@carllee1983/dbcli 1.52.0 → 1.53.0

package/.github/skills/dbcli/reference.md

@@ -4,6 +4,69 @@ Companion to [SKILL.md](SKILL.md). Exhaustive flags, copy-paste examples, `shell
 
 For cross-engine support status, see `docs/feature-matrix.md` in the repository.
 
+## Index
+
+Jump straight to what you need — this file is long, and reading it end to end is
+never the right move.
+
+| Section | What is in it |
+|---|---|
+| [Global options and placement](#global-options-and-placement) | Flags that must precede the command (`--use`, `--config`), and [redirecting output](#redirecting-output). |
+| [Commands](#commands) | Every command, one `###` each — see the command list below. |
+| [Recovery Cookbook](#recovery-cookbook-agent-walkthroughs) | Failure-to-fix walkthroughs S1–S6, the [scenario index](#scenario-index), [risk gate cheat sheet](#risk-gate-cheat-sheet), and [common pitfalls](#common-pitfalls). |
+| [Interactive HTML dashboard](#interactive-html-dashboard) | `--ui` / `--format html`: [entry points](#entry-points), [data injection contract](#data-injection-contract), [`meta` shape](#meta-shape). |
+| [MongoDB Support](#mongodb-support) | Connection shapes, JSON query bodies, write-stage guard. |
+| [Redis Support](#redis-support) | Command permission tiers, size guards, key masking. |
+| [Elasticsearch Support](#elasticsearch-support) | DSL and Lucene queries, scroll export, mapping flattening. |
+
+**Commands** —
+[init](#init) ·
+[use](#use) ·
+[list](#list) ·
+[schema](#schema) ·
+[query](#query) ·
+[explain](#explain) ·
+[lint](#lint) ·
+[plan](#plan) ·
+[q](#q) ·
+[queries](#queries) ·
+[insert](#insert) ·
+[update](#update) ·
+[delete](#delete) ·
+[export](#export) ·
+[blacklist](#blacklist) ·
+[check](#check) ·
+[diff](#diff) ·
+[design](#design) ·
+[impact assess](#impact-assess) ·
+[snapshot](#snapshot) ·
+[assert](#assert) ·
+[proxy](#proxy) ·
+[status](#status) ·
+[inspect](#inspect) ·
+[report](#report) ·
+[guide](#guide) ·
+[recovery](#recovery) ·
+[recover](#recover) ·
+[audit](#audit) ·
+[verify](#verify) ·
+[verification](#verification) ·
+[backfill](#backfill) ·
+[evidence](#evidence) ·
+[doctor](#doctor) ·
+[completion](#completion) ·
+[upgrade](#upgrade) ·
+[shell](#dbcli-shell) ·
+[migrate](#migrate) ·
+[semantic](#semantic) ·
+[contract](#contract) ·
+[skill](#skill) ·
+[skill context](#skill-context) ·
+[skill tasks](#skill-tasks-agent-task-packs)
+
+Also worth knowing before you connect:
+[Agent configuration trust boundary](#agent-configuration-trust-boundary).
+
 ## Global options and placement
 
 These options are available on the root `dbcli` command. Root-level options must
@@ -561,13 +624,15 @@ dbcli q @analytics/revenue --param days=30 --format html > report.html
 - `--ui` — open the rendered HTML dashboard in the system browser (implies `--format html`; writes to a temp file then invokes `open` / `xdg-open` / `start`)
 - `--param <key=value>` — pass a parameter (repeatable)
 - `--param-file <path>` — JSON object whose keys are param names
-- `--no-limit` — skip the `SELECT * FROM (…) AS _dbcli_guard LIMIT 1000` wrap
+- `--no-limit` — skip the `SELECT * FROM (…) AS _dbcli_guard LIMIT 1001` wrap (the effective cap is 1000; one extra row is fetched to detect truncation)
 - `--dry-run` — print the bound SQL + values without executing
-- `--use <name>` — pick a v2 named connection
 - `--slow-ms <number>` — passive slow-query hint threshold (default `1000`; `0` disables). Same contract as `query` — see "Passive slow-query hint" there
 - `--recovery` — emit a `RecoveryEnvelope` on failure (see `recover`)
 - `--verify` — run the snippet's verification assertions after execution (only if the snippet defines them)
 
+`q` has no command-level `--use`. To run a snippet against a v2 named connection,
+use the global form: `dbcli --use <name> q @<snippet>`.
+
 **Permission:** query-only+
 
 #### Snippet file format
@@ -691,8 +756,9 @@ Size guard: `LRANGE` / `ZRANGE` stop overridden when `< 0` or `> 1000`; `SCAN` /
 ##### MongoDB snippets
 
