akm-cli 0.9.4 → 0.9.6

package/dist/workflows/runtime/runs.js

@@ -396,10 +396,9 @@ export function isTruncatedEvidence(value) {
  * `buildEvidence` (exec/step-work.ts) promotes `evidence.output` UNCLIPPED by
  * design: gates judge the full promoted artifact and the in-memory
  * {@link StepExecutionResult} carries it to the caller intact. Nothing bounded
- * the PERSISTED form, though — a `collect` reducer over a fan-out capped only by
- * `WORKFLOW_MAX_MAP_EXPANSION` (10 000 units) can serialize to hundreds of
- * megabytes. This is the write boundary, so the bound lives here rather than in
- * the shared step-semantics module.
+ * the PERSISTED form, though — a `collect` reducer over an unbounded fan-out
+ * can serialize to hundreds of megabytes. This is the write boundary, so the
+ * bound lives here rather than in the shared step-semantics module.
  *
  * Over-cap values are REPLACED (largest top-level entry BY UTF-8 BYTES first —
  * the unit the cap is measured in — until the row fits) with a

package/dist/workflows/source-ir/schema.js

@@ -14,7 +14,7 @@ import { validateExtraParams } from "../../core/extra-params.js";
 import { checkJsonSchemaDefinition } from "../../core/json-schema.js";
 import { parseReference } from "../program/expressions.js";
 import { PROGRAM_RETRY_REASONS } from "../program/schema.js";
-import { jsonBytes, utf8Bytes, WORKFLOW_ENGINE_NAME_PATTERN, WORKFLOW_ENV_VAR_NAME_PATTERN, WORKFLOW_MAX_CONCURRENCY, WORKFLOW_MAX_ENGINE_NAME_LENGTH, WORKFLOW_MAX_EXEC_ARG_BYTES, WORKFLOW_MAX_EXEC_ARGV, WORKFLOW_MAX_EXEC_CWD_LENGTH, WORKFLOW_MAX_EXEC_PASS_ENV, WORKFLOW_MAX_EXTRA_PARAMS_BYTES, WORKFLOW_MAX_GATE_LOOPS, WORKFLOW_MAX_INPUTS, WORKFLOW_MAX_MAP_EXPANSION, WORKFLOW_MAX_OUTPUTS, WORKFLOW_MAX_PARAMS, WORKFLOW_MAX_RETRIES, WORKFLOW_MAX_ROUTE_BRANCHES, WORKFLOW_MAX_SCHEMA_BYTES, WORKFLOW_MAX_TIMEOUT_MS, } from "../resource-limits.js";
+import { jsonBytes, WORKFLOW_ENGINE_NAME_PATTERN, WORKFLOW_ENV_VAR_NAME_PATTERN, WORKFLOW_MAX_CONCURRENCY, WORKFLOW_MAX_ENGINE_NAME_LENGTH, WORKFLOW_MAX_EXTRA_PARAMS_BYTES, WORKFLOW_MAX_SCHEMA_BYTES, WORKFLOW_MAX_TIMEOUT_MS, } from "../resource-limits.js";
 import { compareWorkflowSourceCodePoints } from "./compare.js";
 import { canonicalizeWorkflowCron, canonicalizeWorkflowRun, canonicalizeWorkflowWorkingDirectory, classifyWorkflowStepUses, rejectNulInArgv, validateWorkflowBuiltinCommand, WorkflowSourceSemanticError, } from "./semantics.js";
 export const WORKFLOW_SOURCE_IR_VERSION = 1;
@@ -255,7 +255,7 @@ function validateStep(value, jobId, index, stepIds, options) {
     }
     validateMap(step.map, `step ${id} map`);
     validateRoute(step.route, `step ${id} route`);
-    optionalStringList(step.inputs, `step ${id} inputs`, WORKFLOW_MAX_INPUTS);
+    optionalStringList(step.inputs, `step ${id} inputs`, Infinity);
     if (Array.isArray(step.inputs)) {
         for (const [inputIndex, input] of step.inputs.entries())
             validateReference(input, `step ${id} inputs[${inputIndex}]`);
@@ -277,12 +277,7 @@ function validateExec(value, location, options) {
         return;
     const exec = record(value, location);
     keys(exec, ["command", "cwd", "passEnv"], location);
-    stringList(exec.command, `${location}.command`, WORKFLOW_MAX_EXEC_ARGV, false);
