@moda-ai/cli 1.25.0 → 1.27.0

package/dist/{harness-github-actions-rjyc9a7z.js → harness-github-actions-ct8fwzdr.js}

@@ -1,7 +1,7 @@
 import {
   buildSourceSnapshot
-} from "./cli-jntzpem2.js";
-import"./cli-vc7zddj1.js";
+} from "./cli-g931ntph.js";
+import"./cli-59yacef3.js";
 
 // src/harness-github-actions.ts
 var DEFAULT_MODA_GITHUB_URL = "https://moda-github.modas.workers.dev";

package/dist/{index-fm9ey3ks.js → index-fwxpt51s.js}

@@ -9,7 +9,7 @@ import {
   runPromptSync,
   runSkillSync,
   upsertSkillManifestRecord
-} from "./cli-kqvjt12y.js";
+} from "./cli-cjqzrp9k.js";
 import {
   codingAgentDisplayName,
   describeCodingAgentEvent,
@@ -27,14 +27,14 @@ import {
   runHarnessCommand,
   startRemoteAnalyze,
   stripAnsi
-} from "./cli-jntzpem2.js";
+} from "./cli-g931ntph.js";
 import {
   selectTenantAndCreateKey
-} from "./cli-tdttb4cv.js";
+} from "./cli-pq6rte0w.js";
 import {
   loadToken,
   login
-} from "./cli-8s69sg7f.js";
+} from "./cli-9xw5bq1j.js";
 import {
   CliAuthError,
   CliInputError,
@@ -42,7 +42,7 @@ import {
   createCommandContext,
   resolveIngestUrl,
   resolveModaBaseUrl
-} from "./cli-vc7zddj1.js";
+} from "./cli-59yacef3.js";
 
 // src/init/index.ts
 import { existsSync as existsSync6, mkdirSync as mkdirSync5, readdirSync as readdirSync4, readFileSync as readFileSync5, writeFileSync as writeFileSync4 } from "node:fs";

package/dist/{provision-dphktm75.js → provision-r0cq3g2v.js}

@@ -1,15 +1,15 @@
 import {
   selectTenantAndCreateKey
-} from "./cli-tdttb4cv.js";
+} from "./cli-pq6rte0w.js";
 import {
   loadAuthSession
-} from "./cli-8s69sg7f.js";
+} from "./cli-9xw5bq1j.js";
 import {
   CliAuthError,
   resolveIngestUrl,
   resolveModaBaseUrl,
   stringOption
-} from "./cli-vc7zddj1.js";
+} from "./cli-59yacef3.js";
 
 // src/provision.ts
 async function runProvision(context, options = {}) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@moda-ai/cli",
-  "version": "1.25.0",
+  "version": "1.27.0",
   "description": "CLI for Moda - AI agent analytics and observability",
   "type": "module",
   "bin": {

package/skills/integration/index.json

@@ -1,7 +1,7 @@
 {
   "schema_version": "moda.skill_index.v1",
-  "bundled_at": "2026-07-28T04:14:38.909Z",
-  "cli_version": "1.25.0",
+  "bundled_at": "2026-08-17T23:28:03.271Z",
+  "cli_version": "1.27.0",
   "skills": [
     {
       "id": "integration-node-anthropic",

package/skills/moda-cli/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: moda-cli
-version: 2.2.0
+version: 2.3.0
 description: Query Moda's AI agent conversation analytics and manage code-first prompt versions from the terminal — semantic/keyword/hybrid message search, overview KPIs, topic clusters, message context, user frustration detections, tool failures, and moda prompts status/sync/promote. Use when the user asks about moda, modaflows, conversation analytics, prompt management, user frustrations, agent observability, tool failure debugging, wants to find conversations or tool calls about a topic, or wants to investigate how their AI agent is performing.
 ---
 
@@ -16,7 +16,7 @@ or promote remote state (`prompts sync`, `prompts promote`).
 
 **`moda search` is the primary way to find anything by content.** It runs
 message-grain semantic, keyword, or hybrid retrieval across every message —
-including tool calls and tool results — and returns ranked snippets with
+including tool calls and tool results with `--include-tool-io` — and returns ranked snippets with
 direct anchors (`conversation_id` + `message_index`) you can hand straight to
 `moda context`. Reach for it first whenever the question is "where did X
 happen?" or "find conversations/tool calls about Y". Use `moda conversations`
@@ -55,9 +55,9 @@ Optional:
 
 - `MODA_BASE_URL` defaults to `https://moda.dev`; override only for
   self-hosted or staging.
-- `MODA_SKILL_VERSION` — export to `2.0.0` so search-adoption telemetry can
+- `MODA_SKILL_VERSION` — export to `2.3.0` so search-adoption telemetry can
   attribute usage to this skill version. Set it once per session:
-  `export MODA_SKILL_VERSION=2.0.0`.
+  `export MODA_SKILL_VERSION=2.3.0`.
 