 File extension: `.mongodb.sql`. Frontmatter must declare `engine: mongodb` and
-`operation: find` or `operation: aggregate`. `target: <collection>` provides a default
-collection that `dbcli q --collection <name>` can override. The body is JSON: an object
+`operation: find` or `operation: aggregate`. `target: <collection>` declares the
+collection the snippet runs against; the CLI has no flag to override it, so a different
+collection means a different snippet. The body is JSON: an object
 for `find` and an array for `aggregate`. Each `{{param}}` placeholder is JSON-encoded
 at substitution time — strings are quoted and escaped, so an attacker-supplied string
 cannot escape into operator position.
@@ -1126,6 +1192,7 @@ dbcli design validate [--file <path>] [--format <format>]
 dbcli design render [--file <path>] [--format <format>]
 dbcli design diff (--against-cache | --against-orm <paths>) [options]
 dbcli design propose (--against-cache | --against-orm <paths>) [options]
+dbcli impact assess --design <path> (--against-cache | --against-orm <paths>) --output <path> [options]
 ```
 
 ```bash
@@ -1137,6 +1204,7 @@ dbcli design render --format mermaid
 dbcli design diff --against-cache --format markdown
 dbcli design diff --against-orm ./prisma/schema.prisma --format markdown
 dbcli design propose --against-orm ./prisma/schema.prisma --format markdown
+dbcli impact assess --design ./dbcli.design.json --against-cache --output ./impact.json --fail-on warn
 ```
 
 | Option | Applies to | Default | Meaning |
@@ -1153,8 +1221,14 @@ dbcli design propose --against-orm ./prisma/schema.prisma --format markdown
 `diff` and `propose` require **exactly one** comparison target; passing both or
 neither is an error.
 
+### impact assess
+
+Writes an offline JSON or Markdown report for the known declared impact of a design change. It requires `--design`, exactly one baseline (`--against-cache` or `--against-orm`), and an explicit workspace-contained `--output`. `--format` is `json` (default) or `markdown`; `--fail-on` is `error`, `warn`, or `never` (default). The threshold changes only the successful command exit code after the report is written. The report is limited to declared/partial coverage and never claims complete coverage. It may join the optional reviewed `dbcli.data-access.json` manifest only after validating its canonical semantic references and existing workspace-relative source paths; it never reads those sources. An optional explicit `--events <path>` is bounded and streamed through a redaction-first projection that retains recent safe table metadata only. It never starts a proxy, reads a rotated log, renders SQL/literals/errors/sessions/paths, or makes the command fail solely because workload evidence is absent or advisory. It reads no database, executes no SQL, and never reads saved-query bodies.
+
 #### Artifact shape
 
+This example validates clean (0 errors, 0 warnings):
+
 ```json
 {
   "version": 1,
@@ -1166,14 +1240,24 @@ neither is an error.
       "description": "Completed purchases.",
       "fields": [
         { "name": "id", "type": "bigint", "nullable": false, "primaryKey": true, "unique": true },
-        { "name": "customer_id", "type": "bigint", "nullable": false, "primaryKey": false, "unique": false }
+        { "name": "customer_id", "type": "bigint", "nullable": false },
+        { "name": "created_at", "type": "timestamptz", "nullable": false }
       ],
-      "indexes": [{ "name": "orders_customer_idx", "columns": ["customer_id"], "unique": false }]
+      "indexes": [
+        { "name": "orders_customer_idx", "columns": ["customer_id", "created_at"], "unique": false }
+      ]
+    },
+    {
+      "name": "customers",
+      "table": "customers",
+      "fields": [
+        { "name": "id", "type": "bigint", "nullable": false, "primaryKey": true, "unique": true }
+      ]
     }
   ],
   "relationships": [
     {
-      "name": "orders_customer",
+      "name": "orders-customer",
       "from": { "model": "orders", "field": "customer_id" },
       "to": { "model": "customers", "field": "id" },
       "cardinality": "many-to-one"
@@ -1188,11 +1272,38 @@ It holds no SQL, credentials, rows, or provider configuration. `design init`
 emits this envelope with empty `models`, `relationships`, `accessPatterns`, and
 `decisions`.
 
