@foggy-projects/deepseek-harness-plugin 0.4.0-beta.15 → 0.4.0-beta.17

package/README.md

@@ -14,7 +14,7 @@ modify `PATH`, or register Python globally. Advanced users may explicitly set
 ## Local beta installation
 
 ```powershell
-dsh plugin --profile web add --workspace-root ./foggy-projects-deepseek-harness-plugin-0.4.0-beta.15.tgz
+dsh plugin --profile web add --workspace-root ./foggy-projects-deepseek-harness-plugin-0.4.0-beta.17.tgz
 ```
 
 Restart `dsh web`, use the browser it opens (or the complete printed URL,
@@ -27,6 +27,24 @@ After npm beta publication, the corresponding one-line install is:
 dsh plugin --profile web add --workspace-root @foggy-projects/deepseek-harness-plugin@beta
 ```
 
+DeepSeek Harness 0.1.2-rc.1 applies pnpm's minimum-release-age policy to its
+profile lockfile. An upgrade performed shortly after publication can therefore
+fail while naming either the new plugin or the previously installed Beta. This
+is a DSH profile-policy failure, not a damaged Foggy package. Use the exact
+profile directory printed in the error and rebuild its lockfile, then rerun the
+same `dsh plugin add` command:
+
+```powershell
+Set-Location "<the DSH profile directory printed in the error>"
+npx --yes pnpm@11.7.0 clean --lockfile
+```
+
+The failed add normally records the requested exact version under
+`minimumReleaseAgeExclude` before returning. If the error repeats, confirm that
+both the installed and requested exact Foggy versions named by the error appear
+under that key in the profile's `pnpm-workspace.yaml`; do not disable the policy
+globally.
+
 For development, `FOGGY_ASSET_CACHE_DIRS` can contain platform-delimited verified
 asset-cache directories. It is not required for a normal install.
 
@@ -50,7 +68,10 @@ Opaque CLI profiles remain available for users who prefer them and default to
 the private persistent `<dataRoot>/cli-profiles` directory, but they are not a
 prerequisite for ordinary development. Composite onboarding commands are
 idempotent: unchanged completed phases are resumed rather than re-adding a
-datasource or re-registering a local Bundle. Settings detects legacy temporary
+datasource or re-registering a local Bundle. A resumed datasource checkpoint is
+also reconciled against the live Runtime; if the datasource was removed outside
+the wrapper, only datasource configuration, verification, and schema discovery
+are resumed. Settings detects legacy temporary
 profiles and offers an explicit, validated migration into the persistent store.
 
 Private Python, CLI, Launcher, the analysis Skill, install state, and Runtime

package/docs/PUBLIC-BETA-READINESS.md

@@ -1,10 +1,10 @@
-# Public beta readiness: 0.4.0-beta.15
+# Public beta readiness: 0.4.0-beta.17
 
-Assessment date: 2026-09-04
+Assessment date: 2026-09-05
 
 ## Verdict
 
-`0.4.0-beta.15` is suitable for a scoped public beta on Windows 10/11 x64 and
+`0.4.0-beta.17` is suitable for a scoped public beta on Windows 10/11 x64 and
 Linux/WSL2 x64. It is not a general-availability release.
 
 The plugin package stays small and downloads large components only during
@@ -40,8 +40,10 @@ system PATH, registry, or a user's existing Python installation.
   inspected a SQLite datasource containing one table. Runtime retained the same
   PID, while the persisted onboarding state and command evidence contained no
   copy of the synthetic password.
-- The UI reported Python 3.12.13 as `Foggy private`, and CLI 0.1.23, Launcher
-  0.1.18, analysis Skill 0.1.17, and onboarding Skill 0.4.0-beta.15 as installed.
+- The package and settings contract report Python 3.12.13 as `Foggy private`,
+  CLI 0.1.23, Launcher 0.1.18, analysis Skill 0.1.17, and onboarding Skill
+  0.4.0-beta.17; the same component display was accepted in the preceding
+  native Windows beta run.
 - Runtime startup now reports the current phase, elapsed/timeout seconds, and a
   bounded progress value. A real Windows launch with Temurin 17 reached readiness
   and passed capabilities in 18.9 seconds. A controlled post-launch Java exit was
@@ -61,10 +63,25 @@ system PATH, registry, or a user's existing Python installation.
   downloaded artifacts are not trusted until the complete SHA256 matches.
 - Windows progress writes tolerate transient sharing violations without losing
   the previous valid progress document or leaving temporary files behind.
-- Current automated regression passes 20 Node tests and 20 Python tests. The
+- Current automated regression passes 20 Node tests and 25 Python tests. The
   new datasource tests cover secret-free persisted state, direct Runtime API
-  submission, CLI bypass for inline development credentials, and output
-  redaction.
+  submission, CLI bypass for inline development credentials, stable inline
+  credential resumption, bounded transient connection retry, and output redaction.
+- A real Java 17 / Launcher 0.1.18 regression resumed the same completed inline
+  datasource composite twice, retained the Runtime PID, and emitted a complete
+  `01` through `06` workspace evidence sequence without falling back to granular
+  commands.
+- A separate recovery regression completed a SQLite onboarding composite,
+  removed the managed datasource outside the wrapper, and reran the same
+  composite. The wrapper detected the missing live datasource, reset only the
+  three dependent checkpoints, recreated and tested the datasource, rebound the
+  namespace, and rediscovered the selected table successfully.
+- The full release rehearsal converted an inspected 126-column TMS order table
+  into versioned TM/QM source, passed eight development question-bank cases,
+  deployed the exact Git-tagged source to a separate Ubuntu host, published the
+  bundle, and passed the same question bank before and after a Runtime restart.
+  The remote Runtime retained its datasource binding, active bundle, query
+  model, and 11-tool MCP surface after restart.
 
 ## Beta boundaries
 

package/docs/WINDOWS-BETA-ACCEPTANCE.md

@@ -1,7 +1,7 @@
 # Native Windows Beta acceptance
 
 This runbook validates DeepSeek Harness `0.1.2-rc.1` with Foggy plugin
-`0.4.0-beta.15` on a clean 64-bit Windows 10 or Windows 11 machine. Use a local
+`0.4.0-beta.17` on a clean 64-bit Windows 10 or Windows 11 machine. Use a local
 directory that is not synchronized by OneDrive.
 
 ## Prerequisites
@@ -34,6 +34,14 @@ npx --yes --package=@deepseek-ai/dsh@0.1.2-rc.1 --package=pnpm@11.7.0 `
 npx --yes --package=@deepseek-ai/dsh@0.1.2-rc.1 --package=pnpm@11.7.0 dsh web
 ```
 