-    for (const [index, argument] of exec.command.entries()) {
-        if (utf8Bytes(argument) > WORKFLOW_MAX_EXEC_ARG_BYTES) {
-            fail(`${location}.command[${index}] exceeds ${WORKFLOW_MAX_EXEC_ARG_BYTES} bytes`);
-        }
-    }
+    stringList(exec.command, `${location}.command`, Infinity, false);
     try {
         rejectNulInArgv(exec.command);
     }
@@ -291,9 +286,6 @@ function validateExec(value, location, options) {
     }
     optionalString(exec.cwd, `${location}.cwd`);
     if (exec.cwd !== undefined) {
-        if (exec.cwd.length > WORKFLOW_MAX_EXEC_CWD_LENGTH) {
-            fail(`${location}.cwd exceeds ${WORKFLOW_MAX_EXEC_CWD_LENGTH} characters`);
-        }
         try {
             exec.cwd = canonicalizeWorkflowWorkingDirectory(exec.cwd, options.workspaceRoot);
         }
@@ -302,7 +294,7 @@ function validateExec(value, location, options) {
         }
     }
     if (exec.passEnv !== undefined) {
-        stringList(exec.passEnv, `${location}.passEnv`, WORKFLOW_MAX_EXEC_PASS_ENV, false);
+        stringList(exec.passEnv, `${location}.passEnv`, Infinity, false);
     }
     if (Array.isArray(exec.passEnv)) {
         for (const name of exec.passEnv) {
@@ -359,8 +351,8 @@ function validateRetry(value, location) {
         return;
     const retry = record(value, location);
     keys(retry, ["max", "on"], location);
-    if (!Number.isSafeInteger(retry.max) || retry.max < 0 || retry.max > WORKFLOW_MAX_RETRIES) {
-        fail(`${location}.max must be an integer from 0 through ${WORKFLOW_MAX_RETRIES}`);
+    if (!Number.isSafeInteger(retry.max) || retry.max < 0) {
+        fail(`${location}.max must be a non-negative integer`);
     }
     stringList(retry.on, `${location}.on`, PROGRAM_RETRY_REASONS.length, false);
     for (const reason of retry.on) {
@@ -393,10 +385,8 @@ function validateRoute(value, location) {
     keys(route, ["input", "branches", "defaultStepId"], location);
     nonEmptyString(route.input, `${location}.input`);
     validateReference(route.input, `${location}.input`);
-    if (!Array.isArray(route.branches) ||
-        route.branches.length === 0 ||
-        route.branches.length > WORKFLOW_MAX_ROUTE_BRANCHES) {
-        fail(`${location}.branches must contain 1 through ${WORKFLOW_MAX_ROUTE_BRANCHES} entries`);
+    if (!Array.isArray(route.branches) || route.branches.length === 0) {
+        fail(`${location}.branches must be a non-empty array`);
     }
     const matches = new Set();
     for (const [index, value] of route.branches.entries()) {
@@ -416,11 +406,8 @@ function validateGate(value, location) {
         return;
     const gate = record(value, location);
     keys(gate, ["maxLoops", "rubric"], location);
-    if (gate.maxLoops !== undefined &&
-        (!Number.isSafeInteger(gate.maxLoops) ||
-            gate.maxLoops < 1 ||
-            gate.maxLoops > WORKFLOW_MAX_GATE_LOOPS)) {
-        fail(`${location}.maxLoops must be an integer from 1 through ${WORKFLOW_MAX_GATE_LOOPS}`);
+    if (gate.maxLoops !== undefined && (!Number.isSafeInteger(gate.maxLoops) || gate.maxLoops < 1)) {
+        fail(`${location}.maxLoops must be an integer of at least 1`);
     }
     optionalString(gate.rubric, `${location}.rubric`);