+#### Naming and limits
+
+Two different naming rules apply, and mixing them up is the most common way an
+artifact fails before any review rule runs.
+
+| Applies to | Rule |
+|---|---|
+| `models[].name`, `relationships[].name`, `relationships[].from/to.model`, `accessPatterns[].model`, `decisions[].name` | lowercase kebab-case, `^[a-z][a-z0-9-]*$` — underscores are rejected |
+| `models[].table`, `fields[].name`, `indexes[].name`, `indexes[].columns[]`, `filters[]`, `sort[]` | SQL identifier, `^[A-Za-z_][A-Za-z0-9_]*$` |
+
+Relationship endpoints reference a **model name**, not a table name. Every object
+is strict: an unknown key is an error, not ignored. `description` and `rationale`
+are 1–1000 characters and must not contain SQL keywords or a connection string —
+"rows we delete after 30 days" is rejected for the word `delete`. `fields[].type`
+is at most 100 characters with no `;` or newline. `primaryKey`, `unique`,
+`indexes`, `filters`, `sort`, `relationships`, `accessPatterns`, and `decisions`
+may all be omitted.
+
+Limits: file 256 KiB, 100 models, 100 fields per model, 200 relationships,
+200 access patterns, 100 decisions, 1–16 columns per index, ≤16 entries in
+`filters` and `sort`.
+
 #### Review findings
 
 `validate` is fail-closed: any `error` finding exits `1`, and `render`, `diff`,
 and `propose` refuse to do their work while errors remain.
 
+Structural problems — malformed JSON, an unknown key, a naming or type violation,
+a missing file, or a file over 256 KiB — are rejected before review runs and are
+reported as a single `INVALID_ARTIFACT` finding (`error`) whose `path` points at
+the offending JSON location. None of the codes below appear in that case.
+
 | Severity | Codes |
 |---|---|
 | `error` | `NO_MODELS`, `DUPLICATE_MODEL`, `DUPLICATE_TABLE`, `DUPLICATE_FIELD`, `PRIMARY_KEY_COUNT`, `NULLABLE_PRIMARY_KEY`, `UNKNOWN_INDEX_FIELD`, `DUPLICATE_RELATIONSHIP`, `REVERSE_RELATIONSHIP`, `UNKNOWN_RELATIONSHIP_MODEL`, `UNKNOWN_RELATIONSHIP_FIELD`, `RELATIONSHIP_TYPE_MISMATCH`, `MANY_TO_MANY_REQUIRES_BRIDGE`, `ONE_TO_ONE_REQUIRES_UNIQUE_FK`, `UNKNOWN_ACCESS_MODEL`, `UNKNOWN_ACCESS_FIELD` |
@@ -1297,7 +1408,7 @@ Opt-in flag trio that persists a **VerificationArtifact JSON** (schema v1) under
 | Flag | Required | Description |
 | :--- | :--- | :--- |
 | `--write-verification-artifact` | opt-in | Trigger artifact write. No-op when no verdict has been produced. |
-| `--verification-subject <kind:name>` | yes (when flag is set) | Subject identifier. Format: `<kind>:<name>`. Allowed kinds: `recovery`, `task-pack`, `assertion`, `migration`, `backfill`, `manual`. |
+| `--verification-subject <kind:name>` | yes (when flag is set) | Subject identifier. Format: `<kind>:<name>`. Allowed kinds: `recovery`, `task-pack`, `assertion`, `migration`, `backfill`, `table`, `manual`. |
 | `--verification-summary <text>` | no | Free-text summary line stored in the artifact. Default when pass: "Assertion verified the expected state." Default when fail: "Assertion did not verify the expected state." |
 
 **Output contract:**
@@ -1559,7 +1670,7 @@ Examples:
 
 Boundaries:
 - Recovery only **suggests** commands; agents (or humans) execute them. No automatic remediation in v1.15.0.
-- As of v1.16.0, `--recovery` is honored on `query`, `q`, `insert`, `update`, `delete`, `export`, `schema`, and `inspect`. Other commands (`report`, `guide`, `doctor`, `migrate`, `init`, `use`, `status`, `list`, `check`, `diff`, `plan`, `shell`, `blacklist`, `completion`, `upgrade`, `skill`) keep their existing error behavior.
+- `--recovery` is honored on `query`, `q`, `insert`, `update`, `delete`, `export`, `schema`, `inspect`, `lint`, and `diff`. Other commands (`report`, `guide`, `doctor`, `migrate`, `init`, `use`, `status`, `list`, `check`, `plan`, `shell`, `blacklist`, `completion`, `upgrade`, `skill`) keep their existing error behavior.
 - `dbcli inspect --require-schema-cache` throws `SCHEMA_CACHE_MISSING` when the active SQL connection has no usable schema cache. Combine with `--recovery` for the structured envelope.
 - `BLACKLIST_COLUMN_WRITE` and `PERMISSION_DENIED` envelopes prepend a `risk: 'dry-run'` step (e.g. `dbcli insert <table> --dry-run`) when the failing operation was an INSERT / UPDATE / DELETE.
 - Recovery steps reuse the v1.14.0 `GuideStep` shape, including the full `risk` enum (`readonly` / `dry-run` / `write` / `unknown`).
@@ -1593,7 +1704,7 @@ Boundaries:
 |---|---|---|
 | `readonly` | local read-only | `dbcli inspect`, `dbcli doctor`, `dbcli blacklist list`, `dbcli schema <table>` |
 | `dry-run` | write subcommand invoked with `--dry-run` | `dbcli update orders --where id=1 --dry-run`, `dbcli q @x --dry-run` |
-| `local-write` | writes local config / cache / blacklist | `dbcli blacklist remove <table>`, `dbcli use <name>`, `dbcli schema --refresh` |
+| `local-write` | writes local config / cache / blacklist | `dbcli blacklist table remove <table>`, `dbcli use <name>`, `dbcli schema --refresh` |
 | `db-write` | mutates the connected database | `dbcli update orders --where id=1 --set …` (no `--dry-run`), `dbcli q @x` (no `--dry-run`) |
 | `interactive` | requires TTY | `dbcli init`, `dbcli init --force` |
 