+If an immediate upgrade reports
+`ERR_PNPM_MINIMUM_RELEASE_AGE_VIOLATION`, use the exact DSH profile directory
+printed by the error, run `npx --yes pnpm@11.7.0 clean --lockfile` there, and
+repeat the same plugin-add command. The failed add normally records the new
+exact package version in that profile's `minimumReleaseAgeExclude`; if the
+error names an older installed Beta, confirm that exact version is listed too.
+Do not disable the policy globally.
+
 Let DSH open the browser automatically. If it does not, copy the complete URL
 printed by `dsh web`, including its `?token=...` query, into the browser. A new
 browser session cannot use the bare `http://127.0.0.1:3080/` URL until that
@@ -54,7 +62,7 @@ Expected component state:
 - CLI 0.1.23;
 - Launcher 0.1.18;
 - analysis Skill 0.1.17;
-- onboarding Skill 0.4.0-beta.15;
+- onboarding Skill 0.4.0-beta.17;
 - Java 17+ available;
 - native DSH Skill registration available.
 

package/experience/linux/prepare.sh

@@ -3,7 +3,7 @@ set -euo pipefail
 
 DSH_VERSION="0.1.2-rc.1"
 PNPM_VERSION="11.7.0"
-PLUGIN_VERSION="0.4.0-beta.15"
+PLUGIN_VERSION="0.4.0-beta.17"
 PLUGIN_REF="v${PLUGIN_VERSION}"
 PLUGIN_REPOSITORY="https://github.com/foggy-projects/foggy-deepseek-harness-plugin.git"
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@foggy-projects/deepseek-harness-plugin",
-  "version": "0.4.0-beta.15",
+  "version": "0.4.0-beta.17",
   "description": "Foggy Java data analysis engine integration for DeepSeek Harness",
   "type": "module",
   "main": "./lib/index.js",

package/skills/foggy-deepseek-onboarding/SKILL.md

@@ -62,6 +62,16 @@ are authoritative; do not copy the Skill into the current workspace.
 7. When the model works, recommend committing the model directory to the user's own Git repository.
    Do not initialize a repository, commit, push, or create a remote unless the user requests it.
 
