kibi-cursor 2.10.0 → 2.10.1

package/.cursor-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "kibi-cursor",
   "displayName": "Kibi",
-  "version": "2.10.0",
+  "version": "2.10.1",
   "description": "Optional Cursor adapter for Kibi. Requires project-local kibi-cli, kibi-mcp, and kibi-core (SWI-Prolog is bundled on Linux and macOS) before the bundled MCP server can start.",
   "license": "AGPL-3.0-or-later",
   "author": {

package/CHANGELOG.md

@@ -1,5 +1,21 @@
 # kibi-cursor
 
+## 2.10.1
+
+### Patch Changes
+
+- ce41f02: Agents now keep a signoff in the pull request it approves. The guidance used to time the signoff as an end-of-work step and said only that an agent must never set `approved_by` itself, so agents often opened the pull request first and then carried the person's answer in a separate follow-up pull request.
+  - kibi-usage 2.37.0, kibi-decisions 1.5.0 and kibi-commit-review 1.4.0 (and the signoff-views resource) say to ask for the signoff before opening or merging the pull request, and to record the person's answer (their name in `approved_by` or a `*_waiver_approved_by` field, their words in `origin.ref` or `approval_ref`) on the same branch and pull request. Writing down an answer the person gave is not self-approval. A pull request that only carries a signoff, firmness or waiver approval is ruled out.
+  - The stop-hook signoff reminder (Claude Code, Codex, Cursor, ZCode) adds the same rule. So do the `agent-requirement-unapproved` check action, the `dimension_unrepresented` repair hint and the waiver lines of `kibi proof explain`.
+
+- 43d4366: Usage telemetry data is cleaner and more complete. Telemetry is now off when the home directory lies inside the temp directory, as eval, packaging and end-to-end harnesses set it, so each harness run no longer counts as a new install; Kibi's own eval and packed-test sandboxes also set `KIBI_TELEMETRY=0`. Events now name the host agent from `CLAUDECODE`, `CURSOR_AGENT`, `CODEX_SANDBOX`, `GEMINI_CLI`, `OPENCODE`, `AI_AGENT` or the MCP client name, carry a hashed host session id in every host, and say whether they come from a hosted cloud workspace. Failed CLI commands and MCP tool results now carry a fixed error category: the result's error code (for example `upsert_dry_run_invalid`), the failing `kibi doctor` checks, or the `kibi migrate` approval problem. `kibi migrate --apply-safe` now prints the current plan hash when the approved hash is missing or out of date.
+
+  Hooks in Claude Code, Codex, Cursor, ZCode and OpenCode now all send the anonymous `kibi_hook` event (before, only Claude Code did, and only in diagnostic mode for the others), and each hook remembers, as hashes in `telemetry-sessions.json`, which host session last ran in the workspace. CLI and MCP events take the session id from `KIBI_SESSION_ID`, `CLAUDE_CODE_SESSION_ID` or `CODEX_THREAD_ID` when the host exposes one, and otherwise from that workspace marker (at most 30 minutes old), so events from Cursor and OpenCode sessions, which expose no session variable, can be grouped too. Every host's hook also sends the queue when it is due, and right away at the end of a turn.
+
+- Updated dependencies [ce41f02]
+- Updated dependencies [43d4366]
+  - kibi-agent-core@0.11.0
+
 ## 2.10.0
 
 ### Minor Changes

package/agent-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://agent-plugins.org/schemas/1.0.0/plugin.schema.json",
   "name": "kibi",
-  "version": "2.10.0",
+  "version": "2.10.1",
   "description": "Repo-local, per-branch, queryable knowledge base. Portable Agent Plugin packaging Kibi's Agent Skills and MCP server.",
   "author": { "name": "Kibi" },
   "homepage": "https://looted.github.io/kibi/",

package/agent-plugin/skills/kibi-commit-review/SKILL.md

@@ -2,7 +2,7 @@
 id: kibi-commit-review
 name: kibi-commit-review
 description: "Use when kibi check --staged or the Kibi pre-commit hook blocks a commit (kibi_impact_evidence_missing, symbols_manifest_stale, symbol_granularity_violation, symbol-coverage, kibi check-generated drift, Staged impact review failed), or when preparing a commit or pull-request impact review. Covers reading the per-file Detail lines, authoring .kb/symbols.yaml symbol entries through kb_upsert, staging the tracked Kibi evidence with git add (symbol coordinates are derived and never staged), kb_prepare_impact_review for a staged or diff scope, the kibi signoff requirements summary to show before a commit, and how merge commits and migration renames are judged."
-version: 1.3.0
+version: 1.4.0
 kibiCompatibility: ">=1.0.0"
 tags:
   - kibi
@@ -91,7 +91,11 @@ apps); never attach the `.html` as a file, which chat apps show as raw source;
 past 15 requirements the Markdown stops and only the page lists them all.
 Host details: `kibi-decisions`, `resources/signoff-views.md`. `--no-proof` skips the proof lookup, which reads coverage and is slow
 on a large knowledge base. The summary is not an approval: never set
-`approved_by` or `origin.ref` from it.
+`approved_by` or `origin.ref` from it. Show it before you open or merge the
+pull request. When the person answers, record their answer in the same
+change (their name in `approved_by`, their words in `origin.ref` or
+`approval_ref`) and push it to the same pull request; never move a signoff,
+firmness or waiver approval into a separate pull request.
 
 ## Impact reviews
 

package/agent-plugin/skills/kibi-decisions/SKILL.md

@@ -2,7 +2,7 @@
 id: kibi-decisions
 name: kibi-decisions
 description: "Use when the person states a rule or decision (must, never, always, from now on, we decided), when a Kibi hook asks to confirm or warns about an edit to code that implements a firm requirement, when the Stop hook says source files changed but nothing was recorded, when kb_check quotes requirement_statement_unverified, or when a request conflicts with a recorded requirement or ADR. Covers reading firmness and origin, recording a person's rule as firm with origin.ref, superseding with a Why this changed section, the kb_compile_intent unresolved/reviewCandidates/conflictReviews loop, asking before overriding a firm rule, and putting decisions to the person as one numbered kibi signoff --questions view."
-version: 1.4.0
+version: 1.5.0
 kibiCompatibility: ">=1.0.0"
 tags:
   - kibi
@@ -115,7 +115,12 @@ again until none is open, then record each rule with their words in
 `origin.ref`. Fields and outputs: `resources/signoff-views.md`.
 