 }
@@ -453,8 +440,8 @@ function validateParams(value) {
     if (value === undefined)
         return;
     const params = record(value, "params");
-    if (Object.keys(params).length === 0 || Object.keys(params).length > WORKFLOW_MAX_PARAMS) {
-        fail(`params must contain 1 through ${WORKFLOW_MAX_PARAMS} entries`);
+    if (Object.keys(params).length === 0) {
+        fail(`params must contain at least one entry`);
     }
     for (const [name, schema] of Object.entries(params)) {
         if (!/^[A-Za-z_][A-Za-z0-9_]*$/.test(name))
@@ -472,8 +459,8 @@ function validateOutputs(value) {
     if (value === undefined)
         return;
     const outputs = record(value, "outputs");
-    if (Object.keys(outputs).length === 0 || Object.keys(outputs).length > WORKFLOW_MAX_OUTPUTS) {
-        fail(`outputs must contain 1 through ${WORKFLOW_MAX_OUTPUTS} entries`);
+    if (Object.keys(outputs).length === 0) {
+        fail(`outputs must contain at least one entry`);
     }
     for (const [name, declaration] of Object.entries(outputs)) {
         if (!/^[A-Za-z_][A-Za-z0-9_]*$/.test(name))
@@ -496,9 +483,6 @@ function validateBudget(value) {
             fail(`budget.${key} must be a positive integer`);
         }
     }
-    if (typeof budget.maxUnits === "number" && budget.maxUnits > WORKFLOW_MAX_MAP_EXPANSION) {
-        fail(`budget.maxUnits must be at most ${WORKFLOW_MAX_MAP_EXPANSION}`);
-    }
 }
 function validateLlm(value, location) {
     if (value === undefined)

package/dist/workflows/validator.js

@@ -16,7 +16,7 @@
  * resource limits).
  */
 import { bundleRefToString, parseBundleRef } from "../core/asset/asset-ref.js";
-import { utf8Bytes, WORKFLOW_MAX_INSTRUCTION_BYTES, WORKFLOW_MAX_PARAMS, WORKFLOW_MAX_STEPS } from "./resource-limits.js";
+import { utf8Bytes, WORKFLOW_MAX_INSTRUCTION_BYTES } from "./resource-limits.js";
 export function runSemanticChecks(draft, frontmatterData, frontmatterEndLine, errors) {
     checkXrefs(frontmatterData.xrefs, frontmatterEndLine, errors);
     checkResourceLimits(draft, errors);
@@ -43,12 +43,6 @@ function checkXrefs(value, line, errors) {
     }
 }
 function checkResourceLimits(draft, errors) {
-    if (draft.steps.length > WORKFLOW_MAX_STEPS) {
-        errors.push({ line: 1, message: `Workflow must contain at most ${WORKFLOW_MAX_STEPS} steps.` });
-    }
-    if (Object.keys(draft.params ?? {}).length > WORKFLOW_MAX_PARAMS) {
-        errors.push({ line: 1, message: `Workflow must contain at most ${WORKFLOW_MAX_PARAMS} parameters.` });
-    }
     for (const step of draft.steps) {
         if (step.instructions && utf8Bytes(step.instructions.text) > WORKFLOW_MAX_INSTRUCTION_BYTES) {
             errors.push({

package/docs/reference/cli.md

@@ -2423,9 +2423,10 @@ shell commands. It manages on-disk task definitions under
 (cron / launchd / schtasks). Task source v4 YAML (`version: 4`) is the only
 executable source contract this release accepts; `akm task add` writes v4 —
 see the canonical [Tasks reference](tasks.md). The
-group is `add | run | explain | sync | doctor | history` — there is no `list`
-or `remove`; use `akm search --type task` / `akm show tasks/<id>` to inspect,
-and edit the file + `akm task sync` to change or remove a schedule.