@@ -1746,7 +1857,7 @@ dbcli recover --next --after-step 1 --result '{"status":"ok"}' --format markdown
 
 (v1.20.0+) Inspect, query, and manage the per-connection audit log written to `.dbcli/audit/<connection>.jsonl`.
 
-Audit entries are metadata-only by design — never raw SQL bodies, `--param` values, or result cell contents (D3 lock). Redaction is sourced from `tests/helpers/sensitive-output.ts` (same source as `inspect` / `guide` / `recover` agent contracts).
+Audit entries are metadata-only by design — never raw SQL bodies, `--param` values, or result cell contents (D3 lock). Redaction is sourced from `src/utils/redaction.ts` (same source as `inspect` / `guide` / `recover` agent contracts).
 
 #### Subcommands
 
@@ -1783,7 +1894,9 @@ Examples:
 | `<id-prefix>` | Positional. UUID or prefix ≥ 4 characters; ambiguous prefix exits 1 with disambiguation hint; prefix < 4 chars exits 1. | — |
 | `--recovery-ref <id>` | Find the audit entry whose `recovery_ref` field matches this id (exact, not prefix). Mutually exclusive with positional `<id-prefix>` (D-38). | — |
 | `--all` | Search across all connections. Output is an envelope `{ connection, entry }` (single-hit also envelope, for shape stability — D-36). | off |
-| `--no-brief` | Disable brief mode when a higher-level default enables it. | off |
+| `--brief` | Trim `metadata` and `redacted_query` from the entry. | off |
+| `--for-agent` | Shortcut for `--format json --brief`. | off |
+| `--no-brief` | Disable brief mode when a higher-level default enables it (e.g. `--for-agent`). | off |
 | `--format <fmt>` | `table` \| `json`. | `table` |
 
 Examples:
@@ -1817,7 +1930,7 @@ Output reports: writer enabled/disabled, last write result, file-lock state, rot
 #### Boundaries
 
 - Entries are append-only JSONL; rotation triggers at `~10 MB` or `~1000` entries (whichever first). Previous segment is preserved as `.jsonl.1`.
-- Bi-directional `recovery_ref` / `audit_ref` linkage is wired on every command that accepts `--recovery`: `query`, `inspect`, `insert`, `update`, `delete`, `export`, `q`, and `schema`. Use `audit tail --recovery-ref <id>` to find the audit entry an envelope was emitted alongside.
+- Bi-directional `recovery_ref` / `audit_ref` linkage is wired on every command that accepts `--recovery`: `query`, `inspect`, `insert`, `update`, `delete`, `export`, `q`, and `schema`. Use `audit show --recovery-ref <id>` to find the audit entry an envelope was emitted alongside.
 - Audit writer failures are non-fatal (D6): main command result and exit code are preserved; a stderr warning is emitted. `audit health` surfaces the failure reason.
 - Reader truncation tolerance: a crash-truncated last line is skipped with a stderr warn `[dbcli audit] skipping truncated last line in <file>`; a mid-file non-JSON line is treated as corruption, exits 1, and points at `dbcli audit clear`.
 