 Never invent words for `origin.ref`, never set `approved_by` yourself, and
-mark your own picks `default` and guesses `exploratory`. Payload shapes,
+mark your own picks `default` and guesses `exploratory`. Once the person has
+answered, recording their answer (their name in `approved_by` or a
+`*_waiver_approved_by` field, their words in `origin.ref` or `approval_ref`)
+is yours to do: put it on the branch of the change it approves and push it to
+that pull request, never to a separate one. Ask before you open or merge the
+pull request, so the answer lands in it. Payload shapes,
 refusal messages and the `kb_compile_intent` review loop are in
 `resources/decision-records.md`.
 

package/agent-plugin/skills/kibi-decisions/resources/signoff-views.md

@@ -51,7 +51,9 @@ writes and opens the page; `--format json` prints the view.
 The person answers by number ("1 yes, 2 exploratory, 3 reword: ..."). Copy
 each answer into `answer`, render again, and record the rules only once every
 question is answered. An answer is the person's words: quote it in
-`origin.ref`; a view itself is never an approval.
+`origin.ref`; a view itself is never an approval. Record the answers on the
+branch of the change they approve and push them to its pull request, never to
+a separate one.
 
 ## Summary contents
 

package/agent-plugin/skills/kibi-usage/SKILL.md

@@ -2,7 +2,7 @@
 id: kibi-usage
 name: Kibi Usage
 description: Use Kibi's source-first, exact-Git, migration-aware, proof-aware operations safely across MCP or the trusted local CLI, including partial completion repair.
-version: 2.36.0
+version: 2.37.0
 kibiCompatibility: ">=1.0.0"
 tags:
   - kibi
@@ -372,6 +372,14 @@ human sets `approved_by`; `kb_check` flags approvals an agent recorded without
 corroboration. Never fill in an approval yourself. Record why a requirement
 exists in `rationale`, or link the ADR that explains it.
 
+A signoff belongs to the change it approves. Ask for it before you open or
+merge the pull request, and when the person answers, record their answer
+(`approved_by` or a `*_waiver_approved_by` field set to them, their words in
+`origin.ref` or `approval_ref`) on the same branch and push it to the same pull
+request. Writing down an answer the person gave is not approving it yourself.
+Never open a separate pull request only to carry a signoff, firmness or waiver
+approval.
+
 ## Decision firmness
 
 Requirements and ADRs carry `firmness`, which says what you may do with them.

package/dist/.cursor-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "kibi-cursor",
   "displayName": "Kibi",
-  "version": "2.10.0",
+  "version": "2.10.1",
   "description": "Optional Cursor adapter for Kibi. Requires project-local kibi-cli, kibi-mcp, and kibi-core (SWI-Prolog is bundled on Linux and macOS) before the bundled MCP server can start.",
   "license": "AGPL-3.0-or-later",
   "author": {