+group is `add | run | explain | sync | doctor | history | prune` — there is
+no `list` or `remove`; use `akm search --type task` / `akm show tasks/<id>`
+to inspect, and edit the file + `akm task sync` to change or remove a
+schedule.
 
 ```sh
 akm search --type task                      # List tasks (cross-bundle)
@@ -2439,8 +2440,12 @@ akm task run <id>                           # Execute now (what the scheduler ca
 akm task explain <ref>                      # Read-only: declared inputs, target, schedule — spawns nothing
 akm task history [--id <id>] [--limit <n>]  # Recent runs from state.db
 akm task sync                               # Reconcile on-disk YAML with scheduler
+akm task sync --dry-run                     # Preview the reconcile — zero scheduler writes
 akm task sync --rebind                      # Also capture the current installed runtime
 akm task doctor                             # Report scheduler backend + paths
+akm task prune                              # Preview orphaned scheduler entries — zero writes
+akm task prune --yes                        # Remove every currently-computed orphan
+akm task prune --id ghost,stale --yes       # Remove only the named orphan ids
 ```
 
 `task add` also accepts `--disabled` (register but leave off in the OS
@@ -2466,6 +2471,23 @@ schedule-binding) and run `akm task sync`. To remove one, delete its file
 (`<bundle>/tasks/<id>.yml`) and run `akm task sync` — sync uninstalls the
 orphaned scheduler entry.
 
+`akm task sync --dry-run` prints the planned adds/updates/removes (removals
+carry their owning bundle) without touching the scheduler — zero writes.
+Exits non-zero when removals are pending, so it can gate a CI/health check
+on "sync would change something."
+
+`akm task prune` reclaims installed scheduler entries that `sync` can never
+clean up on its own: entries whose own `--scheduler-context` descriptor no
+longer resolves to a live bundle (a corrupt/missing descriptor, or the
+bundle directory it pointed at is gone). It never touches an entry that
+still resolves to a live bundle — that's `sync`'s job. Like `sync
+--dry-run`, the default is a dry-run preview (zero scheduler writes) that
+exits non-zero when there are candidates to remove; `--yes` executes the
+printed plan, and `--id <id1,id2,...>` narrows a `--yes` run (or a preview)
+to specific binding ids — naming an id that isn't a current orphan
+candidate (not installed, or it still resolves to a live bundle) is
+refused with a usage error and removes nothing.
+
 Scheduler activation captures the installed akm runtime. Ordinary `task sync`
 reconciles definitions, schedules, and enabled state while preserving that
 runtime binding. Use `task sync --rebind` only after intentionally moving or

package/docs/reference/configuration.md

@@ -7,12 +7,33 @@ directory. Project `.akm/config.json` files are not merged.
 
 ## Version 0.9
 
-A present configuration file must set `configVersion` to exactly `"0.9.0"`.
-Missing, older, newer, numeric, and malformed versions are rejected by ordinary
-commands without rewriting the file. Pre-0.9 config and database layouts are
-not runtime inputs and are not migrated by `akm upgrade`. Configure the current
-schema directly. The standalone migrator exists only for explicit task
-migration: task v2 to task v3, then task v3 to task source v4, in one pass.
+A present configuration file must set `configVersion` to a version this
+binary knows: the current `"0.9.0"`, or a known older version it can
+auto-upgrade in memory (see "Version read shim" below). Missing, newer,
+numeric, and any other unrecognized version are rejected by ordinary
+commands without rewriting the file — an older binary never guesses at a
+newer, unknown shape. Pre-0.9 config and database layouts are not runtime
+inputs and are not migrated by `akm upgrade`. Configure the current schema
+directly. The standalone migrator exists only for explicit task migration:
+task v2 to task v3, then task v3 to task source v4, in one pass.
