Skip to content
atif-sql

Debugging guide

View as MarkdownOpen in ChatGPTOpen in ClaudeOpen in Claude CodeOpen in Cursor

atif-sql · Debugging guide

Section titled “atif-sql · Debugging guide”

Something is broken. Where do you look first?

atif-sql is a CLI plus its libraries with no server, no job queue, and no observability platform. That narrows the search surface to a few places, and this guide is organized around them:

  1. The process exit code. EXIT_CODES is one dict (packages/atif-cli/src/atif_cli/errors.py:30-66) and it is the primary diagnostic. Not every key is reachable.
  2. The classified error envelope on stderr: one readable line on a TTY, one JSON object on a pipe (packages/atif-cli/src/atif_cli/output.py:278-285).
  3. The loguru sink. Exactly one, added in main(), stderr, WARNING and up unless ATIF_SQL_LOG_LEVEL sets another level (packages/atif-cli/src/atif_cli/app.py:2556-2559).
  4. The cron refresh log and its marker files, for anything that failed unattended (scripts/atif-sql-refresh.sh:186-188).

The exit-code contract, read from source and confirmed by running the CLI:

code keys meaning
0 ok success
2 empty_session, no_embeddings nothing to work on; not a fault
64 invalid_input, parse_error the caller’s input is malformed; a flag cyclopts refuses exits 64 too
65 catalog_error, validation_error, embedding_mismatch, lake_mismatch the named object cannot be bound, or the lake disagrees with the artifacts
70 runtime_error, sandbox_refused everything else the adapter raises, or a statement kind the query sandbox never runs
77 root_refused query, search, or analyze running as uid 0 without ATIF_SQL_ALLOW_ROOT=1
78 terminal_state, suspicious_scan, extension_missing, lake_unavailable an operator has to act; retrying cannot succeed
127 harbor_missing RETIRED: kept so the table never renumbers; no code path raises it
1 (none) an unhandled exception, outside the taxonomy

One of these codes exists to be told apart from a crash rather than from another code. suspicious_scan: 78 is raised only by materialize (packages/atif-cli/src/atif_cli/app.py:1027-1041), when the source scan finds zero sessions over a non-empty corpus and marking sessions source-removed is refused; it shares 78 with terminal_state because both mean an operator has to act, and carries its own kind string because the remedy is a path rather than a store.

Exit 1 is still a reachable outcome, and it means an unhandled exception: analyze maps no analytics error into the taxonomy (packages/atif-cli/src/atif_cli/app.py:1808-1822 has a finally and no except), and neither does a registration error outside REGISTRATION_ERRORS (packages/atif-cli/src/atif_cli/duck_errors.py:32). Read a 1 as a bug report, never as a diagnosis.

On a pipe the envelope is the last line on stderr, in this shape:

{"error": {"kind": "invalid_input", "message": "a real embed run needs an explicit scope", "hint": "pass --limit N (bounded) or --all (full backfill), or preview with --dry-run"}}