 ## Setup
 
@@ -391,9 +391,10 @@ moda search "checkout error" --time-range=7d --limit=10
 moda search "stripe.charges.create failed" --user-id=<id>
 ```
 
-Searches conversation messages at message grain, **including tool calls and
-tool results** (tool name, input arguments, and output previews are embedded),
-so it finds where an agent *did* something, not just where it talked about it.
+Searches conversation messages at message grain. Pass `--include-tool-io` to
+also search **tool calls and tool results** (tool name, input arguments, and
+output previews), so it finds where an agent *did* something, not just where
+it talked about it — tool-IO search is opt-in, not the default.
 
 Flags: `--mode=keyword|semantic|hybrid` (default `hybrid`), `--user-id`,
 `--time-range` (`all|1h|3d|7d|24h|30d|90d`, default `all`), `--limit` (1–100,
@@ -613,6 +614,56 @@ Do not copy prompt text from the dashboard into source code. The repo prompt
 files are the source of truth; the dashboard is for visibility, usage, labels,
 and future evals.
 
+### 8. Fixing detected Problems (`moda fix` — a fix with proof)
+
+Moda turns detected Problems into **Fixes**: a routed candidate change plus a
+replay **gate** that proves it on held-out production evidence. Use this loop
+whenever you are asked to fix a detected Problem, or before hand-editing
+prompt/skill files in a repo that talks to Moda.
+
+The fail-to-pass loop:
+
+```bash
+moda fixes                                   # ranked queue of fixes with proof
+moda fix <fix_id> --packet                   # moda.fix_packet.v1: cause, target
+                                             # file, evidence links, verify cmd,
+                                             # handback + magic word
+moda fix start <problem_id> --wait           # draft from a Problem and drive
+                                             # scope -> propose -> gate
+moda fix checkout <fix_id>                   # write the candidate at its target
+                                             # source path (NEVER runs git)
+# ...edit the target file...
+moda fix verify <fix_id> --prompt-file=<path>  # re-gate local content on the
+                                               # frozen holdout
+# iterate edit -> verify until exit 0, then hand back:
+moda fix submit <fix_id> --pr                # draft PR via the Moda GitHub App
+moda fix submit <fix_id> --local-ref=<branch>  # or record your own branch
+moda fix dismiss <fix_id> --reason="why"     # reject (feeds problem feedback)
+```
+
+**`moda fix verify` exit codes are the contract — branch on them:**
+
+| Exit | Meaning | How to treat it |
+|---|---|---|
+| `0` | gate pass — the candidate beat baseline on the frozen holdout (wins clear the noise floor) | Safe to submit. |
+| `1` | gate fail — per-case findings ride the envelope's `findings[]` (`case:<id>` entries with baseline/candidate pass) | Edit the candidate and re-`verify`. |
+| `3` | degraded/inconclusive — too many abstentions, blocked coverage, or no verdict | **Never treat as a pass.** Re-verify or read `moda fix <fix_id> --packet`. |
+
+Rules that keep the proof honest:
+
+- The verdict reads only **holdout** cases the proposer never saw; repair
+  rows are training cases, never the headline.
+- The gate compares against a noise floor from a baseline-vs-baseline control
+  run — a "win" inside the noise floor does not pass.
+- `checkout` only writes the file. Branching (`moda/fix/<shortref-lower>`),
+  committing, and pushing are yours; the CLI never runs git.
+- Fix PRs end with the magic word `Fixes MODA-FIX-<SHORTREF>` in the body —
+  Moda-authored PRs carry it automatically; hand-authored PRs must include it
+  so the merge confirms the fix and starts production monitoring.
+- The pipeline is advance-on-poll: `--wait` (on `start`, `verify`, or
+  `moda fix <fix_id> --wait`) drives it; a fix left `GATING` will not finish
+  on its own.
+
 ## Common workflow recipes
 
 ### Find where something happened (search → context)
@@ -710,9 +761,10 @@ moda feedback "a cancelled call is counted as a tool failure" \
 ## Validation — how to know it worked
 
 - Every successful command exits 0 and prints valid JSON to stdout.
-- `moda ask` / `moda investigate` may exit `3` (degraded): a real,
-  lower-trust answer synthesized from local evidence — not a failure. Treat
-  exit `1` as the only hard error (see [Agent protocol](#agent-protocol)).
+- `moda ask` may exit `3` (degraded): a real, lower-trust answer synthesized
