@moda-ai/cli 1.55.39 → 1.55.41

package/skills/moda-cli/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: moda-cli
-version: 3.0.5
+version: 3.0.7
 description: Investigate a production AI agent with the Moda CLI (`moda`) - find where something happened in real traces (traces search, then traces get), check agent health, rank what to fix (investigate, problems), and dig into tool failures, user frustration, and hallucinations. Also prompt registry sync/promote, replays, custom signals, and false-positive marking. Use when the user mentions Moda, production traces or conversations of their agent, agent failures, or asks what to fix.
 ---
 
@@ -60,13 +60,15 @@ Details: `references/output-protocol.md`.
 | Where did X happen / traces about Y | `moda traces search "<words>"` (`--mode=keyword` for ids, error codes, exact strings; `--include-tool-io` for tool inputs/outputs) | `moda traces get <conversation_id> --msg-index=<message_index> --window=3` |
 | Search returned nothing | Read `.warnings` (degraded mode, reason). Try a shorter phrase, `--mode=keyword`, a wider `--time-range` | `moda traces --search="<words>"` |
 | Is the agent healthy? | `moda overview --days-back=7` | `moda investigate` |
-| What to fix first? | `moda investigate` or `moda problems list` (free; the same list the dashboard shows) | `moda problems get <problem_id>` (`--view=dream` on a dream-grouped tenant), then `--evidence` / `--traces` |
+| What to fix first? | `moda investigate` or `moda problems list` (free; both rank the same list the dashboard shows, `data.problemsView` names it) | `moda problems get <problem_id>` (`--view=dream` on a dream-grouped tenant), then `--evidence` / `--traces` |
 | Why is tool T failing? | `moda detections tools` | `moda detections tools <T>` (use each row's `anchor`) |
 | How often is tool T called / what is its schema? | `moda tools list` (every tool: calls, failed calls, success rate; `--sort=failure_rate`) | `moda tools get <T>` (registered schema, observed payload fields, usage over time, top failure categories) |
 | What frustrates users? | `moda detections list --family=frustration --limit=10` | `moda traces get <anchor.conversation_id> --msg-index=<anchor.msg_index>` |
 | Other emotions, laziness, hallucinations | `moda detections list --family=emotion:<f>`, `moda detections list --family=laziness`, `moda detections list --family=hallucination --kind=contradicted` | context on the row's anchor |
 | Counts that match the dashboard, dashboard filters | `moda detections list --family=<f> --group-by=trace` (`--trajectory`, `--elicitor`, `--signal`, `--rule`, `--sort`, ...) | `references/detections.md` 5a |
 | Is an empty result really clean? | `moda overview` → `detectors` (is the detector on?); `traces` rows → `processing_status` (pending = not fully processed; detections may be incomplete) | `references/detections.md` 5b |
+| Everything flagged on one trace | `moda traces findings <id>` (every detector family plus its Problems, no time window; up to 25 ids, or `--ids-from=-`) | `moda traces get <id> --msg-index=<n> --window=2` from its `next_commands` |
+| Cross-signal question over many traces (e.g. contradictions and frustration) | `moda traces findings <id1> <id2> ...` (all families per trace, 25 per call), or one family over up to 50 traces in one request: `moda detections list --family=hallucination --conversation-ids=<id1>,<id2>,...` (or `--ids-from=-`) | Never loop `--conversation-id` one trace at a time |
 | Read one whole trace | `moda traces get <id> --all` (pages of up to 500 messages) | `moda traces spans <id>` for raw spans |
 | Read several traces or turns | `moda traces get <id1>@<msg_index> <id2>@<msg_index> ...` (up to 25; or one per line via `--ids-from=-`) | Same for `moda traces spans <id1> <id2> ...` |
 | One user's experience | `moda users`, `moda user <user_id>` | `moda traces search "<words>" --user-id=<id>` |
@@ -78,14 +80,16 @@ Details: `references/output-protocol.md`.
 | Prompts, replays, registry push | Only on explicit request | `references/prompts.md`, `references/replays-and-registry.md` |
 
 Time windows differ by command: `--days-back=N` (1-90) on overview,
-investigate, problems, and detection commands; `--time-range=1h|24h|3d|7d|30d|90d|all`
+investigate, problems, and detection commands (`detections list
+--conversation-id` is windowed too, so an older trace reads clean there; use
+`moda traces findings`, which has no window); `--time-range=1h|24h|3d|7d|30d|90d|all`
 on search, traces, and clusters. The other spelling works when it names the
 same span (`--days-back=7` is `--time-range=7d`); anything else fails and
 names the closest values. `moda <noun> <verb> --help` shows which.
 