Failure-mode index

Section titled “Failure-mode index”
Symptom Likely surface First check Citation
query exits 64, kind parse_error duckdb.ParserException, or no SQL argument at all Run atif-sql schema --format json for the real view and column names; a bare atif-sql query also exits 64 packages/atif-cli/src/atif_cli/duck_errors.py:38-44, packages/atif-cli/src/atif_cli/app.py:1517-1528
query exits 65, kind catalog_error duckdb.CatalogException: the view, macro, or column does not exist atif-sql schema answers from the static catalog with no DuckDB bind, so it works even on an unmaterialized corpus packages/atif-cli/src/atif_cli/duck_errors.py:45-51, packages/atif-cli/src/atif_cli/app.py:2464
query exits 70 with IO Error: No files found that match the pattern .../sessions/*/meta.json The corpus is not materialized; register_raw’s glob matches nothing atif-sql status; a watermark: never materialized line means run atif-sql materialize packages/atif-cli/src/atif_cli/duck_errors.py:52-58, packages/atif-duck/src/atif_duck/infrastructure/registry.py:737
query or search exits 65, kind embedding_mismatch The Lance store’s stamped (model, dim) differs from the active embedder The message names both sides; either point ATIF_SQL_EMBED_MODEL_ID at the model that wrote the store, or delete the store packages/atif-duck/src/atif_duck/domain/embedding_guard.py:78-85, packages/atif-cli/src/atif_cli/duck_errors.py:75-82
search exits 2, kind no_embeddings The store is empty or absent, so register_vss bound an empty fallback table atif-sql query 'SELECT count(*) FROM message_embeddings'; a count of 0 confirms it packages/atif-cli/src/atif_cli/app.py:2255-2283, packages/atif-duck/src/atif_duck/infrastructure/registry.py:1730-1761
embed, search or status --vector-probe exits 78 with holds a Lance dataset the lance extension can't open The store directory holds a dataset the extension can’t open, most likely a newest manifest cut short; the extension leaves it out of the catalog rather than failing, and query binds it as an empty message_embeddings ls <store>/embeddings.lance/_versions: the newest manifest has the smallest name, and one far smaller than the others is the likely cause. Restore the store from a backup, or move it aside and re-embed packages/atif-embed/src/atif_embed/infrastructure/lance_store.py:320-328, packages/atif-cli/src/atif_cli/app.py:2259-2273, packages/atif-duck/src/atif_duck/infrastructure/registry.py:1810
A bare atif-sql embed exits 64 The scope guard, which runs before any Bedrock client is built Pass --limit N, --all, or --dry-run; this is the accidental-full-backfill guard, not a fault packages/atif-cli/src/atif_cli/app.py:2003-2013
embed exits 78 and the cron lane stops trying A DomainError with terminal = True; the store or its config needs an operator Read the terminal marker file the refresh lane drops: three lines, store path, mtime, reason packages/atif-embed/src/atif_embed/domain/errors.py:14-26, scripts/atif-sql-refresh.sh:489-494
A session directory exists under sessions/ but no view returns its rows Torn artifact set: the dir has no meta.json, so the meta gate excludes it from every reader Grep the WARNING; meta.json is written last and completes the set packages/atif-duck/src/atif_duck/infrastructure/registry.py:567-571, packages/atif-corpus/src/atif_corpus/application/materialize.py:706-707
atif-sql status and SELECT count(*) FROM sessions disagree status counts directories with no meta check; on the per-session path the view applies the meta gate and the session id boundary, and skips a dir with no readable trajectory Run the count with --no-lake: the difference is the number of excluded dirs, and each one logs a WARNING packages/atif-cli/src/atif_cli/app.py:1308-1312, packages/atif-duck/src/atif_duck/infrastructure/registry.py:748
materialize prints all zeroes and reads as an idle, complete corpus Every session was unreadable; an unreadable session lands in no other counter Read the unreadable field, not just materialized packages/atif-cli/src/atif_cli/app.py:780, packages/atif-corpus/src/atif_corpus/application/materialize.py:253-261
