Deal links (separate from brief link blocks — these live in deal_links):
   llama deal link restore <dealId> <linkId>
 
 Ownership:
-  llama claim <dealId>                                       # propose self as owner
-  llama nominate <dealId> --user <userId>                    # partner nominates someone else
-  llama nominations list                                     # pending nominations for me
-  llama nominations decide <approvalId> accepted|declined    # accept/decline a nomination
+  llama claim <dealId> --reason "..."                        # set yourself as Owner
+  llama nominate <dealId> --user <userId> --reason "..."     # set another teammate as Owner
+  Owner is an audited responsibility label, not a permission boundary. Any
+  internal teammate can change it after Partner Support; reason is mandatory.
 
-Approvals (partner queue — self-claim approvals):
+Approvals (Partner queue — contribution decisions):
   llama approvals list
   llama approvals decide <approvalId> approved|rejected [--note "..."]
 
@@ -503,6 +506,9 @@ Deal facts (AI-extracted or human-asserted, with verification):
   llama deal fact list <dealId>
   llama deal fact add <dealId> --category <cat> --claim "<text>" [--source "..."] [--source-url <url>] [--confidence high|medium|low] [--attested]
   llama deal fact verify <dealId> <factId> --status confirmed|disputed [--corrected-value "..."]
+  llama deal fact uncontest <dealId> <factId> --reason "<why the contest was wrong>"
+    contest and trust are separate axes: 'verify --status confirmed' does NOT lift a contest,
+    and a contested fact stays excluded from what the Deal Agent treats as current.
 
 Mentions / Inbox:
   llama mentions                                       # default: my unresolved cues
@@ -625,7 +631,7 @@ Command groups — run \`llama help <group>\` for that group's commands:
   deal        create · show · feed · update · enrich · search · collaborators · links · delete
   activity    new-deals · updated-deals · events for agent read models
   brief       brief blocks: list · add · edit · history · refresh
-  facts       deal facts + skill corrections (the sourced, trust-rated layer)
+  facts       deal facts — the sourced, trust-rated layer
   timeline    timeline · posts · mentions
   wiki        cross-deal knowledge entries (markdown or HTML)
   pref        standing agent preferences: list · add · retire · approve
@@ -645,10 +651,20 @@ the CLI auto-detects it — no token needed (\`llc_\` tokens are a fallback).`;
 
 // Area → which top-level sections of HELP_FULL belong to it.
 const HELP_AREA_MATCH = {
-  deal: [/^Deals/, /^Collaborators/, /^Soft-delete/, /^Deal links/, /^Deal soft-delete/],
+  deal: [
+    /^Deals/,
+    /^Collaborators/,
+    /^Soft-delete/,
+    /^Deal links/,
+    /^Deal soft-delete/,
+    // `llama deal refresh-brief` / `revert-run` live under this heading. Without
+    // it here the section matched no area and was reachable only via `help all`,
+    // which read as "the command does not exist".
+    /^Brief refresh/,
+  ],
   activity: [/^Agent activity/],
-  brief: [/^Brief blocks/],
-  facts: [/^Deal facts/, /^Skill corrections/],
+  brief: [/^Brief blocks/, /^Brief refresh/],
+  facts: [/^Deal facts/],
   timeline: [/^Timeline/, /^Mentions/],
   wiki: [/^Wiki/, /^Where does this HTML/],
   memo: [/^Memo/],
@@ -1816,7 +1832,29 @@ async function main() {
       return;
     }
 
-    throw new Error(`Unknown fact sub-command "${sub}". Use list, add, or verify.`);
+    // Lift a contest that turned out to be wrong. Contesting a claim removes
+    // it from everything the Deal Agent treats as current; before this existed
+    // there was no way back, so a TRUE claim contested on weak evidence stayed
+    // suppressed in every regenerated brief and memo. `verify --status
+    // confirmed` does NOT do this — trust and contest are separate axes.
+    if (sub === "uncontest") {
+      const factId = rest[2];
+      const reason = flags.reason !== undefined && flags.reason !== true ? String(flags.reason).trim() : "";
+      if (!factId || !reason) {
+        throw new Error(
+          `Usage: llama deal fact uncontest <dealId> <factId> --reason "<why the contest was wrong>"`
+        );
+      }
+      // @core-api-operation POST /api/deals/{dealId}/facts/{factId}/uncontest
+      print(await request(
+        "POST",
+        `/api/deals/${encodeURIComponent(dealId)}/facts/${encodeURIComponent(factId)}/uncontest`,
+        { reason }
+      ));
+      return;
+    }
+
+    throw new Error(`Unknown fact sub-command "${sub}". Use list, add, verify, or uncontest.`);
   }
 
   // ----- Brief refresh: trigger stale-section re-eval watcher run -----
@@ -1880,12 +1918,16 @@ async function main() {
   // ----- Ownership: self-claim -----
   if (area === "claim") {
     const dealId = action; // second positional
-    if (!dealId) throw new Error("Usage: llama claim <dealId>");
+    const { flags } = parseFlags(rest, ["reason"]);
+    const reason = typeof flags.reason === "string" ? flags.reason.trim() : "";
+    if (!dealId || !reason) {
+      throw new Error('Usage: llama claim <dealId> --reason "<why this person should own it>"');
+    }
     const me = await request("GET", "/api/me");
     print(await request(
       "POST",
       `/api/deals/${encodeURIComponent(dealId)}/propose-owner`,
-      { userId: me.id }
+      { userId: me.id, reason }
     ));
     return;
   }
@@ -1893,15 +1935,16 @@ async function main() {
   // ----- Ownership: partner nominates someone else -----
   if (area === "nominate") {
     const dealId = action;
-    const { flags } = parseFlags(rest);
+    const { flags } = parseFlags(rest, ["user", "reason"]);
     const userId = Number(flags.user);
-    if (!dealId || !Number.isFinite(userId)) {
-      throw new Error("Usage: llama nominate <dealId> --user <userId>");
+    const reason = typeof flags.reason === "string" ? flags.reason.trim() : "";
+    if (!dealId || !Number.isFinite(userId) || !reason) {
+      throw new Error('Usage: llama nominate <dealId> --user <userId> --reason "<why>"');
     }
     print(await request(
       "POST",
       `/api/deals/${encodeURIComponent(dealId)}/propose-owner`,
-      { userId }
+      { userId, reason }
     ));
     return;
   }

package/contracts/required-operations.json

@@ -186,6 +186,10 @@
       "method": "PATCH",
       "path": "/api/deals/{dealId}/facts/{factId}"
     },
+    {
+      "method": "POST",
+      "path": "/api/deals/{dealId}/facts/{factId}/uncontest"
+    },
     {
       "method": "GET",
       "path": "/api/deals/{dealId}/feed"

package/lib/build-manifest.json

@@ -1,8 +1,8 @@
 {
   "format": "llama.cli-build.v1",
   "packageName": "@llamaventures/cli",
-  "packageVersion": "1.22.0",
-  "sourceSha": "3e3ede496804de031650e23fc29e1a6763c41daf",
+  "packageVersion": "1.23.0",
+  "sourceSha": "3f3fb4ae90469e4c561bc8ad319c59d877d0fcf2",
   "sourceKind": "github",
   "sourceDirty": false,
   "coreApiContract": {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llamaventures/cli",
-  "version": "1.22.0",
+  "version": "1.23.0",
   "description": "CLI + MCP server for the Llama Ventures investment workbench (command.llamaventures.vc).",
   "type": "module",
   "scripts": {