package/dist/hook-runner.js

@@ -4,7 +4,7 @@ import fs from "node:fs";
 import path from "node:path";
 import { fileURLToPath, pathToFileURL } from "node:url";
 import { captureReminder, claimCaptureReminder, claimSignoffReminder, confirmFirmEdits, firmEditNotice, firmRequirementsTouched, hasDecisionFlag, holdRuleNote, isRequirementUpsert, markDecisionFlag, recordRequest, settleFirmRequirements, settledFirmRequirements, signoffReminder, takeRuleNote, } from "kibi-agent-core/decision-guard";
-import { appendHookUsageRows, editTraces, hookTelemetryEnabled, kbUsageTrace, readPackageVersion, } from "kibi-agent-core/hook-usage-log";
+import { appendHookUsageRows, editTraces, finishHookTelemetry, hookTracesWanted, kbUsageTrace, readPackageVersion, } from "kibi-agent-core/hook-usage-log";
 import { stampKibiWorkspace } from "kibi-agent-core/kb-mcp-tools";
 import { createEntitySummarizer, editFocus, fileKnowledgeSnippet, implementedRequirementIds, } from "kibi-agent-core/snippets";
 import { preEditGuidance, readGuidance, writeGuidance } from "./guidance.js";
@@ -68,7 +68,7 @@ function recordToolTelemetry(input, workspaceRoot, stateDir, startedAt, env) {
     // rationale: appendHookUsageRows applies the same opt-in; this guard only
     // skips reading the index when telemetry is off.
     // Stryker disable next-line ConditionalExpression
-    if (!hookTelemetryEnabled(env))
+    if (!hookTracesWanted(env))
         return;
     const kbUsage = kbUsageTrace(input.toolName, input.toolInput);
     if (!kbUsage && !isKnownEditableTool(input.toolName))
@@ -166,157 +166,89 @@ export async function runHook(rawInput, environment = {}) {
         stateDir,
         sessionId: input.conversationId,
     };
-    switch (input.event) {
-        case "sessionStart":
-            if (!kibiReady) {
-                const advisory = interfaceAdvisory(initialState.mcpState, workspaceTrusted);
-                return {
-                    additional_context: advisory
-                        ? `${BOOTSTRAP_REMINDER}\n${advisory}`
-                        : BOOTSTRAP_REMINDER,
-                };
-            }
-            {
-                const advisory = interfaceAdvisory(initialState.mcpState, workspaceTrusted);
-                return advisory ? { additional_context: advisory } : emptyResult();
-            }
-        case "preToolUse": {
-            // Name the agent's workspace on every Kibi MCP call so the launcher
-            // answers from it, also for agents working in a git worktree. The
-            // per-call cwd wins over workspace_roots, which can be the window's
-            // folders rather than the agent's checkout.
-            const stamped = stampKibiWorkspace(input.toolName, input.toolInput, kibiWorkspaceFor(input.cwd ?? input.workspaceRoots?.[0]));
-            if (stamped) {
-                return { updated_input: stamped };
-            }
-            const explicitPaths = extractExplicitPathFields(input.toolInput);
-            const hasDirectKbEdit = isEditLikeTool(input.toolName) && explicitPaths.some(isDirectKbPath);
-            if (hasDirectKbEdit) {
-                return {
-                    permission: "allow",
-                    agent_message: DIRECT_KB_EDIT_WARNING,
-                };
-            }
-            const primaryPath = resolvePrimaryPath(input);
-            if (!primaryPath || !kibiReady || !isKnownEditableTool(input.toolName)) {
-                return emptyResult();
-            }
-            const relativePath = toRepoRelativePath(primaryPath, cwd);
-            const state = loadHookState(stateDir);
-            const linked = getSourceLinkedSymbols(cwd, primaryPath, stateDir);
-            const focus = editFocus(path.resolve(cwd, primaryPath), input.toolInput);
-            const ruleNote = takeRuleNote(session);
-            const notice = firmNotice(cwd, linked, focus, session, environment.env ?? process.env);
-            const guidance = hasGuidedPath(state, "pre-edit", relativePath)
-                ? undefined
-                : preEditGuidance(primaryPath, {
-                    cwd,
-                    hasKibi: kibiReady,
-                    mcpState: state.mcpState,
-                    workspaceTrusted,
-                    // A file that implements requirements always gets the shared
-                    // snippet, so the bare linked-requirement wording is never needed.
-                    knowledge: knowledgeFor(cwd, linked, "edit", focus),
-                });
-            if (guidance) {
-                rememberGuidedPath(stateDir, "pre-edit", relativePath);
-            }
-            const parts = [ruleNote, notice, guidance].filter((part) => part !== undefined);
-            return parts.length > 0
-                ? { permission: "allow", agent_message: parts.join("\n\n") }
-                : emptyResult();
-        }
-        // implements REQ-agent-hosts-decision-guard
-        case "beforeSubmitPrompt": {
-            // Cursor's prompt hook cannot add context: the request's rule note
-            // waits for the next pre-edit message.
-            if (kibiReady && input.prompt) {
-                const rule = recordRequest(session, input.prompt);
-                if (rule !== undefined)
-                    holdRuleNote(session, rule);
-            }
-            return { continue: true };
-        }
-        case "beforeReadFile": {
-            const primaryPath = resolvePrimaryPath(input);
-            if (!primaryPath || !kibiReady) {
-                return { permission: "allow" };
-            }
-            const relativePath = toRepoRelativePath(primaryPath, cwd);
-            const state = loadHookState(stateDir);
-            if (hasGuidedPath(state, "read", relativePath)) {
-                return { permission: "allow" };
-            }
-            const guidance = readGuidance(primaryPath, {
-                cwd,
-                hasKibi: kibiReady,
-                mcpState: state.mcpState,
-                workspaceTrusted,
-                knowledge: knowledgeFor(cwd, getSourceLinkedSymbols(cwd, primaryPath, stateDir), 
-                // rationale: the snippet only special-cases "edit", so any other
-                // surface literal renders the same read snippet.
-                // Stryker disable next-line StringLiteral
-                "read"),
-            });
-            if (!guidance) {
-                return { permission: "allow" };
-            }
-            rememberGuidedPath(stateDir, "read", relativePath);
-            return {
-                permission: "allow",
-                agent_message: guidance,
-            };
-        }
-        case "postToolUse": {
-            if (kibiReady) {
-                try {
-                    recordToolTelemetry(input, cwd, stateDir, startedAt, environment.env);
+    try {
+        switch (input.event) {
+            case "sessionStart":
+                if (!kibiReady) {
+                    const advisory = interfaceAdvisory(initialState.mcpState, workspaceTrusted);
+                    return {
+                        additional_context: advisory
+                            ? `${BOOTSTRAP_REMINDER}\n${advisory}`
+                            : BOOTSTRAP_REMINDER,
+                    };
                 }
-                catch {
-                    // Telemetry is best effort and must never change the hook's output.
+                {
+                    const advisory = interfaceAdvisory(initialState.mcpState, workspaceTrusted);
+                    return advisory ? { additional_context: advisory } : emptyResult();
                 }
-            }
-            const kbToolCall = extractKbMcpToolCall(input.toolName, input.toolInput);
-            if (kbToolCall) {
-                recordKbMcpTool(stateDir, kbToolCall.toolName, {
-                    impactCheckRun: kbToolCall.impactCheckRun,
-                    sourceFiles: kbToolCall.sourceFiles,
-                });