@@ -1835,7 +1948,7 @@ Output reports: writer enabled/disabled, last write result, file-lock state, rot
 ### verify
 
 Run a verification scenario. `verify` **runs** verification scenarios (safe-backfill,
-migration, rollback) and never executes writes/DDL. `verification` **inspects and manages**
+migration, rollback, constraint) and never executes writes/DDL. `verification` **inspects and manages**
 the local result artifacts those scenarios produce under `.dbcli/verification/`.
 
 ```bash
@@ -1847,7 +1960,8 @@ dbcli verify safe-backfill \
   --expect "value == 0"
 
 # After-write: re-run guards, run the read-back assertion, write a v1 artifact.
-dbcli verify safe-backfill ... --after-write
+# An optional receipt is provenance only, never approval to execute a write.
+dbcli verify safe-backfill ... --after-write --evidence-receipt .dbcli/evidence/verify-receipt.json
 
 # JSON for agents.
 dbcli verify safe-backfill ... --format json
@@ -1855,7 +1969,11 @@ dbcli verify safe-backfill ... --format json
 
 Options: `--table` (req), `--query` (req, analyzed not executed), `--verify-query`
 (req, **plain SELECT only**), `--expect` (req), `--after-write`, `--format <table|json>`,
-`--subject-name <name>`, `--summary <text>`.
+`--subject-name <name>`, `--summary <text>`, `--evidence-receipt <workspace-relative-path>`
+(after-write only). A receipt is safe provenance for the already-executed verification,
+not approval to execute a write. Its `succeeded`/`failed` outcome remains distinct from
+the artifact status (`verified`, `not_verified`, `indeterminate`, or `blocked`); task-pack
+`planned` evidence stays plan-only.
 
 Guard constraints (fail closed): `--verify-query` must be a **plain `SELECT`** —
 `EXPLAIN`/`EXPLAIN ANALYZE`, `SHOW`, `DESCRIBE`, and data-modifying CTEs are rejected
@@ -2078,7 +2196,7 @@ dbcli verification list --include-invalid --format json
 | `--format <json\|table>` | Output format. | `json` |
 | `--limit <n>` | Maximum number of entries to return. | `20` |
 | `--status <status>` | Filter by status. One of: `verified`, `not_verified`, `indeterminate`, `blocked`. | all |
-| `--subject <kind[:name]>` | Filter by subject kind or exact `kind:name`. Allowed kinds: `recovery`, `task-pack`, `assertion`, `migration`, `backfill`, `manual`. | all |
+| `--subject <kind[:name]>` | Filter by subject kind or exact `kind:name`. Allowed kinds: `recovery`, `task-pack`, `assertion`, `migration`, `backfill`, `table`, `manual`. | all |
 | `--include-invalid` | Surface malformed artifact files (normally skipped silently). Invalid files are returned as a separate top-level `invalid` array in JSON output, each entry shaped `{ "path": "...", "filename": "...", "error": "..." }`. When off, `invalid` is `[]`. | off |
 
 **Missing directory:** if `.dbcli/verification/` does not exist, exits `0` with an
@@ -2192,6 +2310,7 @@ includes `storageDir`, `dryRun`, `cutoff`, `criteria`, `protected`, `candidates`
 | `assertion` | General-purpose inline assertions. |
 | `migration` | Schema migration pre/post checks. |
 | `backfill` | Data backfill verification assertions. |
+| `table` | `verify constraint` artifacts — the subject name is the table checked. |
 | `manual` | Manually triggered or ad-hoc verification runs. |
 
 **Storage root:** `<cwd>/.dbcli/verification/` (cwd-relative; independent of `--config`).
@@ -2359,6 +2478,66 @@ dbcli migrate drop-enum status --execute --force
 
 **AI agent note:** Always use dry-run first (no `--execute`) to preview generated SQL. Only add `--execute` after confirming the SQL is correct. For DROP operations, both `--execute` and `--force` are required.
 
+### evidence
+
+Compose a canonical, offline **Evidence Pack** from existing verification artifacts
+and audit entries. This command never opens a database connection. A pack retains only
+safe pointers and outcome fields: it excludes SQL, targets, audit metadata, verification
+summaries, result data, and credentials. Claims are external statements, never dbcli
+verification verdicts.
+
+#### `evidence compose`
+
+```bash
+dbcli evidence compose --claims ./claims.json --verification ver_abcd --audit 1a2b \
+  --receipt .dbcli/evidence/verify-receipt.json \
+  --output .dbcli/evidence/review.json --format json
+```
+
+| Flag | Purpose | Default |
+|---|---|---|
+| `--claims <file>` | Required JSON file with exactly `subject` and `claims`; every claim has an `id` and plain-language `text`. SQL-shaped text, credentials, error content, and blacklisted identifiers are rejected. | — |