+  from local evidence — not a failure (`investigate`/`failures` always exit
+  `0` on success). Treat exit `1` as the only hard error (see
+  [Agent protocol](#agent-protocol)).
 - Errors print to stderr with non-zero exit. Check stderr before claiming
   success.
 - If the JSON has `"conversations": []` or `"frustrations": []`, the call
@@ -768,18 +820,34 @@ npx: `npx -p @moda-ai/cli moda <command>`.
 | `moda overview` | Dashboard KPIs, top clusters, recent activity |
 | `moda ask "<question>"` | Natural-language production/harness answer with evidence (exit `3` = degraded local fallback) |
 | `moda investigate` | Rank production issues with evidence + next commands |
-| `moda clusters` | Browse topic cluster hierarchy |
+| `moda clusters` | Browse topic cluster hierarchy; `--search="q"` finds clusters by meaning, `--node-id=ID` resolves a deep link |
 | `moda cluster-conversations <node_id>` | Conversations in a cluster |
 | `moda conversations` | List/filter conversations by structured fields |
 | `moda context <conversation_id>` | Windowed message context (max 5 per side) |
-| `moda frustrations` | User frustration detections with evidence |
+| `moda frustrations` | User frustration detections with evidence (legacy single-family; prefer `emotions`) |
+| `moda emotions` | Multi-family emotion detections: frustration, sadness, confusion, anxiety, trust, positive (`--family=F`, limit 1–20) |
+| `moda hallucinations` | Grounding detections: contradicted/verified outputs, rule breakdown (`--kind=contradicted\|verified`, `--conversation-id=ID`) |
 | `moda tool-failures` | Tool failure overview |
 | `moda tool-failure-detail <tool_name>` | Per-tool failure breakdown + examples |
+| `moda problems` | Rank cross-signal Problems by root cause (what to fix first) |
+| `moda problem <problem_id>` | One Problem: dossier, or `--evidence`/`--reports`/`--conversations`/`--feedback` pages (`--limit` 1–50, `--cursor` verbatim keyset token) |
+| `moda problem-feedback <problem_id>` | Write: `--action=mark_fixed\|dismiss\|flag_attribution\|rename`. `--reason` required for dismiss/flag_attribution; `--new-name` for rename; `--attribution-id` (UUID) required for flag_attribution |
+| `moda step-scores <conversation_id>` | Graph-PRM step scores: per-segment curves, first bad step, rollup |
+| `moda world-state <conversation_id>` | Agent memory: slots/threads/events; `--summary-only`; `--snapshot --msg-index=N` (state at a turn); `--replay --message-count=N` (state over time) |
+| `moda failures` | Production failures worth fixing first |
+| `moda tail` | Live tail: one JSON line per new conversation/detection (`--signal=conversations\|emotions\|all`, `--interval=N`, `--once`, `--max-events=N`). **Emotions caveat:** `/emotions` is ranked by score with no time ordering or cursor, so the tail follows the *highest-scoring* detections rather than everything; each poll emits a `tail_coverage` line with `scanned`/`total`/`coverage_pct`/`complete`. Pass `--full-scan` for a complete window scan (many more requests, capped by the API's offset ceiling of 10000). |
 | `moda feedback "<note>"` | Flag wrong/missing data or CLI quirks to the Moda team |
 | `moda prompts status` | Read-only local prompt status |
 | `moda prompts diff` | Read-only local prompt diff/status |
 | `moda prompts sync` | Upload changed prompt versions and update lockfile |
 | `moda prompts promote <key>` | Move a remote prompt label |
+| `moda fixes` | Ranked queue of Fixes ("a fix with proof"); `--status=S`, `--limit=N`, `--cursor=TOKEN` |
+| `moda fix <fix_id>` | One Fix: status + gate result; `--packet` prints `moda.fix_packet.v1`; `--wait` drives a running pipeline to rest |
+| `moda fix start <problem_id>` | Draft a Fix from a Problem (`--type=auto\|prompt`); `--wait` drives scope → propose → gate |
+| `moda fix verify <fix_id>` | Re-gate on the frozen holdout; `--prompt-file=P` gates local content (exit `0` pass / `1` fail with per-case findings / `3` degraded — never a pass) |
+| `moda fix checkout <fix_id>` | Write the candidate at its target source path (never runs git) |
+| `moda fix submit <fix_id>` | Deliver: `--pr` opens a draft PR \| `--local-ref=<branch>` records your branch |
+| `moda fix dismiss <fix_id>` | Reject with `--reason=TEXT` (feeds problem feedback) |
 
 `moda search` flags: `--mode=keyword|semantic|hybrid` (default `hybrid`),
 `--user-id`, `--time-range=<range>`, `--limit=N` (1–100, default 20).