-            }
-            if (isPlanDeliveryTool(input.toolName)) {
-                recordPlanDelivered(stateDir);
-            }
-            const explicitPaths = extractExplicitPathFields(input.toolInput);
-            if (isKnownEditableTool(input.toolName)) {
-                const dirtyPaths = explicitPaths
-                    .map((candidate) => toRepoRelativePath(candidate, cwd))
-                    .filter((candidate) => isMeaningfulTrackedPath(candidate) ||
-                    isKbFreshnessRelevantPath(candidate));
-                if (dirtyPaths.length > 0) {
-                    addDirtyPaths(stateDir, dirtyPaths);
+            case "preToolUse": {
+                // Name the agent's workspace on every Kibi MCP call so the launcher
+                // answers from it, also for agents working in a git worktree. The
+                // per-call cwd wins over workspace_roots, which can be the window's
+                // folders rather than the agent's checkout.
+                const stamped = stampKibiWorkspace(input.toolName, input.toolInput, kibiWorkspaceFor(input.cwd ?? input.workspaceRoots?.[0]));
+                if (stamped) {
+                    return { updated_input: stamped };
                 }
-                if (dirtyPaths.some(isSourceImpactRelevantPath)) {
-                    markDecisionFlag(session, "source-edited");
+                const explicitPaths = extractExplicitPathFields(input.toolInput);
+                const hasDirectKbEdit = isEditLikeTool(input.toolName) && explicitPaths.some(isDirectKbPath);
+                if (hasDirectKbEdit) {
+                    return {
+                        permission: "allow",
+                        agent_message: DIRECT_KB_EDIT_WARNING,
+                    };
                 }
-            }
-            if (kbToolCall?.toolName === "kb_upsert" ||
-                kbUsageTrace(input.toolName, input.toolInput)?.kbOperation ===
-                    "kb_upsert") {
-                markDecisionFlag(session, "kb-recorded");
-                // implements REQ-kibi-signoff-views
-                if (isRequirementUpsert(input.toolInput)) {
-                    markDecisionFlag(session, "requirement-recorded");
+                const primaryPath = resolvePrimaryPath(input);
+                if (!primaryPath ||
+                    !kibiReady ||
+                    !isKnownEditableTool(input.toolName)) {
+                    return emptyResult();
                 }
+                const relativePath = toRepoRelativePath(primaryPath, cwd);
+                const state = loadHookState(stateDir);
+                const linked = getSourceLinkedSymbols(cwd, primaryPath, stateDir);
+                const focus = editFocus(path.resolve(cwd, primaryPath), input.toolInput);
+                const ruleNote = takeRuleNote(session);
+                const notice = firmNotice(cwd, linked, focus, session, environment.env ?? process.env);
+                const guidance = hasGuidedPath(state, "pre-edit", relativePath)
+                    ? undefined
+                    : preEditGuidance(primaryPath, {
+                        cwd,
+                        hasKibi: kibiReady,
+                        mcpState: state.mcpState,
+                        workspaceTrusted,
+                        // A file that implements requirements always gets the shared
+                        // snippet, so the bare linked-requirement wording is never needed.
+                        knowledge: knowledgeFor(cwd, linked, "edit", focus),
+                    });
+                if (guidance) {
+                    rememberGuidedPath(stateDir, "pre-edit", relativePath);
+                }
+                const parts = [ruleNote, notice, guidance].filter((part) => part !== undefined);
+                return parts.length > 0
+                    ? { permission: "allow", agent_message: parts.join("\n\n") }
+                    : emptyResult();
             }
-            const primaryPath = resolvePrimaryPath(input);
-            if (!primaryPath || !kibiReady) {
-                return emptyResult();
+            // implements REQ-agent-hosts-decision-guard
+            case "beforeSubmitPrompt": {
+                // Cursor's prompt hook cannot add context: the request's rule note
+                // waits for the next pre-edit message.
+                if (kibiReady && input.prompt) {
+                    const rule = recordRequest(session, input.prompt);
+                    if (rule !== undefined)
+                        holdRuleNote(session, rule);
+                }
+                return { continue: true };
             }
-            const relativePath = toRepoRelativePath(primaryPath, cwd);
-            const state = loadHookState(stateDir);
-            if (isReadLikeTool(input.toolName)) {
+            case "beforeReadFile": {
+                const primaryPath = resolvePrimaryPath(input);
+                if (!primaryPath || !kibiReady) {
+                    return { permission: "allow" };
+                }
+                const relativePath = toRepoRelativePath(primaryPath, cwd);
+                const state = loadHookState(stateDir);
                 if (hasGuidedPath(state, "read", relativePath)) {
-                    return emptyResult();
+                    return { permission: "allow" };
                 }
                 const guidance = readGuidance(primaryPath, {
                     cwd,
@@ -324,77 +256,158 @@ export async function runHook(rawInput, environment = {}) {
                     mcpState: state.mcpState,
                     workspaceTrusted,
                     knowledge: knowledgeFor(cwd, getSourceLinkedSymbols(cwd, primaryPath, stateDir), 
-                    // rationale: as above, only "edit" changes the snippet.
+                    // rationale: the snippet only special-cases "edit", so any other
+                    // surface literal renders the same read snippet.
                     // Stryker disable next-line StringLiteral
                     "read"),
                 });
                 if (!guidance) {
-                    return emptyResult();
+                    return { permission: "allow" };
                 }
                 rememberGuidedPath(stateDir, "read", relativePath);
-                return { additional_context: guidance };
+                return {
+                    permission: "allow",
+                    agent_message: guidance,
+                };
             }
-            if (isEditLikeTool(input.toolName)) {
-                // rationale: the runner only ever passes "read" or "write"; an empty
-                // kind selects the same write bucket.
-                // Stryker disable StringLiteral
-                if (hasGuidedPath(state, "write", relativePath)) {
-                    return emptyResult();
+            case "postToolUse": {
+                if (kibiReady) {
+                    try {
+                        recordToolTelemetry(input, cwd, stateDir, startedAt, environment.env);
+                    }
+                    catch {
+                        // Telemetry is best effort and must never change the hook's output.
+                    }
                 }
-                // Stryker restore
-                const guidance = writeGuidance(primaryPath, {
-                    cwd,
-                    hasKibi: kibiReady,
-                    mcpState: state.mcpState,
-                    workspaceTrusted,
-                });
-                if (!guidance) {
+                const kbToolCall = extractKbMcpToolCall(input.toolName, input.toolInput);
+                if (kbToolCall) {
+                    recordKbMcpTool(stateDir, kbToolCall.toolName, {
+                        impactCheckRun: kbToolCall.impactCheckRun,
+                        sourceFiles: kbToolCall.sourceFiles,
+                    });
+                }
+                if (isPlanDeliveryTool(input.toolName)) {