+
+### Version read shim
+
+Like the task-source v2/v3 auto-shim (`akm migrate apply`'s in-memory
+counterpart, documented under Migration below), a known older `configVersion`
+is converted to the current shape in memory on load — with a one-line stderr
+deprecation warning — rather than hard-failing every command. Nothing is
+written back to disk by the shim itself; the very next config-mutating
+command (`akm config set`, etc.) persists the upgrade for free, since every
+config write already forces `configVersion` to the current value, which
+silences the warning. A `configVersion` this binary does not recognize at
+all — including anything newer than current — still fails closed with
+`UNSUPPORTED_CONFIG_VERSION`.
+
+As of this writing `"0.9.0"` is the only `configVersion` akm has ever
+shipped, so there is no real older shape for the shim to convert yet; the
+mechanism (`src/core/config/config-version-shim.ts`) is established ahead of
+the first bump that will need it, per #863.
 
 ```jsonc
 {
@@ -345,7 +366,7 @@ reference, same rules as engine `apiKey`), `dimension`, `localModel`,
 
 | Key | Purpose |
 | --- | --- |
-| `search.graphBoost.*` | Entity-graph relevance boost: `directBoostPerEntity`/`directBoostCap` (directly related entities), `hopBoostPerEntity`/`hopBoostCap` (multi-hop, capped at `maxHops` ≤ 3), `confidenceMode` (`off`\|`blend`\|`multiply`, default `blend`), `confidenceWeight` (0–1, default `0.2`) |
+| `search.graphBoost.*` | Entity-graph relevance boost: `directBoostPerEntity`/`directBoostCap` (directly related entities), `hopBoostPerEntity`/`hopBoostCap` (multi-hop, capped at `maxHops` ≤ 3), `confidenceMode` (`blend`, the only supported value), `confidenceWeight` (0–1, default `0.2`) |
 
 ## Feedback
 

package/docs/reference/tasks.md

@@ -389,6 +389,16 @@ for full before/after examples and recovery guidance.
 - Disable a binding by editing the source and syncing: set that schedule
   entry's own `enabled: false`.
 - Delete the `.yml` source and sync to remove its derived binding(s).
+- `akm task sync --dry-run` previews the reconcile (adds/updates/removes,
+  removals annotated with their owning bundle) without writing to the
+  scheduler; exits non-zero when removals are pending.
+- `akm task prune` removes installed scheduler entries `sync` cannot reach
+  because their own descriptor no longer resolves to a live bundle
+  (corrupt/missing `--scheduler-context`, or the owning bundle directory is
+  gone). It never touches an entry that still resolves to a live bundle.
+  Defaults to a dry-run preview (zero writes); `--yes` executes it; `--id
+  <id1,id2,...>` scopes to specific ids and refuses any id that isn't a
+  current orphan candidate.
 - Use `akm task sync --rebind` only when deliberately changing the captured
   AKM runtime, then verify with `akm task doctor`.
