@softspark/ai-toolkit 4.15.0 → 4.16.0

package/app/hooks/save-session.sh

@@ -8,15 +8,17 @@
 source "$(dirname "$0")/_profile-check.sh"
 # shellcheck source=_session-paths.sh
 source "$(dirname "$0")/_session-paths.sh"
+# shellcheck source=_hook-io.sh
+source "$(dirname "$0")/_hook-io.sh"
 
 # Read from stdin (Claude Code passes JSON with .session_id, .last_assistant_message)
 INPUT=$(cat)
-SESSION_ID=$(echo "$INPUT" | jq -r '.session_id // empty' 2>/dev/null)
+SESSION_ID=$(hook_session_id)
 LAST_MSG=$(echo "$INPUT" | jq -r '.last_assistant_message // "No summary available"' 2>/dev/null | head -5)
 
 SESSION_FILE="$SESSION_CONTEXT_FILE"
 
-if [ -n "$SESSION_ID" ]; then
+if [ "$SESSION_ID" != "default" ]; then
     mkdir -p "$SESSION_DIR"
 
     # Gather git state for richer context

package/app/hooks/session-end.sh

@@ -3,12 +3,45 @@
 #
 # Fires on: SessionEnd
 # Matcher: all
-# Skipped when TOOLKIT_HOOK_PROFILE=minimal.
+# The handoff snapshot is skipped for the minimal profile. Owned recovery
+# cleanup still runs for every profile.
 
-# shellcheck source=_profile-check.sh
-source "$(dirname "$0")/_profile-check.sh"
 # shellcheck source=_session-paths.sh
 source "$(dirname "$0")/_session-paths.sh"
+# shellcheck source=_hook-io.sh
+source "$(dirname "$0")/_hook-io.sh"
+
+# Cleanup must run before _profile-check.sh can exit for the minimal profile.
+INPUT=""
+if [ ! -t 0 ]; then
+    # shellcheck disable=SC2034  # INPUT is consumed via sourced _hook-io.sh
+    INPUT=$(cat)
+fi
+SESSION_ID=$(hook_session_id)
+RECOVERY_ROOT="$SESSION_DIR/output-filter"
+OUTPUT_FILTER_CLI="${AI_TOOLKIT_OUTPUT_FILTER_CLI:-$HOME/.softspark/ai-toolkit/scripts/output_filter_cli.py}"
+if [ "$SESSION_ID" != "default" ] &&
+   [ -d "$RECOVERY_ROOT" ] &&
+   [ ! -L "$RECOVERY_ROOT" ] &&
+   [ -f "$OUTPUT_FILTER_CLI" ] &&
+   [ ! -L "$OUTPUT_FILTER_CLI" ] &&
+   command -v python3 >/dev/null 2>&1; then
+    python3 -S "$OUTPUT_FILTER_CLI" clean \
+        --base-directory "$SESSION_DIR" \
+        --session-id "$SESSION_ID" >/dev/null 2>&1 || true
+fi
+
+SESSION_STATE_CLI="${AI_TOOLKIT_SESSION_STATE_CLI:-$HOME/.softspark/ai-toolkit/scripts/session_state.py}"
+if [ "$SESSION_ID" != "default" ] &&
+   [ -f "$SESSION_STATE_CLI" ] &&
+   [ ! -L "$SESSION_STATE_CLI" ] &&
+   command -v python3 >/dev/null 2>&1; then
+    python3 -S "$SESSION_STATE_CLI" clean \
+        --session-id "$SESSION_ID" >/dev/null 2>&1 || true
+fi
+
+# shellcheck source=_profile-check.sh
+source "$(dirname "$0")/_profile-check.sh"
 
 SESSION_FILE="$SESSION_CONTEXT_FILE"
 HANDOFF_FILE="$SESSION_END_FILE"
@@ -35,4 +68,3 @@ else
 fi
 
 exit 0
-

package/app/hooks/session-start.sh

@@ -8,6 +8,8 @@
 source "$(dirname "$0")/_locate-toolkit.sh"
 # shellcheck source=_session-paths.sh
 source "$(dirname "$0")/_session-paths.sh"
+# shellcheck source=_hook-io.sh
+source "$(dirname "$0")/_hook-io.sh"
 
 emit_context() {
     [ "${AI_TOOLKIT_HOOK_QUIET:-0}" = "1" ] && return 0
@@ -20,14 +22,18 @@ emit_context "MANDATORY: Before answering ANY technical question, apply ALL rule
 emit_context "REMINDER: When writing features or fixing bugs, ensure tests cover the changes. When modifying API, config, or setup, update relevant documentation. Propose these steps to the user — do not silently skip them."
 
 # 1a. Reset per-session edit state (used by revert-guard, test-cohesion, quality-gate)
-SESSION_ID_INPUT=""
+INPUT=""
 if [ ! -t 0 ]; then
-    STDIN_PAYLOAD="$(cat)"
-    SESSION_ID_INPUT="$(printf '%s' "$STDIN_PAYLOAD" | jq -r '.session_id // empty' 2>/dev/null)"
+    # shellcheck disable=SC2034  # INPUT is consumed via sourced _hook-io.sh
+    INPUT="$(cat)"
 fi
-if [ -n "$TOOLKIT_DIR" ] && command -v python3 >/dev/null 2>&1; then
+SESSION_ID_INPUT=$(hook_session_id)
+SESSION_SOURCE=$(hook_json '.source // empty')
+if [ "$SESSION_SOURCE" != "compact" ] &&
+   [ -n "$TOOLKIT_DIR" ] &&
+   command -v python3 >/dev/null 2>&1; then
     python3 "$TOOLKIT_DIR/scripts/session_state.py" reset \
-        ${SESSION_ID_INPUT:+--session-id "$SESSION_ID_INPUT"} >/dev/null 2>&1 || true
+        --session-id "$SESSION_ID_INPUT" >/dev/null 2>&1 || true
 fi
 