+                    recordPlanDelivered(stateDir);
+                }
+                const explicitPaths = extractExplicitPathFields(input.toolInput);
+                if (isKnownEditableTool(input.toolName)) {
+                    const dirtyPaths = explicitPaths
+                        .map((candidate) => toRepoRelativePath(candidate, cwd))
+                        .filter((candidate) => isMeaningfulTrackedPath(candidate) ||
+                        isKbFreshnessRelevantPath(candidate));
+                    if (dirtyPaths.length > 0) {
+                        addDirtyPaths(stateDir, dirtyPaths);
+                    }
+                    if (dirtyPaths.some(isSourceImpactRelevantPath)) {
+                        markDecisionFlag(session, "source-edited");
+                    }
+                }
+                if (kbToolCall?.toolName === "kb_upsert" ||
+                    kbUsageTrace(input.toolName, input.toolInput)?.kbOperation ===
+                        "kb_upsert") {
+                    markDecisionFlag(session, "kb-recorded");
+                    // implements REQ-kibi-signoff-views
+                    if (isRequirementUpsert(input.toolInput)) {
+                        markDecisionFlag(session, "requirement-recorded");
+                    }
+                }
+                const primaryPath = resolvePrimaryPath(input);
+                if (!primaryPath || !kibiReady) {
                     return emptyResult();
                 }
-                // rationale: same bucket equivalence as the lookup above.
-                // Stryker disable StringLiteral
-                rememberGuidedPath(stateDir, "write", relativePath);
-                // Stryker restore
-                return { additional_context: guidance };
+                const relativePath = toRepoRelativePath(primaryPath, cwd);
+                const state = loadHookState(stateDir);
+                if (isReadLikeTool(input.toolName)) {
+                    if (hasGuidedPath(state, "read", relativePath)) {
+                        return emptyResult();
+                    }
+                    const guidance = readGuidance(primaryPath, {
+                        cwd,
+                        hasKibi: kibiReady,
+                        mcpState: state.mcpState,
+                        workspaceTrusted,
+                        knowledge: knowledgeFor(cwd, getSourceLinkedSymbols(cwd, primaryPath, stateDir), 
+                        // rationale: as above, only "edit" changes the snippet.
+                        // Stryker disable next-line StringLiteral
+                        "read"),
+                    });
+                    if (!guidance) {
+                        return emptyResult();
+                    }
+                    rememberGuidedPath(stateDir, "read", relativePath);
+                    return { additional_context: guidance };
+                }
+                if (isEditLikeTool(input.toolName)) {
+                    // rationale: the runner only ever passes "read" or "write"; an empty
+                    // kind selects the same write bucket.
+                    // Stryker disable StringLiteral
+                    if (hasGuidedPath(state, "write", relativePath)) {
+                        return emptyResult();
+                    }
+                    // Stryker restore
+                    const guidance = writeGuidance(primaryPath, {
+                        cwd,
+                        hasKibi: kibiReady,
+                        mcpState: state.mcpState,
+                        workspaceTrusted,
+                    });
+                    if (!guidance) {
+                        return emptyResult();
+                    }
+                    // rationale: same bucket equivalence as the lookup above.
+                    // Stryker disable StringLiteral
+                    rememberGuidedPath(stateDir, "write", relativePath);
+                    // Stryker restore
+                    return { additional_context: guidance };
+                }
+                return emptyResult();
             }
-            return emptyResult();
-        }
-        case "stop": {
-            const state = loadHookState(stateDir);
-            const shouldClearSession = state.dirtyPaths.length > 0 || state.kbCheckRun || state.planDelivered;
-            if (input.status === "aborted" || input.status === "error") {
+            case "stop": {
+                const state = loadHookState(stateDir);
+                const shouldClearSession = state.dirtyPaths.length > 0 ||
+                    state.kbCheckRun ||
+                    state.planDelivered;
+                if (input.status === "aborted" || input.status === "error") {
+                    if (shouldClearSession) {
+                        clearSessionHookState(stateDir);
+                    }
+                    return emptyResult();
+                }
+                // implements REQ-agent-hosts-decision-guard
+                // Source changed and nothing was recorded: the follow-up asks for the
+                // decision to be recorded, once per conversation.
+                const capture = claimCaptureReminder(session)
+                    ? captureReminder(hasDecisionFlag(session, "rule-prompt"))
+                    : undefined;
+                // implements REQ-kibi-signoff-views
+                // Requirements were written without a signoff view since: ask once.
+                // A recorded requirement also counts as recorded knowledge, so this
+                // never coincides with the capture request above.
+                const signoff = claimSignoffReminder(session, cwd)
+                    ? signoffReminder()
+                    : undefined;
+                const followupMessage = [capture, signoff, stopFollowupMessage(state)]
+                    .filter((part) => part !== undefined)
+                    .join("\n");
+                if (followupMessage) {
+                    clearSessionHookState(stateDir);
+                    return { followup_message: followupMessage };
+                }
                 if (shouldClearSession) {
                     clearSessionHookState(stateDir);
                 }
                 return emptyResult();
             }
-            // implements REQ-agent-hosts-decision-guard
-            // Source changed and nothing was recorded: the follow-up asks for the
-            // decision to be recorded, once per conversation.
-            const capture = claimCaptureReminder(session)
-                ? captureReminder(hasDecisionFlag(session, "rule-prompt"))
-                : undefined;
-            // implements REQ-kibi-signoff-views
-            // Requirements were written without a signoff view since: ask once.
-            // A recorded requirement also counts as recorded knowledge, so this
-            // never coincides with the capture request above.
-            const signoff = claimSignoffReminder(session, cwd)
-                ? signoffReminder()
-                : undefined;
-            const followupMessage = [capture, signoff, stopFollowupMessage(state)]
-                .filter((part) => part !== undefined)
-                .join("\n");
-            if (followupMessage) {
-                clearSessionHookState(stateDir);
-                return { followup_message: followupMessage };
-            }
-            if (shouldClearSession) {
-                clearSessionHookState(stateDir);
-            }
-            return emptyResult();
+            default:
+                return emptyResult();
         }
-        default:
-            return emptyResult();
+    }
+    finally {
+        await finishHookTelemetry({
+            event: input.event,
+            workspaceRoot: kibiReady ? cwd : undefined,
+            sessionId: input.conversationId,
+        }, environment.env ?? process.env);
     }
 }
 export async function main() {