materialize exits 78, kind suspicious_scan The scan found zero sessions over a non-empty corpus while every source directory listed, almost always a wrong source_root Compare the source_root line in atif-sql status against where the transcripts actually are; nothing was marked packages/atif-cli/src/atif_cli/app.py:1027-1041, raise at packages/atif-corpus/src/atif_corpus/application/materialize.py:1661-1667
A changed session never materializes Quiescence: the newest source mtime is younger than quiesce_seconds (default 300) atif-sql status reports it under live; --force overrides staleness but never quiescence packages/atif-corpus/src/atif_corpus/domain/sessions.py:84, packages/atif-corpus/src/atif_corpus/domain/sessions.py:218-219, packages/atif-corpus/src/atif_corpus/domain/sessions.py:220-221
A session stays live forever and warns about a FUTURE mtime The writing host’s clock is ahead, so the age is negative and never meets the threshold Fix the clock on the host writing the transcript packages/atif-corpus/src/atif_corpus/domain/sessions.py:107-113
materialize re-converts every quiescent session after watermark.json is damaged An unreadable, wrong-shaped, or non-numeric watermark degrades to empty with a WARNING, so every session reads as stale Grep treating corpus as unmaterialized; the pass writes a fresh watermark, and the cost is one full re-materialization packages/atif-corpus/src/atif_corpus/application/materialize.py:424-451
Every session fails conversion at once, on the same ConversionError The converter is a port of harbor 0.24.0’s, so a systematic failure is a bug in the port or a transcript shape it never saw Run the parity oracle: ATIF_PARITY_LIMIT=0 uv run pytest packages/atif-converter/tests/test_parity_live.py, and read the diff paths packages/atif-converter/tests/harbor_oracle.py:358
One session fails with N source file(s) changed while converting … The session is still being written; the snapshot re-check refused to publish inconsistent artifacts Retry once the session goes quiet; this is self-clearing packages/atif-converter/src/atif_converter/application/convert_and_audit.py:199-218, packages/atif-converter/src/atif_converter/domain/errors.py:57
convert prints a trajectory and then exits 65 The enriched trajectory failed the TrajectoryValidator vendored from harbor Read validation_errors in the JSON it already printed packages/atif-cli/src/atif_cli/app.py:760-772, packages/atif-converter/src/atif_converter/infrastructure/harbor_adapter.py:76
loss_report.json shows records_dropped > 0 Fidelity gap 2: only user and assistant records become steps, and a record session_events.jsonl doesn’t keep is dropped Check gaps_observed against the named gaps; these are documented losses, not defects packages/atif-converter/src/atif_converter/domain/fidelity.py:71-77
analyze summary carries budget_exhausted: true and a stage {"skipped": "budget_exhausted", "consecutive_skips": N} The dollar ceiling was crossed against running actual usage Read llm_spent_usd in the same summary; raise --max-cost-usd or wait for the next run packages/atif-analytics/src/atif_analytics/application/analyze.py:112-128, packages/atif-analytics/src/atif_analytics/application/analyze.py:150-152
An LLM stage skips the same sessions every run, forever The retry queue is exhausted at 5 attempts; attempts >= 5 with completed_at IS NULL is permanent and nothing resets it sqlite3 <corpus_root>/analytics/state.db 'SELECT * FROM retry_queue' (note the pipeline is user_friction, never friction) packages/atif-analytics/src/atif_analytics/infrastructure/sqlite_state/retry_queue.py:199-204, packages/atif-analytics/src/atif_analytics/infrastructure/sqlite_state/checkpointer.py:37-42
An LLM run stalls for minutes with no output Bedrock throttling under tenacity, in either adapter: 10 attempts, exponential 2 s to 60 s The per-backoff WARNING names the attempt number and the sleep packages/atif-models/src/atif_models/infrastructure/openai_bedrock.py:183-189, packages/atif-models/src/atif_models/infrastructure/anthropic_bedrock.py:221-227, packages/atif-models/src/atif_models/infrastructure/bedrock_common.py:76-85