+| `--verification <selector...>` | One or more verification artifact ids, unique prefixes, filenames, or in-bounds paths. | none |
+| `--audit <selector...>` | One or more audit ids or unique prefixes (minimum four characters) from the active connection. | none |
+| `--receipt <path...>` | Explicit workspace-relative receipt path(s). Receipts contribute safe provenance only; they are not execution approval. | none |
+| `--output <path>` | Required new JSON path under the current workspace. Existing files and paths outside the workspace are refused. | — |
+| `--format <json\|markdown>` | Compose receipt format printed to stdout. | `json` |
+
+At least one `--verification`, `--audit`, or `--receipt` reference is required. The resulting pack
+contains a canonical SHA-256 integrity digest and `coverage.completeForDeclaredEvidence:
+true` for its explicitly selected references.
+
+#### `evidence validate`
+
+```bash
+dbcli evidence validate --file .dbcli/evidence/review.json --format json
+```
+
+| Flag | Purpose | Default |
+|---|---|---|
+| `--file <path>` | Required evidence-pack JSON file. | — |
+| `--format <json\|markdown>` | Validation report format. | `json` |
+
+Exit `0` when digest and all selected sources are valid. If a referenced audit or
+verification artifact later rotates, is cleared, disappears, or belongs to a different
+active connection, output stays parseable with `integrity: "valid"` and
+`references: "source-expired"`, then exits `1`.
+
+#### `evidence render`
+
+```bash
+dbcli evidence render --file .dbcli/evidence/review.json --format markdown
+```
+
+| Flag | Purpose | Default |
+|---|---|---|
+| `--file <path>` | Required evidence-pack JSON file. | — |
+| `--format <json\|markdown>` | Render format. | `markdown` |
+
+`render` validates the pack digest and active blacklist policy but does not resolve its
+original references. It is therefore available for historical review after `validate`
+reports `source-expired`, unless an exposed pack field is now blacklisted.
+
 ### semantic
 
 Validate or print the optional, version-controlled `dbcli.semantic.json` in the
@@ -2485,6 +2664,39 @@ limits, audit, and recovery safeguards remain in force.
 
 **Permission:** n/a (local files only; no database connection).
 
+### contract
+
+Inspect the optional, version-controlled `dbcli.contracts.json` without opening a
+database connection. A contract adds a reviewed owner and descriptive evidence policy
+to canonical semantic references; it is not an executable data contract, assertion,
+query, or verification scenario.
+
+```bash
+dbcli contract validate --format json
+dbcli contract context --format json
+dbcli contract context --format markdown
+dbcli contract search customer --format json
+dbcli contract drift --format json
+```
+
+The strict v1 artifact contains `version` and `contracts`. Each contract has a unique
+canonical `name`, `status` (`draft`, `approved`, or `deprecated`), bounded plain-text
+`description` and `owner`, one or more canonical semantic `subjects`, optional aliases,
+and an `evidencePolicy` of `none`, `receipt-required`, or `verification-required`.
+Subjects must remain in the existing visible semantic registry; protected identifiers,
+unknown keys, duplicate names or subjects, SQL-like text, credentials, and stale
+references fail closed without exposing protected names.
+
+`contract validate` requires an explicit valid artifact. `contract context` and
+`contract search` return only valid approved contracts; draft and deprecated contracts
+remain local review artifacts. `contract drift` reports `valid`, `stale`, `invalid`, or
+`unavailable` evidence offline and exits non-zero except for `valid`. A missing default
+file is allowed by `skill context` and leaves ordinary semantic context unchanged; a
+present invalid file fails closed. No contract command writes a file, opens a database
+connection, widens a `QueryDraft`, or changes permission/blacklist behavior.
+
+**Permission:** n/a (local files only; no database connection).
+
 ### skill
 
 Emit `SKILL.md` (and the companion `reference.md`) to stdout, a file, or an
@@ -3024,9 +3236,9 @@ Permission is derived from the command's first token (case-insensitive). Unknown
 
 | Tier | Commands |
 |------|----------|
-| `query-only` | `GET`, `MGET`, `STRLEN`, `EXISTS`, `TTL`, `PTTL`, `TYPE`, `SCAN`, `HGET`, `HGETALL`, `HKEYS`, `HVALS`, `HLEN`, `HEXISTS`, `HMGET`, `LRANGE`, `LLEN`, `LINDEX`, `SMEMBERS`, `SCARD`, `SISMEMBER`, `ZRANGE`, `ZREVRANGE`, `ZRANGEBYSCORE`, `ZCARD`, `ZSCORE`, `PING`, `ECHO` |