package/dist/skills/.canon-hash.json

@@ -4,11 +4,11 @@
   "kibi-bootstrap/resources/branch-lifecycle.md": "c854264ce888439140883a3fc636db5e522f020dd27b30a5c1ecc4105b64ed13",
   "kibi-bootstrap/resources/operation-access.md": "b5a6a59e72d800c376e7fad79132897623c32a0ce152a1e8d73f9628b444db51",
   "kibi-bootstrap/resources/source-authoring.md": "bf296e62f69474e01797d6cb9469bf624551dca834dd188abb6e705bd96fa586",
-  "kibi-commit-review/SKILL.md": "cd3a0d5ed5e2e7562995180e440dce1aa1607d3d072d3400e966e988725ed399",
+  "kibi-commit-review/SKILL.md": "bec768ad3e4cd4e782e9e13360b69bcd40a0964b305909818115d7f9e8d282d9",
   "kibi-commit-review/resources/staged-gate.md": "37f750f327bb9202c1f30989763f4f1205b906cf672318ca50887d7984bc3e08",
-  "kibi-decisions/SKILL.md": "02685fa50f093025782986bfe009940e934c3a64d47858920815e7eb66b0c523",
+  "kibi-decisions/SKILL.md": "d91f784410251095e9d881bce751728cc4596858a518fb24f5e5d38c940bab50",
   "kibi-decisions/resources/decision-records.md": "b201e619c48daf6ebe5989991a96f6c0573adb6c37abf814b9d34b7a1cd423c6",
-  "kibi-decisions/resources/signoff-views.md": "123da13a3f2b30b5672044a1284626e218d3088411f84f2692746f95e06c1393",
+  "kibi-decisions/resources/signoff-views.md": "84eb2b7123e54047205b84af6fbf60c910ebcaaf98023e5812f3de9b65a4871c",
   "kibi-freshness/SKILL.md": "b5c4f24bebdb5b4e0d04c2fa1176fc33ace35334b0c0b6ca943d6f445dc9b69e",
   "kibi-freshness/resources/branch-lifecycle.md": "ebc01b19fe3dca8df5fc0175978963f6f1204f71f7c805cb2a5fb1b277ac78e4",
   "kibi-freshness/resources/operation-access.md": "4e1db9294ee540cd29b8b3a2532a81e7894d5b4228b39e99ba48b37212f489f3",
@@ -18,7 +18,7 @@
   "kibi-traceability/resources/operation-access.md": "39f7972ab7f4ae7479e78f7072bb50e8e6c4cdf4f857425f69d2522d37db437c",
   "kibi-traceability/resources/source-authoring.md": "408c6ee2bef1b84e0194c0538ef929c26a0a617e631b2234247388140dbbdd08",
   "kibi-traceability/resources/traceability.md": "38c290a0bc71bd1692e115f22694743946e2f69630e134acfd43cee5f613fb88",
-  "kibi-usage/SKILL.md": "d5c2e9e2e643f51f068a4e586002976d28125001c33e791e70994efdbbc8130f",
+  "kibi-usage/SKILL.md": "24c7148c78501a1fa8d9374e96c9a2c5503d04347c186c958cbd768cd67abd88",
   "kibi-usage/resources/branch-lifecycle.md": "f1bf359e9826e83c548f3cd0287f8175a62456527b0132b06b3f9f74aa59b7d6",
   "kibi-usage/resources/fact-lanes.md": "7d487ffb794f6a7c32afabc7756a265a258b072ca83510ddb40d275863d44648",
   "kibi-usage/resources/kb-improvement.md": "a8408fce27342bdce6415a6718599e53fe00bb33aa1fd3dc42353ec6f61adcd7",

package/dist/skills/kibi-commit-review/SKILL.md

@@ -2,7 +2,7 @@
 id: kibi-commit-review
 name: kibi-commit-review
 description: "Use when kibi check --staged or the Kibi pre-commit hook blocks a commit (kibi_impact_evidence_missing, symbols_manifest_stale, symbol_granularity_violation, symbol-coverage, kibi check-generated drift, Staged impact review failed), or when preparing a commit or pull-request impact review. Covers reading the per-file Detail lines, authoring .kb/symbols.yaml symbol entries through kb_upsert, staging the tracked Kibi evidence with git add (symbol coordinates are derived and never staged), kb_prepare_impact_review for a staged or diff scope, the kibi signoff requirements summary to show before a commit, and how merge commits and migration renames are judged."
-version: 1.3.0
+version: 1.4.0
 kibiCompatibility: ">=1.0.0"
 tags:
   - kibi
@@ -91,7 +91,11 @@ apps); never attach the `.html` as a file, which chat apps show as raw source;
 past 15 requirements the Markdown stops and only the page lists them all.
 Host details: `kibi-decisions`, `resources/signoff-views.md`. `--no-proof` skips the proof lookup, which reads coverage and is slow
 on a large knowledge base. The summary is not an approval: never set
-`approved_by` or `origin.ref` from it.
+`approved_by` or `origin.ref` from it. Show it before you open or merge the
+pull request. When the person answers, record their answer in the same
+change (their name in `approved_by`, their words in `origin.ref` or
+`approval_ref`) and push it to the same pull request; never move a signoff,
+firmness or waiver approval into a separate pull request.
 
 ## Impact reviews
 

package/dist/skills/kibi-decisions/SKILL.md

@@ -2,7 +2,7 @@
 id: kibi-decisions
 name: kibi-decisions
 description: "Use when the person states a rule or decision (must, never, always, from now on, we decided), when a Kibi hook asks to confirm or warns about an edit to code that implements a firm requirement, when the Stop hook says source files changed but nothing was recorded, when kb_check quotes requirement_statement_unverified, or when a request conflicts with a recorded requirement or ADR. Covers reading firmness and origin, recording a person's rule as firm with origin.ref, superseding with a Why this changed section, the kb_compile_intent unresolved/reviewCandidates/conflictReviews loop, asking before overriding a firm rule, and putting decisions to the person as one numbered kibi signoff --questions view."
-version: 1.4.0
+version: 1.5.0
 kibiCompatibility: ">=1.0.0"
 tags:
   - kibi
@@ -115,7 +115,12 @@ again until none is open, then record each rule with their words in
 `origin.ref`. Fields and outputs: `resources/signoff-views.md`.
 