A classify row reads [refused], or a refusals sidecar row appears RefusalError: a content filter or a refusal finish_reason (openai), or stop_reason: refusal (anthropic, with its stop_details.category in the reason) Nothing to fix here: a refusal is terminal, checkpointed, and never retried packages/atif-models/src/atif_models/domain/ports.py:43-48, packages/atif-models/src/atif_models/infrastructure/openai_bedrock.py:219-225, packages/atif-models/src/atif_models/infrastructure/anthropic_bedrock.py:252-257
An LLM unit lands in the retry queue with a truncation error finish_reason=length (openai) or stop_reason=max_tokens (anthropic) twice, or once already at the effort floor Raise max_completion_tokens or shrink the prompt; the automatic ladder has only one reachable rung packages/atif-models/src/atif_models/infrastructure/openai_bedrock.py:226-232, packages/atif-models/src/atif_models/infrastructure/openai_bedrock.py:160-181, packages/atif-models/src/atif_models/infrastructure/anthropic_bedrock.py:258-264, packages/atif-models/src/atif_models/infrastructure/anthropic_bedrock.py:195-219
An anthropic unit lands in the retry queue naming unexpected stop_reason or no text content The response stopped somewhere other than end_turn, or carried only a thinking block Read the reason string; it names the stop reason or the block types the response held packages/atif-models/src/atif_models/infrastructure/anthropic_bedrock.py:265-276
llm_spent_usd runs well past the dry-run plan’s estimate The plan prices at the standard rates only; a call over the spec’s long-context threshold (272,000 input tokens on GPT-6, 100,000 on Haiku 5.5) bills the whole request at the long rates, and running actuals price that Divide the plan’s estimated_input_tokens by its llm_calls: an average near the threshold means some calls bill at the long rates. The long-context split isn’t in the analyze summary packages/atif-analytics/src/atif_analytics/domain/costs.py:35-49, packages/atif-models/src/atif_models/domain/registry.py:175-210
A classify, conflicts, or perceived verdict ignores what a tool returned The transcript renders each tool result as a size stub, (N chars omitted), so the model never saw the content Set ATIF_SQL_SESSION_TEXT_TOOL_RESULT_BODIES=true to send bodies clipped at 50,000 characters, and expect the input tokens to grow with them packages/atif-analytics/src/atif_analytics/domain/transcript.py:306-312, packages/atif-analytics/src/atif_analytics/infrastructure/settings.py:90-94
analyze crashes with a raw traceback and exit 1 The analyze command body has no except, so nothing maps analytics errors into EXIT_CODES Read the traceback’s innermost frame; the exit code carries no information here packages/atif-cli/src/atif_cli/app.py:1808-1822
atif-sql examples raises a bare KeyError naming ARG_EXEMPLARS or DESCRIPTIONS A catalog drift tripwire firing at example-build time, not query time Add the missing exemplar or description entry named in the message packages/atif-duck/src/atif_duck/domain/examples.py:164-171, packages/atif-duck/src/atif_duck/domain/examples.py:180-188
A cron lane logs analytics not yet installed, skipping and exits 0 The resolved CLI predates the analyze subcommand; ~/.local/bin/atif-sql wins over the workspace venv Check which binary the lane resolved, in the CLI-resolution order the script documents scripts/atif-sql-refresh.sh:414-419, scripts/atif-sql-refresh.sh:301-308
The refresh log shows [structural] lane removed 2026-09-27 ... once an hour A crontab line still calls the structural lane, which the script accepts and exits 0 on Delete the hourly structural crontab line; atif-sql cron install prints a reminder comment for it scripts/atif-sql-refresh.sh:199-202, packages/atif-cli/src/atif_cli/cron.py:201-202
The refresh log shows vector-probe <corpus>: TRIP median_7=... after a compact run The median of that corpus’s last 7 nightly probe timings is over 2.0 s: search ranks every stored vector, and the store has grown past the point an exact scan stays quick Move search onto a vector index for that store; until then each search pays about the logged scan_s. atif-sql status --vector-probe repeats one timing on demand scripts/atif-sql-refresh.sh:357-365, scripts/atif-sql-refresh.sh:398, packages/atif-cli/src/atif_cli/vector_probe.py:43