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "akm-cli",
-  "version": "0.9.4",
+  "version": "0.9.6",
   "type": "module",
   "description": "akm (Agent Knowledge Manager) — a portable, local-first capability library for AI agents. Discover, load, share, and improve reusable skills, scripts, workflows, and knowledge across any shell-capable coding agent, including Claude Code, OpenCode, and Cursor.",
   "keywords": [

package/schemas/akm-config.json

@@ -57,9 +57,6 @@
                 "type": "integer",
                 "exclusiveMinimum": 0
               },
-              "supportsJsonSchema": {
-                "type": "boolean"
-              },
               "extraParams": {
                 "type": "object",
                 "additionalProperties": {}
@@ -572,9 +569,7 @@
             "confidenceMode": {
               "type": "string",
               "enum": [
-                "off",
-                "blend",
-                "multiply"
+                "blend"
               ],
               "default": "blend"
             },
@@ -1747,9 +1742,6 @@
                     "type": "integer",
                     "exclusiveMinimum": 0
                   },
-                  "supportsJsonSchema": {
-                    "type": "boolean"
-                  },
                   "extraParams": {
                     "type": "object",
                     "additionalProperties": {}
@@ -2262,9 +2254,7 @@
                 "confidenceMode": {
                   "type": "string",
                   "enum": [
-                    "off",
-                    "blend",
-                    "multiply"
+                    "blend"
                   ],
                   "default": "blend"
                 },
@@ -3420,14 +3410,8 @@
           "type": "integer",
           "exclusiveMinimum": 0
         },
-        "capabilities": {
-          "type": "object",
-          "properties": {
-            "structuredOutput": {
-              "type": "boolean"
-            }
-          },
-          "additionalProperties": true
+        "supportsJsonSchema": {
+          "type": "boolean"
         },
         "extraParams": {
           "type": "object",
@@ -3443,9 +3427,6 @@
         "reasoningEffort": {
           "type": "string",
           "minLength": 1
-        },
-        "supportsJsonSchema": {
-          "type": "boolean"
         }
       },
       "required": [

package/schemas/akm-task.json

@@ -27,7 +27,6 @@
     },
     "inputs": {
       "type": "object",
-      "maxProperties": 128,
       "propertyNames": { "$ref": "#/definitions/taskInputName" },
       "additionalProperties": { "$ref": "#/definitions/inputDeclaration" }
     },
@@ -105,7 +104,7 @@
       "items": { "type": "string", "pattern": "^[A-Za-z_][A-Za-z0-9_]*$" }
     },
     "maxSteps": { "type": "integer", "minimum": 1, "maximum": 9007199254740991 },
-    "maxRetries": { "type": "integer", "minimum": 0, "maximum": 100 }
+    "maxRetries": { "type": "integer", "minimum": 0 }
   },
   "oneOf": [
     {

package/schemas/akm-workflow.json

@@ -22,7 +22,6 @@
     "params": {
       "type": "object",
       "minProperties": 1,
-      "maxProperties": 128,
       "description": "Run parameters: param name -> JSON Schema declaration. Names must be params.<name>-addressable identifiers.",
       "propertyNames": {
         "pattern": "^[A-Za-z_][A-Za-z0-9_]*$"
@@ -34,7 +33,6 @@
     "outputs": {
       "type": "object",
       "minProperties": 1,
-      "maxProperties": 64,
       "description": "Named, optionally schema-validated projections of step artifacts, exported when the run completes. Names must be steps.<child>.output.<name>-addressable identifiers. Each entry: { from: steps.<id>.output(.<seg>)*, schema?: <bounded JSON Schema> }. Markdown-frontmatter only.",
       "propertyNames": {
         "pattern": "^[A-Za-z_][A-Za-z0-9_]*$"
@@ -52,7 +50,6 @@
     "steps": {
       "type": "array",
       "minItems": 1,
-      "maxItems": 256,
       "items": {
         "$ref": "#/definitions/step"
       }
@@ -188,7 +185,6 @@
         "max": {
           "type": "integer",
           "minimum": 0,
-          "maximum": 100,
           "description": "Maximum retry attempts per unit."
         },
         "on": {
@@ -213,7 +209,6 @@
         "max_units": {
           "type": "integer",
           "minimum": 1,
-          "maximum": 10000,
           "description": "Ceiling on total dispatched units across the run (seeded from journaled unit rows)."