 Never invent words for `origin.ref`, never set `approved_by` yourself, and
-mark your own picks `default` and guesses `exploratory`. Payload shapes,
+mark your own picks `default` and guesses `exploratory`. Once the person has
+answered, recording their answer (their name in `approved_by` or a
+`*_waiver_approved_by` field, their words in `origin.ref` or `approval_ref`)
+is yours to do: put it on the branch of the change it approves and push it to
+that pull request, never to a separate one. Ask before you open or merge the
+pull request, so the answer lands in it. Payload shapes,
 refusal messages and the `kb_compile_intent` review loop are in
 `resources/decision-records.md`.
 

package/dist/skills/kibi-decisions/resources/signoff-views.md

@@ -51,7 +51,9 @@ writes and opens the page; `--format json` prints the view.
 The person answers by number ("1 yes, 2 exploratory, 3 reword: ..."). Copy
 each answer into `answer`, render again, and record the rules only once every
 question is answered. An answer is the person's words: quote it in
-`origin.ref`; a view itself is never an approval.
+`origin.ref`; a view itself is never an approval. Record the answers on the
+branch of the change they approve and push them to its pull request, never to
+a separate one.
 
 ## Summary contents
 

package/dist/skills/kibi-usage/SKILL.md

@@ -2,7 +2,7 @@
 id: kibi-usage
 name: Kibi Usage
 description: Use Kibi's source-first, exact-Git, migration-aware, proof-aware operations safely across MCP or the trusted local CLI, including partial completion repair.
-version: 2.36.0
+version: 2.37.0
 kibiCompatibility: ">=1.0.0"
 tags:
   - kibi
@@ -372,6 +372,14 @@ human sets `approved_by`; `kb_check` flags approvals an agent recorded without
 corroboration. Never fill in an approval yourself. Record why a requirement
 exists in `rationale`, or link the ADR that explains it.
 
+A signoff belongs to the change it approves. Ask for it before you open or
+merge the pull request, and when the person answers, record their answer
+(`approved_by` or a `*_waiver_approved_by` field set to them, their words in
+`origin.ref` or `approval_ref`) on the same branch and push it to the same pull
+request. Writing down an answer the person gave is not approving it yourself.
+Never open a separate pull request only to carry a signoff, firmness or waiver
+approval.
+
 ## Decision firmness
 
 Requirements and ADRs carry `firmness`, which says what you may do with them.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "kibi-cursor",
-  "version": "2.10.0",
+  "version": "2.10.1",
   "description": "Kibi Cursor plugin - thin adapter for Cursor MCP integration and editor hooks",
   "type": "module",
   "main": "dist/index.js",
@@ -59,7 +59,7 @@
     }
   },
   "dependencies": {
-    "kibi-agent-core": "^0.10.0"
+    "kibi-agent-core": "^0.11.0"
   },
   "devDependencies": {
     "@types/node": "latest",

package/skills/.canon-hash.json

@@ -4,11 +4,11 @@
   "kibi-bootstrap/resources/branch-lifecycle.md": "c854264ce888439140883a3fc636db5e522f020dd27b30a5c1ecc4105b64ed13",
   "kibi-bootstrap/resources/operation-access.md": "b5a6a59e72d800c376e7fad79132897623c32a0ce152a1e8d73f9628b444db51",
   "kibi-bootstrap/resources/source-authoring.md": "bf296e62f69474e01797d6cb9469bf624551dca834dd188abb6e705bd96fa586",
-  "kibi-commit-review/SKILL.md": "cd3a0d5ed5e2e7562995180e440dce1aa1607d3d072d3400e966e988725ed399",
+  "kibi-commit-review/SKILL.md": "bec768ad3e4cd4e782e9e13360b69bcd40a0964b305909818115d7f9e8d282d9",
   "kibi-commit-review/resources/staged-gate.md": "37f750f327bb9202c1f30989763f4f1205b906cf672318ca50887d7984bc3e08",
-  "kibi-decisions/SKILL.md": "02685fa50f093025782986bfe009940e934c3a64d47858920815e7eb66b0c523",
+  "kibi-decisions/SKILL.md": "d91f784410251095e9d881bce751728cc4596858a518fb24f5e5d38c940bab50",
   "kibi-decisions/resources/decision-records.md": "b201e619c48daf6ebe5989991a96f6c0573adb6c37abf814b9d34b7a1cd423c6",
-  "kibi-decisions/resources/signoff-views.md": "123da13a3f2b30b5672044a1284626e218d3088411f84f2692746f95e06c1393",
+  "kibi-decisions/resources/signoff-views.md": "84eb2b7123e54047205b84af6fbf60c910ebcaaf98023e5812f3de9b65a4871c",
   "kibi-freshness/SKILL.md": "b5c4f24bebdb5b4e0d04c2fa1176fc33ace35334b0c0b6ca943d6f445dc9b69e",
   "kibi-freshness/resources/branch-lifecycle.md": "ebc01b19fe3dca8df5fc0175978963f6f1204f71f7c805cb2a5fb1b277ac78e4",
   "kibi-freshness/resources/operation-access.md": "4e1db9294ee540cd29b8b3a2532a81e7894d5b4228b39e99ba48b37212f489f3",
@@ -18,7 +18,7 @@
   "kibi-traceability/resources/operation-access.md": "39f7972ab7f4ae7479e78f7072bb50e8e6c4cdf4f857425f69d2522d37db437c",
   "kibi-traceability/resources/source-authoring.md": "408c6ee2bef1b84e0194c0538ef929c26a0a617e631b2234247388140dbbdd08",
   "kibi-traceability/resources/traceability.md": "38c290a0bc71bd1692e115f22694743946e2f69630e134acfd43cee5f613fb88",
-  "kibi-usage/SKILL.md": "d5c2e9e2e643f51f068a4e586002976d28125001c33e791e70994efdbbc8130f",
+  "kibi-usage/SKILL.md": "24c7148c78501a1fa8d9374e96c9a2c5503d04347c186c958cbd768cd67abd88",
   "kibi-usage/resources/branch-lifecycle.md": "f1bf359e9826e83c548f3cd0287f8175a62456527b0132b06b3f9f74aa59b7d6",
   "kibi-usage/resources/fact-lanes.md": "7d487ffb794f6a7c32afabc7756a265a258b072ca83510ddb40d275863d44648",
   "kibi-usage/resources/kb-improvement.md": "a8408fce27342bdce6415a6718599e53fe00bb33aa1fd3dc42353ec6f61adcd7",