 ## Fixing a Problem
 
-1. `moda investigate` or `moda problems` to pick one; note its `problem_id`.
+1. `moda investigate` or `moda problems list` to pick one (both show the dashboard's Problems in its order); note its `problem_id`.
 2. `moda problems get <problem_id> --evidence` and open two or three anchors with
    `moda traces get <id> --msg-index=<n>` before deciding on a cause.
 3. Make the change in code; cite the problem id and trace ids in the commit or PR.

package/skills/moda-cli/references/commands.md

@@ -44,6 +44,7 @@ Loaded on demand from the moda-cli skill. This table can lag the CLI: `moda <com
 | `moda signal <signal_id>` | One custom signal; `--detections` pages recent hits (`--days-back` 1–90, `--limit` 1–20, `--offset`) |
 | `moda detections list --family=custom_signal --conversation-id=<conversation_id>` | Custom signal hits on one trace |
 | `moda review custom-signal <signal_id>` | Write: `--conversation-id`, `--verdict=correct\|false_positive\|clear`, optional `--note`, `--msg-index`/`--msg-dedup-token` |
+| `moda traces findings <conversation_id>...` | Everything flagged on a trace, no time window: per family a `state` (`found`/`none`/`not_run`/`unknown`), count and top rows with `msg_index`, plus its Problems; up to 25 ids or `--ids-from`; `--limit` rows per family |
 | `moda traces 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` | Tool failures only, ranked (no frustrations or Problems) |
 | `moda tail` | Live tail: one JSON line per new trace/detection (`--signal=traces\|emotions\|all`, legacy alias `--signal=conversations`, `--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). |

package/skills/moda-cli/references/detections.md

@@ -10,8 +10,37 @@ moda detections list --family=frustration --days-back=14 --limit=20
 moda detections list --family=frustration --include-window --window=1 --limit=5   # agent output omits windows unless asked
 moda detections list --family=emotion:frustration --user-id=<id>        # one end user
 moda detections list --family=emotion --conversation-id=<id>                     # every family, one trace
+moda detections list --family=frustration --conversation-ids=<id1>,<id2>,<id3>    # several traces, one request
+moda detections list --family=hallucination --ids-from=ids.txt --days-back=30     # ids one per line (or - for stdin)
 ```
 
+Several traces: `--conversation-ids=a,b,c` (or a JSON array, or `--ids-from=-|<file>`,
+one id per line) takes up to 50 trace ids on `emotion`, `frustration`, `laziness`
+and `hallucination`, and the server filters them in one request. Every figure
+(summary, list, total) then covers just those traces. `--days-back` still
+applies. For `custom_signal` and `tool_failure` per trace, or for every family
+of a trace with no window, use `moda traces findings <id> ...` (below).
+
+### Everything flagged on one trace
+
+```bash
+moda traces findings <conversation_id>                 # every family + its Problems, no time window
+moda traces findings <id1> <id2> <id3>                 # up to 25 ids, one envelope (or --ids-from=-|<file>)
+moda traces findings <conversation_id> --limit=10      # more rows per family (default 3; counts cover all)
+```
+
+One read per trace returns emotion, laziness, hallucination (contradicted claims),
+tool failure (with its category) and custom signal detections, each with the
+`msg_index` it anchors to, plus the Problems the trace is evidence for (id, title,
+attribution count, `in_book`). There is no time window, so a trace older than
+`--days-back` does not read clean. Each family has a `state`:
+`found`, `none` (its detector ran and found nothing), `not_run` (detector off for
+the workspace, or the trace not processed yet) or `unknown` (nothing found, and the
+detector records nothing for a clean trace, as laziness does). Only `none` means
+clean. `next_commands` open each anchor with
+`moda traces get <id> --msg-index=<n> --window=2`. `--json` returns the full
+response (up to 50 rows per family).
+
 In agent output, `frustrations` and `emotions` rows leave out the embedded
 conversation window (most of each row) unless you pass `--include-window`;
 each row keeps `anchor` and a `context_command` to read the turns around it.

package/skills/moda-cli/references/production-intelligence.md

@@ -55,7 +55,11 @@ Data API fallback. Abbreviated envelope:
 narrows the tool-failure evidence to one tool, and `--trace` (legacy alias
 `--conversation`) scopes only the frustration evidence (overview, tool
 failures and Problems stay tenant-wide). Its findings include the top 1–3
-Problems from `moda problems`, with `moda problems get <id>` as the next command.
+Problems of the list `moda problems list` shows (the dream grouping when the
+dashboard shows one, otherwise the Book), in the same order, with
+`moda problems get <id>` as the next command. `data.problemsView` names the
+view, the window its counts cover, and the `moda problems list` command that
+lists the same Problems.
 `moda failures` is the tool-failures-only view.
 
 After reporting the top behavioral failure and the harness layer it routes to