Log and error surfaces

Section titled “Log and error surfaces”

There is no structured-logging platform, no log file the application writes, and no observability bootstrap. The whole surface is loguru on stderr plus the shell-side refresh log.

Surface Where it emits What to grep for Citation
The one loguru sink stderr, WARNING and above unless ATIF_SQL_LOG_LEVEL sets another level; logger.remove() deletes loguru’s default handler first WARNING / ERROR; INFO and DEBUG lines appear only under ATIF_SQL_LOG_LEVEL=INFO or DEBUG packages/atif-cli/src/atif_cli/app.py:2556-2564
Classified error envelope, pipe form stderr, one JSON object "error", "kind", "hint" packages/atif-cli/src/atif_cli/errors.py:78-86
Classified error, TTY form stderr, one line plus an optional hint line [<kind>] at line start, then hint: packages/atif-cli/src/atif_cli/output.py:279-282
Process exit code the shell the reachable EXIT_CODES values packages/atif-cli/src/atif_cli/errors.py:30-66
Per-session materialize failures stderr, table format only FAILED <session_id>:, UNREADABLE <session_id>, REJECTED '<name>', and LAKE PENDING packages/atif-cli/src/atif_cli/app.py:843-854
Materialize report stdout, JSON on a pipe unreadable, failures, removed_session_ids packages/atif-cli/src/atif_cli/app.py:798-824
Corpus freshness atif-sql status stdout watermark_age_seconds, staleness, source_root packages/atif-cli/src/atif_cli/app.py:1355-1380
DDL failure narration stderr via logger.exception, with loguru’s decorated traceback, then re-raised Failed to register raw readers over, Failed to register derived views, Failed to register macros packages/atif-duck/src/atif_duck/infrastructure/registry.py:996, packages/atif-duck/src/atif_duck/infrastructure/registry.py:1568, packages/atif-duck/src/atif_duck/infrastructure/registry.py:2163
Torn-session exclusion stderr WARNING, one line per excluded dir Skipping incomplete session dir packages/atif-duck/src/atif_duck/infrastructure/registry.py:567-571
Unreadable-store fallback stderr WARNING at VSS bind holds a dataset the lance extension can't open packages/atif-duck/src/atif_duck/infrastructure/registry.py:1731-1742
Empty-store fallback stderr WARNING at VSS bind No Lance embeddings table at packages/atif-duck/src/atif_duck/infrastructure/registry.py:1744-1749
Unreadable-source warnings stderr WARNING during the scan NOT treated as deleted, NOT treating it as deleted packages/atif-corpus/src/atif_corpus/infrastructure/scanner.py:169, packages/atif-corpus/src/atif_corpus/infrastructure/scanner.py:269
Source-removal marking suppression stderr WARNING not marking any session source-removed, keeping session packages/atif-corpus/src/atif_corpus/application/materialize.py:1675, packages/atif-corpus/src/atif_corpus/application/materialize.py:1204
Watermark degradation stderr WARNING treating corpus as unmaterialized packages/atif-corpus/src/atif_corpus/application/materialize.py:435, packages/atif-corpus/src/atif_corpus/application/materialize.py:438, packages/atif-corpus/src/atif_corpus/application/materialize.py:449
Bedrock retry narration stderr WARNING, one per backoff bedrock invoke retry (LLM, both adapters), Retrying … seconds as it raised (embeddings) packages/atif-models/src/atif_models/infrastructure/bedrock_common.py:76-85, packages/atif-embed/src/atif_embed/infrastructure/cohere_bedrock.py:94-101
Degraded-effort retry stderr WARNING finish_reason=length at effort= (openai), stop_reason=max_tokens at effort= (anthropic) packages/atif-models/src/atif_models/infrastructure/openai_bedrock.py:169-175, packages/atif-models/src/atif_models/infrastructure/anthropic_bedrock.py:205-210
Cost-ceiling events stderr, WARNING or ERROR at 3 consecutive skips cost ceiling hit packages/atif-analytics/src/atif_analytics/application/analyze.py:116-126, packages/atif-analytics/src/atif_analytics/application/use_cases/classify.py:230
Terminally failed embed batch stderr ERROR failed terminally, the next run re-picks these rows packages/atif-embed/src/atif_embed/infrastructure/cohere_bedrock.py:378-385
Clipped embedding input stderr WARNING Clipping text at position, content past the cap will not match a search packages/atif-embed/src/atif_embed/infrastructure/cohere_bedrock.py:333-339
Refresh-script log a file appended under the script’s .run/ sibling (or ATIF_SQL_REFRESH_RUN_DIR), one date -Is-stamped line per event; the directory is gitignored, so the path is named in prose only refresh complete (mode=, skip[, TERMINAL:, FATAL: scripts/atif-sql-refresh.sh:186-189, scripts/atif-sql-refresh.sh:638
Refresh log, machine-read subset atif-sql cron status stdout only the completion and skip line shapes are parsed packages/atif-cli/src/atif_cli/cron.py:61-62
Lane lock state atif-sql cron status, from a nonblocking flock probe plus the pidfile RUNNING (pid …) versus idle packages/atif-cli/src/atif_cli/cron.py:115-130, packages/atif-cli/src/atif_cli/cron.py:133-150
Embed terminal marker one file per corpus beside the refresh log; three lines: store path, store mtime, reason the reason string, parsed out of the exit-78 envelope scripts/atif-sql-refresh.sh:489-494
Analytics durable state <corpus_root>/analytics/state.db, sqlite in WAL mode, no read command retry_queue, budget_skips, session_checkpoint packages/atif-analytics/src/atif_analytics/domain/layout.py:81-83, packages/atif-analytics/src/atif_analytics/infrastructure/sqlite_state/checkpointer.py:44-62
Analyze summary stdout JSON budget_exhausted, llm_spent_usd, consecutive_skips packages/atif-analytics/src/atif_analytics/application/analyze.py:127, packages/atif-analytics/src/atif_analytics/application/analyze.py:150-152
Per-session loss accounting loss_report.json inside each corpus session dir gaps_observed, records_dropped, records_total packages/atif-converter/src/atif_converter/domain/fidelity.py:114-137

These properties of this surface change how you read it.

Everything below WARNING is hidden by default. main() takes the level from ATIF_SQL_LOG_LEVEL (default WARNING; an unknown level falls back to it with a warning). There is no --verbose and no --log-level, and LOGURU_LEVEL has no effect: logger.remove() deletes the default handler that would honor it (packages/atif-cli/src/atif_cli/app.py:2556-2564). The INFO lines that narrate a materialize pass (packages/atif-corpus/src/atif_corpus/application/materialize.py:1811-1825) or an embed backfill (packages/atif-embed/src/atif_embed/application/embed.py:276-282) appear only under ATIF_SQL_LOG_LEVEL=INFO, and each materialize pool worker reads the same variable (packages/atif-cli/src/atif_cli/app.py:859).