         }
       }
@@ -252,25 +247,21 @@
           "type": "array",
           "description": "argv; command[0] is the program, resolved through PATH. Never shell-parsed.",
           "minItems": 1,
-          "maxItems": 64,
           "items": {
             "type": "string",
-            "minLength": 1,
-            "maxLength": 4096
+            "minLength": 1
           }
         },
         "cwd": {
           "type": "string",
           "description": "Optional RELATIVE working directory inside the unit's working directory (its fresh worktree under `isolation: worktree`). Absolute paths, Windows drive letters, \"~\", and \"..\" segments are rejected here, and containment is re-checked against the resolved base — symlinks included — before the command is spawned.",
           "minLength": 1,
-          "maxLength": 1024,
           "pattern": "^(?!/|\\\\|~|[A-Za-z]:)(?!.*(?:^|[/\\\\])\\.\\.(?:[/\\\\]|$)).+$"
         },
         "pass_env": {
           "type": "array",
           "description": "Extra environment variable NAMES copied through from akm's own environment on top of the default allowlist — for a per-machine toolchain variable (e.g. CARGO_HOME) that no committed `env:` asset could carry. NAMES ONLY: values never appear in a workflow, and unlike `env:` bindings these values are NOT redacted from the command's output, so never list a credential here.",
           "minItems": 1,
-          "maxItems": 32,
           "uniqueItems": true,
           "items": {
             "type": "string",
@@ -398,7 +389,6 @@
         "when": {
           "type": "array",
           "minItems": 1,
-          "maxItems": 256,
           "items": {
             "$ref": "#/definitions/routeBranch"
           }
@@ -416,7 +406,6 @@
         "max_loops": {
           "type": "integer",
           "minimum": 1,
-          "maximum": 100,
           "description": "Evaluator-optimizer loop bound."
         }
       }
@@ -424,7 +413,6 @@
     "inputs": {
       "type": "array",
       "minItems": 1,
-      "maxItems": 64,
       "uniqueItems": true,
       "items": {
         "$ref": "#/definitions/reference"

package/dist/indexer/search/semantic-status.js

@@ -1,142 +0,0 @@
-// This Source Code Form is subject to the terms of the Mozilla Public
-// License, v. 2.0. If a copy of the MPL was not distributed with this
-// file, You can obtain one at https://mozilla.org/MPL/2.0/.
-import fs from "node:fs";
-import { writeFileAtomic } from "../../core/common.js";
-import { getCacheDir, getSemanticStatusPath } from "../../core/paths.js";
-import { DETERMINISTIC_EMBED_MODEL_ID, isDeterministicEmbedEnabled } from "../../llm/embedders/deterministic.js";
-import { DEFAULT_LOCAL_MODEL } from "../../llm/embedders/local.js";
-let semanticStatusMutationForTests;
-/** TEST-ONLY. Inject a semantic-status filesystem publication fault. */
-export function _setSemanticStatusMutationForTests(mutation) {
-    semanticStatusMutationForTests = mutation;
-}
-export function deriveSemanticProviderFingerprint(embedding) {
-    if (isDeterministicEmbedEnabled()) {
-        return `deterministic:${DETERMINISTIC_EMBED_MODEL_ID}`;
-    }
-    if (embedding?.endpoint) {
-        // Fingerprint keys on vector identity only (model + dimension). The endpoint
-        // is transport/routing and has no bearing on vector compatibility, so moving
-        // the same model+dimension to a different host must not force a full re-embed.
