dsh-harbor-evolution 0.9.2 → 0.9.3

package/README.md

@@ -2,7 +2,7 @@
 
 Installable DeepSeek Harness Plugin + Skill for running stable Harbor evaluation and controlled Agent evolution loops, with a native DSH Web dashboard.
 
-The package gives DSH nineteen strict Harbor tools, dedicated Tool cards, an object-first Evaluation Workbench, a same-session Copilot Dock, an installation Doctor, and the model- and user-invocable `evolve-agent-with-harbor` Skill. The Skill starts with four user-facing concepts—Dataset (what to test), Generator (who answers), Evaluator plus criteria (what good means), and Optimizer (who improves it)—then compiles accepted choices into the strict Evaluation Stack. When no Dataset is supplied, it can instead preview recent completed DSH Sessions and evaluate each immutable Session as one Historical Trial without rerunning a Candidate. A DSH Generator may explicitly pin the current default model as a non-secret Candidate identity while retaining the per-Job Host Broker credential boundary. The Plugin validates Dataset identity, checks Trial Lifecycle and Score Validity, governs independent Ground Truth meta-evaluation, diagnoses evidence provenance, limits each iteration to one controlled Candidate change, and invokes the Promotion Gate only as an explicit action.
+The package gives DSH nineteen strict Harbor tools, native Tool cards in the same-session conversation, an object-first Evaluation Workbench, an installation Doctor, and the model- and user-invocable `evolve-agent-with-harbor` Skill. The Skill starts with four user-facing concepts—Dataset (what to test), Generator (who answers), Evaluator plus criteria (what good means), and Optimizer (who improves it)—then compiles accepted choices into the strict Evaluation Stack. When no Dataset is supplied, it can instead preview recent completed DSH Sessions and evaluate each immutable Session as one Historical Trial without rerunning a Candidate. A DSH Generator may explicitly pin the current default model as a non-secret Candidate identity while retaining the per-Job Host Broker credential boundary. The Plugin validates Dataset identity, checks Trial Lifecycle and Score Validity, governs independent Ground Truth meta-evaluation, diagnoses evidence provenance, limits each iteration to one controlled Candidate change, and invokes the Promotion Gate only as an explicit action.
 
 ## Install
 
@@ -14,8 +14,8 @@ npx --yes dsh-harbor-evolution@latest setup --project-root "$PWD"
 
 The setup command installs both required runtimes:
 
-- `harbor-dsh-evolution==0.9.2` in a managed Python environment.
-- `dsh-harbor-evolution@0.9.2` in the selected DSH profile.
+- `harbor-dsh-evolution==0.9.3` in a managed Python environment.
+- `dsh-harbor-evolution@0.9.3` in the selected DSH profile.
 
 It then stores the absolute Harbor executable paths and a fallback `projectRoot` in the profile's `harbor-evolution` block and verifies the integration. Agent Tool calls always use the calling session's absolute working directory as their project root; the configured value remains the Web Workbench and non-Agent fallback. Existing unrelated profile entries are preserved, and rerunning setup updates the same block.
 
@@ -61,13 +61,14 @@ The Plugin registers:
 In the `web` profile, the same package also registers:
 
 - a localized object-first Workbench (Summary, Trials, Pipeline, Optimization, Compare/Gate, Evaluator/Rubric, Artifacts, Audit) that directly exposes fixed experiment identities, Agent-visible Dataset queries/instructions, safe business-artifact previews, Ground Truth meta-evaluation, paginated per-Trial evidence and recommendations, Population validity/coverage, controlled optimization hypotheses, and Baseline/Gate deltas; raw JSON remains in the audit drawer;
-- a Composer Context Capsule that freezes a Job, Trial, Criterion, or Evidence selection only after an explicit `Ask AI` / `Ask about this` action or `@harbor` selection; ordinary sends never inherit the visible Harbor page, and the one-shot reference clears after its turn is sent;
-- a same-session Harbor Copilot Dock that renders the real Chat Session's running state, Tool progress, final answer, failure, and Stop action without copying conversation history; a typed `harbor.navigate` result performs only an allowlisted, read-only Harbor location change when its object, page-Session, and generation preconditions still match, while Back restores the prior workspace, page, stage, Trial filters/sort/focus, Compare Baseline, and scroll position;
-- a first-class `Evaluate recent Sessions` action that previews up to ten safe Session records, shows the frozen Evaluator/Judge and diagnostic boundaries, requires explicit confirmation, runs in the background, and opens the completed Job;
+- the existing native Composer and conversation, without a Context Capsule or Copilot panel above the input; optional one-shot `Ask AI` / `@harbor` references freeze a Job, Trial, Criterion, or Evidence selection and clear after sending. Plain messages do not automatically attach the visible page on the current rc.8 Host;
+- native Tool result cards for evidence navigation and reviewed AI proposals; typed `harbor.navigate` actions retain allowlisted, read-only Harbor navigation and Back restoration of the prior workspace, page, stage, Trial filters/sort/focus, Compare Baseline, and scroll position. Cards ask you to open the Harbor tab after preparing the object; they do not automatically switch Host tabs;
+- background operations in the main plugin page, retaining cancellation, recovery inspection and result navigation. The entry disappears only after a successful empty read, not on a read failure;
+- a first-class `Evaluate recent Sessions` quickstart that automatically samples up to three completed conversations from history available to the current DSH, independent of the evaluation output directory. It previews the review model and redacted-data/cost disclosure, requires confirmation, runs in the background, and opens the completed Job. No history path, project or date picker is required; the bounded recent sample is not a claim about all history;
 - descriptor-authorized Evaluator/Rubric source editing for `script` and `llm-as-judge` implementations, with optimistic concurrency and mandatory new identities;
 - a `harbor-dsh-evaluator/v1` interface shared by deterministic scripts and LLM-as-Judge implementations;
 - compact result cards for all Harbor Tool calls;
-- explicit local-object selection and frozen Trial-set selectors (fixed IDs/revisions or a query snapshot, at most 1000 members), independently removable context chips, and source fragment Ask;
+- explicit local-object selection and frozen Trial-set selectors (fixed IDs/revisions or a query snapshot, at most 1000 members), removable native references, and source fragment Ask;
 - AI proposal cards with deterministic Preflight, explicit review, idempotent confirmation and append-only local operation journals; Candidate/Gate/handoff output is a saved draft, not an applied resource change. Evaluator source proposals can be opened in the reviewed version editor. Selected Compare executes a read only;
 - a `Harbor Evolution` Settings section that checks the configured project, Evaluation Stack, Jobs directory, and CLI paths, supports process-local `projectRoot` reload, and checks npm for a newer formal release without silently installing it.
 