+For a normal end-to-end request, take the documented composite-command path first. Do not inspect
+`onboarding.py`, enumerate CLI help, search installation directories, load the sales-drop example, or
+fetch online syntax documentation before trying that path. Use those troubleshooting or reference
+sources only when a concrete error or unsupported modeling requirement makes them relevant.
+
+Use the user's business terms to narrow schema discovery with `--pattern` or repeated `--table` when
+there are clear candidates. Otherwise keep the bounded default, then inspect only the few tables and
+columns needed for the requested model. Full results belong in the evidence files; retain concise
+summaries and paths in the conversation instead of repeatedly reading complete JSON artifacts.
+
 Read [references/onboarding-workflow.md](references/onboarding-workflow.md) for the wrapper commands and
 credential shapes. Use `foggy-ai-analysis` references for detailed TM/QM modeling and query tuning.
 

package/skills/foggy-deepseek-onboarding/assets/versions.json

@@ -1,6 +1,6 @@
 {
   "schemaVersion": "foggy-deepseek-onboarding-versions/v1",
-  "packageVersion": "0.4.0-beta.15",
+  "packageVersion": "0.4.0-beta.17",
   "validatedAt": "2026-09-04",
   "components": {
     "deepseekHarness": {

package/skills/foggy-deepseek-onboarding/references/onboarding-workflow.md

@@ -72,7 +72,18 @@ tables list -> bounded schema inspection
 ```
 
 It does not stop or restart Runtime. Repeated calls reuse completed checkpoints when the public
-connection contract is unchanged. Inline password values are excluded from the comparison.
+connection contract is unchanged. Before reusing a datasource checkpoint, the wrapper verifies that
+the named datasource still exists with the expected public type in the live Runtime. If it was
+removed outside the wrapper, datasource configuration, verification, and schema discovery resume
+automatically from the first affected step. Inline password values are excluded from the comparison.
+Transient connection-pool readiness timeouts are retried a small, bounded number of times; invalid
+credentials, invalid URLs, and other configuration errors fail immediately.
+
+When the user's request clearly identifies a business subject or likely tables, add `--pattern` or
+one or more `--table` arguments so discovery remains focused. Do not enumerate CLI help, inspect the
+wrapper source, load unrelated examples, or fetch online documentation on the successful normal path.
+Keep full command payloads in the numbered evidence files and use only their concise result fields in
+the conversation.
 
 After discovery, use `foggy-ai-analysis` to create TM/QM drafts from actual tables, columns, keys,
 small read-only samples when useful, and the user's business questions. Keep the files in the current

package/skills/foggy-deepseek-onboarding/scripts/invoke-onboarding.ps1

@@ -26,5 +26,7 @@ if (-not $pythonCommand -or -not (Test-Path -LiteralPath $pythonCommand)) {
     exit 1
 }
 
+$env:PYTHONUTF8 = '1'
+$env:PYTHONIOENCODING = 'utf-8'
 & $pythonCommand (Join-Path $PSScriptRoot 'onboarding.py') @OnboardingArgs
 exit $LASTEXITCODE

package/skills/foggy-deepseek-onboarding/scripts/onboarding.py

@@ -993,12 +993,15 @@ def validate_connection(payload: dict) -> dict:
 def persisted_connection(connection: dict) -> dict:
     """Return the resumable connection contract without an inline development password."""
     persisted = {key: value for key, value in connection.items() if key != "password"}
-    persisted["credentialMode"] = (
-        "inline-development" if connection.get("password") is not None
-        else "agent-environment" if connection.get("passwordEnv")
-        else "opaque-profile" if connection.get("connectionMode") == "opaque-profile"
-        else "none"
-    )
+    credential_mode = connection.get("credentialMode")
+    if credential_mode not in {"inline-development", "agent-environment", "opaque-profile", "none"}:
+        credential_mode = (
+            "inline-development" if connection.get("password") is not None
+            else "agent-environment" if connection.get("passwordEnv")
+            else "opaque-profile" if connection.get("connectionMode") == "opaque-profile"
+            else "none"
+        )
+    persisted["credentialMode"] = credential_mode
     return persisted
 
 
@@ -1797,6 +1800,50 @@ def cli_json(install_state: dict, runtime_state: dict, namespace: str, command:
     return payload
 
 
+TRANSIENT_DATASOURCE_TEST_MARKERS = (
+    "connection is not available",
+    "connection pool",
+    "request timed out",
+    "timeout waiting for connection",
+)
+
+
+def datasource_test_with_retry(
+    install_state: dict,
+    runtime_state: dict,
+    namespace: str,
+    datasource_name: str,
+    *,
+    attempts: int = 3,
+    initial_delay_seconds: float = 1.0,
+) -> dict:
+    """Retry only transient pool/readiness failures; authentication and configuration failures fail fast."""
+    for attempt in range(1, attempts + 1):
+        try:
+            result = cli_json(
+                install_state,
+                runtime_state,
+                namespace,
+                ["datasources", "test", datasource_name],
+                "datasources test",
+            )
+            if attempt > 1:
+                result = {