-        return `remote:${embedding.model}|${embedding.dimension ?? "default"}`;
-    }
-    return `local:${embedding?.localModel ?? DEFAULT_LOCAL_MODEL}`;
-}
-export function readSemanticStatus() {
-    try {
-        const raw = JSON.parse(fs.readFileSync(getSemanticStatusPath(), "utf8"));
-        if ((raw.status === "pending" ||
-            raw.status === "ready-js" ||
-            raw.status === "ready-vec" ||
-            raw.status === "blocked") &&
-            typeof raw.providerFingerprint === "string" &&
-            typeof raw.lastCheckedAt === "string") {
-            const status = {
-                status: raw.status,
-                providerFingerprint: raw.providerFingerprint,
-                lastCheckedAt: raw.lastCheckedAt,
-            };
-            if (typeof raw.reason === "string")
-                status.reason = raw.reason;
-            if (typeof raw.message === "string")
-                status.message = raw.message;
-            if (typeof raw.entryCount === "number")
-                status.entryCount = raw.entryCount;
-            if (typeof raw.embeddingCount === "number")
-                status.embeddingCount = raw.embeddingCount;
-            return status;
-        }
-    }
-    catch {
-        // ignore corrupt or missing semantic status
-    }
-    return undefined;
-}
-export function writeSemanticStatus(status) {
-    semanticStatusMutationForTests?.("write");
-    const dir = getCacheDir();
-    fs.mkdirSync(dir, { recursive: true });
-    writeFileAtomic(getSemanticStatusPath(), `${JSON.stringify(status, null, 2)}\n`);
-}
-export function clearSemanticStatus() {
-    semanticStatusMutationForTests?.("clear");
-    try {
-        fs.unlinkSync(getSemanticStatusPath());
-    }
-    catch {
-        // ignore missing file
-    }
-}
-/** How long a "blocked" status is retained before the system retries. 24 hours. */
-export const BLOCKED_TTL_MS = 24 * 60 * 60 * 1000;
-export function getEffectiveSemanticStatus(config, status = readSemanticStatus()) {
-    if (config.semanticSearchMode === "off")
-        return "disabled";
-    if (!status)
-        return "pending";
-    const fingerprint = deriveSemanticProviderFingerprint(config.embedding);
-    if (status.providerFingerprint !== fingerprint)
-        return "pending";
-    // Auto-recovery: if blocked status is older than BLOCKED_TTL_MS, treat as pending
-    // so the next index run will re-attempt semantic setup.
-    if (status.status === "blocked") {
-        const checkedAt = new Date(status.lastCheckedAt).getTime();
-        if (Number.isNaN(checkedAt) || Date.now() - checkedAt > BLOCKED_TTL_MS) {
-            return "pending";
-        }
-    }
-    return status.status;
-}
-export function isSemanticRuntimeReady(status) {
-    return status === "ready-js" || status === "ready-vec";
-}
-export function classifySemanticFailure(message) {
-    const lower = message.toLowerCase();
-    if (lower.includes("401") || lower.includes("403") || lower.includes("auth") || lower.includes("unauthorized")) {
-        return "remote-auth";
-    }
-    if (lower.includes("429") || lower.includes("rate limit") || lower.includes("quota")) {
-        return "remote-rate-limit";
-    }
-    if (lower.includes("eacces") || lower.includes("permission denied")) {
-        return "permission-denied";
-    }
-    // Native library / linker errors must be checked before the generic ONNX
-    // match because Alpine/musl linker errors often contain "onnxruntime" in
-    // the library path (e.g. onnxruntime_binding.node).
-    if (lower.includes("shared library") ||
-        lower.includes("glibc") ||
-        lower.includes("musl") ||
-        lower.includes("libc.so")) {
-        return "native-lib-missing";
-    }
-    if (lower.includes("onnx") || lower.includes("onnxruntime")) {
-        return "onnx-runtime-failed";
-    }
-    if (lower.includes("404") || lower.includes("model not found") || lower.includes("bad request")) {
-        return "remote-model";
-    }
-    if (lower.includes("transformers") ||
-        lower.includes("missing-package") ||
-        lower.includes("local embedding dependency is unavailable")) {
-        return "missing-package";
-    }
-    if (lower.includes("download")) {
-        return "local-model-download";
-    }
-    if (lower.includes("dimension mismatch")) {
-        return "dimension-mismatch";
-    }
-    if (lower.includes("db") || lower.includes("sqlite") || lower.includes("cache dir")) {
-        return "db-open";
-    }
-    if (lower.includes("timeout") ||
-        lower.includes("unreachable") ||
-        lower.includes("refused") ||
-        lower.includes("network") ||
-        lower.includes("fetch")) {
-        return "remote-network";
-    }
-    return "unknown";
-}