-| `read-write` | `SET`, `SETEX`, `SETNX`, `PSETEX`, `MSET`, `MSETNX`, `APPEND`, `INCR`/`INCRBY`, `DECR`/`DECRBY`, `HSET`/`HSETNX`/`HMSET`/`HINCRBY`, `LPUSH`/`RPUSH`/`LPOP`/`RPOP`/`LSET`, `SADD`/`SREM`, `ZADD`/`ZREM`, `EXPIRE`/`EXPIREAT`/`PEXPIRE`/`PERSIST`, `RENAME` |
-| `data-admin` | `DEL`, `UNLINK`, `HDEL` |
+| `query-only` | `GET`, `MGET`, `STRLEN`, `EXISTS`, `TTL`, `PTTL`, `TYPE`, `SCAN`, `HGET`, `HGETALL`, `HKEYS`, `HVALS`, `HLEN`, `HEXISTS`, `HMGET`, `LRANGE`, `LLEN`, `LINDEX`, `SMEMBERS`, `SCARD`, `SISMEMBER`, `ZRANGE`, `ZREVRANGE`, `ZRANGEBYSCORE`, `ZCARD`, `ZSCORE`, `XLEN`, `XREAD`, `XRANGE`, `XREVRANGE`, `PING`, `ECHO` |
+| `read-write` | `SET`, `SETEX`, `SETNX`, `PSETEX`, `MSET`, `MSETNX`, `APPEND`, `INCR`/`INCRBY`, `DECR`/`DECRBY`, `HSET`/`HSETNX`/`HMSET`/`HINCRBY`, `LPUSH`/`RPUSH`/`LPOP`/`RPOP`/`LSET`/`LREM`, `SADD`/`SREM`, `ZADD`/`ZREM`, `XADD`, `EXPIRE`/`EXPIREAT`/`PEXPIRE`/`PERSIST`, `RENAME` |
+| `data-admin` | `DEL`, `UNLINK`, `HDEL`, `XDEL` |
 | `admin` | `FLUSHDB`, `FLUSHALL`, `CONFIG`, `INFO`, `CLIENT`, `DEBUG`, `SHUTDOWN`, `KEYS`, `MONITOR`, `SAVE`, `BGSAVE`, `BGREWRITEAOF`, `REPLICAOF`, `SLAVEOF`, `ACL` |
 
 ### Schema inspection
@@ -3237,5 +3449,5 @@ GET /orders/_search
 
 - Writes (`insert`/`update`/`delete`) are not exposed yet — the adapter implements them, but the CLI currently only routes them for SQL and MongoDB. Read-only `export` (v1.22) and the interactive `shell` (v1.22) are available.
 - No `_search/scroll` or PIT pagination at the CLI layer; large pulls need a saved external script.
-- `check`, `diff`, `migrate`, and `q` are SQL-only and exit with errors (or fall through to a generic "unsupported" path).
+- `check`, `diff`, and `migrate` are SQL-only and exit with errors (or fall through to a generic "unsupported" path). `q` **is** supported — Elasticsearch snippets use the `.elasticsearch.sql` extension (see `@diag/es-cluster-health`).
 - Blacklist column rules are applied to flattened hit rows on `query`; table-level blacklist rejects an index up front.

package/CHANGELOG.md

@@ -5,6 +5,44 @@ All notable changes to dbcli are documented here.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [1.53.0] - 2026-08-09 - Offline evidence, semantic contracts, and impact assessment
+
+### Added
+
+- **Offline evidence packs.** `dbcli evidence compose`, `validate`, and `render` create, verify, and render workspace-contained evidence packs from safe claim text plus existing verification artifacts, audit entries, and optional assert receipts. Packs omit SQL, rows, targets, credentials, audit metadata, and verification summaries; source retention loss remains visible without preventing historical rendering.
+- **Evidence receipts for post-write assertions.** `assert --evidence-receipt <path>` atomically records safe provenance only after the verdict, audit attempt, and optional verification artifact are authoritative. Receipts are workspace-contained, contain no SQL or returned data, cannot be used as execution approval, and may be composed into an evidence pack.
+- **Reviewable semantic contracts.** `dbcli contract validate|context|search|drift` governs optional `dbcli.contracts.json` evidence expectations for canonical semantic terms. The commands are offline and read-only; only valid approved contracts enter ordinary agent context.
+- **Offline impact assessment.** `dbcli impact assess` creates a declared-coverage report for a design change against exactly one local schema-cache or ORM baseline, optionally incorporating reviewed data-access metadata and redaction-first proxy workload evidence. It never connects, executes SQL, or claims complete coverage.
+
+### Changed
+
+- **Shared execution and adapter boundaries.** Query execution now uses an injectable command runtime, SQL adapters share readiness and driver-error handling, and doctor accepts a non-SQL collector runtime. These internal changes keep the CLI behavior stable while making offline evidence and impact workflows testable.
+- **Documentation and agent skills cover the new surfaces.** English and Traditional Chinese Markdown and HTML guides, installed skill copies, and reference material document evidence packs, assert receipts, semantic contracts, and impact assessment.
+
+### Fixed
+