+                    **result,
+                    "onboardingRetry": {
+                        "attempts": attempt,
+                        "reason": "transient-datasource-readiness",
+                    },
+                }
+            return result
+        except OnboardingError as exc:
+            transient = any(marker in str(exc).lower() for marker in TRANSIENT_DATASOURCE_TEST_MARKERS)
+            if not transient or attempt == attempts:
+                raise
+            time.sleep(initial_delay_seconds * attempt)
+    raise AssertionError("unreachable")
+
+
 def runtime_api_json(
     runtime_state: dict,
     namespace: str,
@@ -2092,9 +2139,8 @@ def datasource_verify_command(args: argparse.Namespace) -> dict:
         raise OnboardingError("Datasource is not configured; run datasource-configure --apply first")
     connection = state["connection"]
     namespace = connection["namespace"]
-    tested = redact_connection_material(cli_json(
-        install_state, runtime_state, namespace,
-        ["datasources", "test", connection["name"]], "datasources test",
+    tested = redact_connection_material(datasource_test_with_retry(
+        install_state, runtime_state, namespace, connection["name"],
     ))
     if not args.bind:
         mark_step(state, "datasourceVerified", "waiting-for-binding", connectionTested=True)
@@ -2714,6 +2760,38 @@ def step_completed(state: dict, name: str) -> bool:
     return state.get("steps", {}).get(name, {}).get("status") == "completed"
 
 
+def reconcile_datasource_checkpoint(
+    state: dict,
+    install_state: dict,
+    data_root: Path,
+    runtime_state: dict,
+) -> dict | None:
+    """Reset dependent checkpoints when the live Runtime lost the datasource."""
+    if not step_completed(state, "datasourceConfigured"):
+        return None
+    connection = state["connection"]
+    listed = cli_json(
+        install_state,
+        runtime_state,
+        connection["namespace"],
+        ["datasources", "list"],
+        "datasources list for checkpoint reconciliation",
+    )
+    if matching_datasource(listed, connection) is not None:
+        return None
+    reset_steps = ("datasourceConfigured", "datasourceVerified", "schemaDiscovered")
+    for name in reset_steps:
+        mark_step(state, name, "pending", reason="runtime-datasource-missing")
+    state.get("artifacts", {}).pop("schemaDiscovery", None)
+    write_onboarding_state(data_root, state)
+    return {
+        "recovered": True,
+        "reason": "runtime-datasource-missing",
+        "resetSteps": list(reset_steps),
+        "dataSource": connection["name"],
+    }
+
+
 def bound_project_roots(state: dict) -> list[Path]:
     values = [state.get("projectRoot"), *state.get("workspaceBindings", [])]
     result: list[Path] = []
@@ -2746,7 +2824,7 @@ def bind_completed_workspace(state: dict, data_root: Path, project_root: Path) -
 
 
 def datasource_run_command(args: argparse.Namespace) -> dict:
-    _install_root, install_state, data_root, _runtime_state = onboarding_context(args, require_runtime=True)
+    _install_root, install_state, data_root, runtime_state = onboarding_context(args, require_runtime=True)
     project_root = normalized(args.project_root or Path.cwd())
     requested_connection = validate_connection(read_json_object(normalized(args.connection_file), "Connection plan"))
     resumable_connection = persisted_connection(requested_connection)
@@ -2788,6 +2866,9 @@ def datasource_run_command(args: argparse.Namespace) -> dict:
         ))
     save_composite_result(evidence_dir, "01-plan.json", plan_result, files)
     state = read_onboarding_state(data_root, profile)
+    checkpoint_recovery = reconcile_datasource_checkpoint(
+        state, install_state, data_root, runtime_state,
+    )
 
     if step_completed(state, "datasourceConfigured"):
         configured = resumed_phase(profile, "datasourceConfigured")
@@ -2798,6 +2879,8 @@ def datasource_run_command(args: argparse.Namespace) -> dict:
             install_root=args.install_root, data_root=args.data_root, profile=profile, apply=False, replace=False,
             runtime_password=requested_connection.get("password"),
         ))
+        if checkpoint_recovery:
+            configure_dry["checkpointRecovery"] = checkpoint_recovery
         save_composite_result(evidence_dir, "02-datasource-dry.json", configure_dry, files)
         if not args.approve_configure:
             return {
@@ -2814,6 +2897,8 @@ def datasource_run_command(args: argparse.Namespace) -> dict:
             install_root=args.install_root, data_root=args.data_root, profile=profile, apply=True, replace=False,
             runtime_password=requested_connection.get("password"),
         ))
+        if checkpoint_recovery:
+            configured["checkpointRecovery"] = checkpoint_recovery
         save_composite_result(evidence_dir, "03-datasource-apply.json", configured, files)
         state = read_onboarding_state(data_root, profile)
 
@@ -3425,6 +3510,10 @@ def build_parser() -> argparse.ArgumentParser:
 
 
 def main() -> None:
+    for stream in (sys.stdout, sys.stderr):
+        reconfigure = getattr(stream, "reconfigure", None)
+        if callable(reconfigure):
+            reconfigure(encoding="utf-8", errors="replace")
     args = build_parser().parse_args()
     try:
         result = args.handler(args)