package/skills/kibi-commit-review/SKILL.md

@@ -2,7 +2,7 @@
 id: kibi-commit-review
 name: kibi-commit-review
 description: "Use when kibi check --staged or the Kibi pre-commit hook blocks a commit (kibi_impact_evidence_missing, symbols_manifest_stale, symbol_granularity_violation, symbol-coverage, kibi check-generated drift, Staged impact review failed), or when preparing a commit or pull-request impact review. Covers reading the per-file Detail lines, authoring .kb/symbols.yaml symbol entries through kb_upsert, staging the tracked Kibi evidence with git add (symbol coordinates are derived and never staged), kb_prepare_impact_review for a staged or diff scope, the kibi signoff requirements summary to show before a commit, and how merge commits and migration renames are judged."
-version: 1.3.0
+version: 1.4.0
 kibiCompatibility: ">=1.0.0"
 tags:
   - kibi
@@ -91,7 +91,11 @@ apps); never attach the `.html` as a file, which chat apps show as raw source;
 past 15 requirements the Markdown stops and only the page lists them all.
 Host details: `kibi-decisions`, `resources/signoff-views.md`. `--no-proof` skips the proof lookup, which reads coverage and is slow
 on a large knowledge base. The summary is not an approval: never set
-`approved_by` or `origin.ref` from it.
+`approved_by` or `origin.ref` from it. Show it before you open or merge the
+pull request. When the person answers, record their answer in the same
+change (their name in `approved_by`, their words in `origin.ref` or
+`approval_ref`) and push it to the same pull request; never move a signoff,
+firmness or waiver approval into a separate pull request.
 
 ## Impact reviews
 

package/skills/kibi-decisions/SKILL.md

@@ -2,7 +2,7 @@
 id: kibi-decisions
 name: kibi-decisions
 description: "Use when the person states a rule or decision (must, never, always, from now on, we decided), when a Kibi hook asks to confirm or warns about an edit to code that implements a firm requirement, when the Stop hook says source files changed but nothing was recorded, when kb_check quotes requirement_statement_unverified, or when a request conflicts with a recorded requirement or ADR. Covers reading firmness and origin, recording a person's rule as firm with origin.ref, superseding with a Why this changed section, the kb_compile_intent unresolved/reviewCandidates/conflictReviews loop, asking before overriding a firm rule, and putting decisions to the person as one numbered kibi signoff --questions view."
-version: 1.4.0
+version: 1.5.0
 kibiCompatibility: ">=1.0.0"
 tags:
   - kibi
@@ -115,7 +115,12 @@ again until none is open, then record each rule with their words in
 `origin.ref`. Fields and outputs: `resources/signoff-views.md`.
 
 Never invent words for `origin.ref`, never set `approved_by` yourself, and
-mark your own picks `default` and guesses `exploratory`. Payload shapes,
+mark your own picks `default` and guesses `exploratory`. Once the person has
+answered, recording their answer (their name in `approved_by` or a
+`*_waiver_approved_by` field, their words in `origin.ref` or `approval_ref`)
+is yours to do: put it on the branch of the change it approves and push it to
+that pull request, never to a separate one. Ask before you open or merge the
+pull request, so the answer lands in it. Payload shapes,
 refusal messages and the `kb_compile_intent` review loop are in
 `resources/decision-records.md`.
 

package/skills/kibi-decisions/resources/signoff-views.md

@@ -51,7 +51,9 @@ writes and opens the page; `--format json` prints the view.
 The person answers by number ("1 yes, 2 exploratory, 3 reword: ..."). Copy
 each answer into `answer`, render again, and record the rules only once every
 question is answered. An answer is the person's words: quote it in
-`origin.ref`; a view itself is never an approval.
+`origin.ref`; a view itself is never an approval. Record the answers on the
+branch of the change they approve and push them to its pull request, never to
+a separate one.
 
 ## Summary contents
 

package/skills/kibi-usage/SKILL.md

@@ -2,7 +2,7 @@
 id: kibi-usage
 name: Kibi Usage
 description: Use Kibi's source-first, exact-Git, migration-aware, proof-aware operations safely across MCP or the trusted local CLI, including partial completion repair.
-version: 2.36.0
+version: 2.37.0
 kibiCompatibility: ">=1.0.0"
 tags:
   - kibi
@@ -372,6 +372,14 @@ human sets `approved_by`; `kb_check` flags approvals an agent recorded without
 corroboration. Never fill in an approval yourself. Record why a requirement
 exists in `rationale`, or link the ADR that explains it.
 
+A signoff belongs to the change it approves. Ask for it before you open or
+merge the pull request, and when the person answers, record their answer
+(`approved_by` or a `*_waiver_approved_by` field set to them, their words in
+`origin.ref` or `approval_ref`) on the same branch and push it to the same pull
+request. Writing down an answer the person gave is not approving it yourself.
+Never open a separate pull request only to carry a signoff, firmness or waiver
+approval.
+
 ## Decision firmness
 
 Requirements and ADRs carry `firmness`, which says what you may do with them.