+- **Release audit no longer resolves a vulnerable `nanoid`.** The build-only PostCSS and Tailwind dependencies now resolve `nanoid@3.3.18`, satisfying the security advisory without adding a runtime dependency.
+
+## [1.52.1] - 2026-08-08 - Skill audit: correct claims, per-platform install, drift gate
+
+### Fixed
+
+- **The skill documented three flags that do not exist.** `q --use` (the global `dbcli --use <name> q` form is the real one), `q --collection` (a MongoDB snippet's `target:` is the only collection source), and `audit tail --recovery-ref` (it lives on `audit show`). The same wrong `audit tail --recovery-ref` is corrected in `docs/user` (en/zh, Markdown and HTML).
+- **The `design` artifact example failed its own validator.** It is replaced with one that validates clean, and the naming rules it has to satisfy are now documented: lowercase kebab-case for design names versus SQL identifiers for tables and columns, endpoints referencing a model rather than a table, strict objects, descriptions that must not contain SQL keywords, and the size limits. Structural violations report `INVALID_ARTIFACT`, a code the severity table did not list.
+- **Stale lists corrected against the CLI.** `--recovery` covers `lint` and `diff`; verification subject kinds include `table` (what `verify constraint` writes); the Redis permission table includes `XLEN` / `XREAD` / `XRANGE` / `XREVRANGE` / `XADD` / `XDEL` / `LREM`; `audit show` lists `--brief` and `--for-agent`; the snippet guard emits `LIMIT 1001`, fetching one extra row to detect truncation. Elasticsearch supports `q`, and the 10 000 bound belongs to `query` — `export --no-limit` streams via the scroll API. `schema --help` claimed `--sample-size` defaults to 50; it is 100.
+- **`design` was unreachable from the skill entry point.** `SKILL.md` did not mention it at all, so an agent asked to design or review a schema would hand-write DDL and bypass the review-only `propose` contract. It now has a command row, both workflows, and a guardrail that a proposed plan is never executed.
+- **Six drifts between the English and Traditional Chinese skills.** The most serious dropped the permission-tier semantics (multi-statement SQL rejected below `admin`, snippets free of write and DDL keywords, `$out` / `$merge` requiring `data-admin`). The MongoDB connection guidance also disagreed between languages — field-by-field is the recommendation, full URI the escape hatch — and the zh-TW side was missing the env-refs MongoDB exception, the `uri`-wins-silently gotcha, `--slow-ms` on `query` and `q`, and the `proxy analyze` action guidance.
+- **Cursor and Windsurf installs were the Claude skill verbatim.** Windsurf does not parse frontmatter, so roughly 900 characters of `description:` were read as rule text; Cursor reads `description` / `globs` / `alwaysApply` and received none of them. Both platforms keep `reference.md` outside the primary file's directory, so every mention of it resolved to nothing. `dbcli skill --install` now shapes the file per platform — Cursor as an Agent Requested rule, Windsurf with the frontmatter stripped and the description kept as prose — and repoints the reference path. Recognizing an existing dbcli install no longer depends on the frontmatter, so a Windsurf reinstall stops backing up dbcli's own file. See ADR 0006.
+
+### Added
+
+- **`reference.md` has an index.** 3300 lines previously offered no way in but a full read or a guessed grep, and the skill pointed at it in prose ("reference.md Redis section"). Those pointers are real anchors now, and a test derives the anchors from the headings so a renamed heading cannot break them silently.
+
+### Changed
+
+- **The bilingual parity gate compares content, not just shape.** It checked heading levels, fence counts, table rows, and a curated token list for mere presence — every drift above kept that structure intact. It now compares per-section counts of every code token and list item; run against 1.52.0 it reports 48 problems it used to pass. Its success message no longer reads as "the docs are aligned" when it only checked the skeleton.
+- **The skill entry point carries less that an agent cannot act on.** The always-loaded `description` drops from 990 to 626 characters with every trigger branch intact, release markers such as `(v1.23)` and an internal ticket id are gone (an agent has exactly one installed version), and the `proxy` row's flag wall becomes an anchor.
+
 ## [1.52.0] - 2026-08-07 - Offline database design assistant and slow-query hints
 
 ### Added