@@ -75,7 +76,7 @@ In the `web` profile, the same package also registers:
 
 1. Open an evaluation result. Select a task, score, evidence item, or saved source fragment.
 2. Choose **Ask AI** or a suggested question. Harbor prepares a visible reference in the existing Composer; review, add your question, and send.
-3. Continue chatting in the same Session. The Copilot shows ordinary follow-ups and recent discussion history. An ordinary answer does not inherit the old answer's evidence-validity badge. Reattach the original object when you need fresh evidence or a modification.
+3. Read answers and proposal cards in the existing conversation. Harbor does not create another answer surface. Ordinary follow-ups use conversation history; reattach an object when you need fresh evidence or a modification. The current rc.8 public input contract has no general page-context contributor at message submission, so automatic page binding is not claimed.
 4. For scoring rules, select saved lines and choose **Suggest a change**. **Review and edit** opens the matching file directly. The AI may populate an unchanged editor, but never replaces your manual edits. Review the diff and explicitly save to create new identities; this does not run an evaluation or Gate.
 
 Unsaved source edits are isolated by Session, workspace, Job, and file and retained in this browser tab's `sessionStorage`. File/view switches and refresh can recover them; closing the tab may discard them. Storage failures are shown, with an in-memory fallback and a leave-page warning for unpersisted edits. Source conflicts preserve the original base and edited text; review the latest source before accepting a new base. Saving or explicitly discarding clears only that file's draft. Expired authorizations never erase suggestion text or human edits; changed source or expired task subsets require an explicit new selection, not an automatically widened scope.
@@ -88,7 +89,7 @@ A direct evaluation requires `candidatePath`, `datasetPath`, `stackPath`, and ex
 
 ## Historical Session cold start
 
-When the user does not provide a Dataset, the simplest entry is the `Evaluate recent Sessions` button in the Harbor tab. It previews up to ten recent completed business Sessions, shows only safe metadata plus the Evaluator, Judge, same-model coupling, estimated requests, expiry and local retention boundaries, and starts nothing until the user confirms. The Host keeps the short-lived selection token in memory; the browser receives only an opaque Preview id. The bundled Skill remains the conversational entry and uses the same Preview/Run services from the Agent's exact working directory.
+When the user does not provide a Dataset, the simplest entry is the `Evaluate recent Sessions` button in the Harbor tab. It automatically finds up to three completed conversations through the current DSH Session Query service, including histories from other project directories. It reads recent candidates in small batches, stops once enough are found, and records the scan boundary instead of claiming a full-history ranking. The current conversation, unfinished and internal evaluation conversations remain excluded. The user sees a short sample preview, the review model and a redacted-data/cost disclosure, then confirms once to start. No storage path or project selection is needed. Results stay in the selected evaluation workspace; each source is revalidated against its own frozen identity. The Host keeps the short-lived selection token in memory; the browser receives only an opaque Preview id. The bundled Skill remains the conversational entry and retains its explicit exact-working-directory selection mode (up to ten) for existing Agent workflows.
 
 After explicit confirmation, `harbor_session_diagnostic_run` receives only the `selectionToken` and an optional Job name. It revalidates the frozen Session and Feedback digests, materializes an immutable Historical Batch plus matching Dataset and Stack, and evaluates one Session Observation per Harbor Trial. The Job does not rerun a Candidate, cannot enter Promotion Gate, and records Evaluator Meta-Evaluation as `not-run` because evaluator reliability requires a separate independent Ground Truth workflow.