A logger.exception traceback on stderr is not a crash. The registration path logs and re-raises, so the decorated traceback is followed by the JSON envelope and a classified exit code (packages/atif-duck/src/atif_duck/infrastructure/registry.py:994-997). Read the last stderr line, not the first.

First-checks ladder

Section titled “First-checks ladder”

Cheapest first. Steps 1 through 6 are free and read-only; step 10 spends money.

  1. Read the exit code. It partitions the problem before you read any text: 64 means your input, 65 means the object cannot be bound, 70 means the adapter or the query sandbox, 77 means the process runs as root, 78 means an operator is required, and 1 means the taxonomy did not cover it. packages/atif-cli/src/atif_cli/errors.py:30-66
  2. Read the last line on stderr. On a pipe it is the JSON envelope with kind, message, and hint; on a TTY it is [kind] message plus a hint: line. The hint names the recovery command in most cases. packages/atif-cli/src/atif_cli/output.py:271-285
  3. Run atif-sql status. It replays the exact planning decision materialize would make, without converting anything, so its stale / up-to-date / live split answers “would a pass do anything right now?”, and it prints the resolved source_root and corpus_root, which is how a wrong-root problem becomes visible. packages/atif-cli/src/atif_cli/app.py:1293-1300, packages/atif-cli/src/atif_cli/app.py:1328-1329
  4. Run atif-sql schema or atif-sql examples. Both answer from the static catalog with no DuckDB connection, so they work on an unmaterialized corpus and separate “my SQL is wrong” from “the corpus is not there”. Every example is executed by the test suite against a fixture corpus, so a verbatim copy is known to parse. packages/atif-cli/src/atif_cli/app.py:2470-2474, packages/atif-cli/src/atif_cli/app.py:2383-2384, packages/atif-cli/src/atif_cli/app.py:2380-2382
  5. Cross-check status’s session count against the view. atif-sql query --no-lake 'SELECT count(*) FROM sessions' applies the meta gate; status counts directories. A gap is the number of dirs the gates excluded (a torn dir, a rejected name, no readable trajectory), and each one also emits a WARNING naming itself. packages/atif-duck/src/atif_duck/infrastructure/registry.py:567-571
  6. Run atif-sql cron status. For anything that failed unattended: per lane it reports whether the flock is held and by which pid, the last completion with its exit code, the last skip, and a log tail. The lock probe acquires and releases nonblocking, so it perturbs no running lane. packages/atif-cli/src/atif_cli/cron.py:207, packages/atif-cli/src/atif_cli/cron.py:228-253
  7. Look for a terminal marker file beside the refresh log. Its presence is why an embed stopped being attempted; it holds the store path, the mtime it was keyed on, and the reason. The mtime is the newest one of anything under the store directory, so any change inside the store clears it on the next tick, an in-place repair of a manifest under embeddings.lance/_versions included. scripts/atif-sql-refresh.sh:471-481
  8. Query state.db directly at <corpus_root>/analytics/state.db. There is no CLI or SQL surface over it. retry_queue rows with attempts >= 5 and completed_at IS NULL are permanently blocked and need manual clearing; budget_skips rows are the consecutive-skip streak. The pipeline name is user_friction. packages/atif-analytics/src/atif_analytics/infrastructure/sqlite_state/retry_queue.py:42-56
  9. Raise the log level if the WARNING lines were not enough. ATIF_SQL_LOG_LEVEL=DEBUG shows the DEBUG narration of the staging sweep, the mid-scan races, the atomic-write tmp cleanups, and every view and macro registration. packages/atif-cli/src/atif_cli/app.py:2525
  10. Only now re-run a billable command, and bound it: analyze defaults to a dry run and needs --no-dry-run to spend, and embed refuses to run without --limit N or --all. atif-sql search embeds the query once per run, on Bedrock under the default Cohere provider. packages/atif-cli/src/atif_cli/app.py:1731, packages/atif-cli/src/atif_cli/app.py:2003, packages/atif-cli/src/atif_cli/app.py:2286

Known incident patterns

Section titled “Known incident patterns”

See also

Section titled “See also”