 # 1b. GC stale per-session search-required flags (older than 60 min)

package/app/hooks.json

@@ -158,6 +158,16 @@
                         "command": "\"$HOME/.softspark/ai-toolkit/hooks/search-tracker.sh\""
                     }
                 ]
+            },
+            {
+                "_source": "ai-toolkit",
+                "matcher": "Bash",
+                "hooks": [
+                    {
+                        "type": "command",
+                        "command": "\"$HOME/.softspark/ai-toolkit/hooks/filter-tool-output.sh\""
+                    }
+                ]
             }
         ],
         "Stop": [

package/app/output-filter-policy.json

@@ -0,0 +1,15 @@
+{
+  "mode": "off",
+  "profiles": [
+    "repeat-lines",
+    "tap-success"
+  ],
+  "maxInputBytes": 8388608,
+  "minSavingsBytes": 1024,
+  "minSavingsRatio": 0.15,
+  "recovery": {
+    "mode": "ephemeral",
+    "ttlMinutes": 60,
+    "maxSessionBytes": 33554432
+  }
+}

package/app/skills/brand-voice/scripts/measure.py

@@ -8,8 +8,9 @@ and aggregate ratios. Asserts:
     concise <= CONCISE_BUDGET (default 0.60)
     strict  <= STRICT_BUDGET  (default 0.40)
 
-Also asserts fact preservation: every file path, identifier, line:number, and
-fenced code block from default.md must appear in concise.md and strict.md.
+Also asserts fact preservation: every extracted file path, identifier, line
+reference, and number with a unit from default.md must appear in concise.md
+and strict.md. Fenced code bodies are intentionally excluded.
 
 Usage:
     python3 app/skills/brand-voice/scripts/measure.py \\
@@ -36,6 +37,7 @@ LINE_REF_SUFFIX_RE = re.compile(r":\d+$")
 CODE_BLOCK_RE = re.compile(r"```[\w]*\n.*?```", re.DOTALL)
 NUMBER_WITH_UNIT_RE = re.compile(r"\b\d+(?:\.\d+)?(?:ms|s|kb|mb|gb|%)\b", re.IGNORECASE)
 IDENT_BACKTICK_RE = re.compile(r"`([^`\n]+)`")
+IDENTIFIER_RE = re.compile(r"^[A-Za-z_$][A-Za-z0-9_$.-]*(?:\(\))?$")
 
 
 def count_tokens(text: str) -> int:
@@ -65,7 +67,7 @@ def extract_facts(text: str) -> set[str]:
 
     for match in IDENT_BACKTICK_RE.findall(text_no_code):
         cleaned = match.strip()
-        if len(cleaned) >= 3 and not cleaned.startswith("```"):
+        if len(cleaned) >= 3 and IDENTIFIER_RE.fullmatch(cleaned):
             facts.add(normalize_path(cleaned))
 
     for match in NUMBER_WITH_UNIT_RE.findall(text_no_code):
@@ -138,8 +140,8 @@ def evaluate_fixture(
 
     budget_ok_concise = concise_ratio <= concise_budget
     budget_ok_strict = strict_ratio <= strict_budget
-    facts_ok_concise = not concise_required_missing
-    facts_ok_strict = not strict_required_missing
+    facts_ok_concise = not concise_missing and not concise_required_missing
+    facts_ok_strict = not strict_missing and not strict_required_missing
 
     return {
         "fixture": fixture_dir.name,

package/benchmarks/ecosystem-doctor-snapshot.json

@@ -1,5 +1,5 @@
 {
-  "last_run": "2026-07-15T05:28:09Z",
+  "last_run": "2026-07-24T03:52:58Z",
   "schema_version": 1,
   "tools": {
     "aider": {
@@ -24,7 +24,7 @@
       }
     },
     "augment": {
-      "docs_hash": "480f89501d47f5ec",
+      "docs_hash": "388c0b5e25ab4906",
       "headings": [
         "Admin",
         "Auggie CLI",
@@ -66,7 +66,7 @@
       }
     },
     "claude-app": {
-      "docs_hash": "01981c240e67db02",
+      "docs_hash": "3c7ff437aec0ab98",
       "headings": [
         "Add global and folder instructions",
         "Availability",
@@ -107,7 +107,7 @@
       }
     },
     "claude-code": {
-      "docs_hash": "517ae09f5f1feba9",
+      "docs_hash": "a6ae93657498688c",
       "headings": [
         "Core concepts",
         "Documentation Index",
@@ -135,15 +135,15 @@
         "PermissionRequest": false,
         "PostCompact": false,
         "PostToolBatch": false,
-        "PostToolUse": true,
+        "PostToolUse": false,
         "PostToolUseFailure": false,
         "PreCompact": false,
-        "PreToolUse": true,
-        "SKILL.md": true,
+        "PreToolUse": false,
+        "SKILL.md": false,
         "SessionEnd": false,
         "SessionStart": false,
         "Setup": true,
-        "Stop": true,
+        "Stop": false,
         "StopFailure": false,
         "SubagentStart": false,
         "SubagentStop": false,
@@ -159,14 +159,14 @@
         "hook handler: http": false,
         "hook handler: mcp_tool": false,
         "hook handler: prompt": false,
-        "output style": true,
-        "slash command": true,
+        "output style": false,
+        "slash command": false,
         "sub-agent": true
       },
-      "version": "2.1.210 (Claude Code)"
+      "version": "2.1.218 (Claude Code)"
     },
     "cline": {
-      "docs_hash": "9a024e9356105f48",
+      "docs_hash": "3eaa20e45d385a1f",
       "headings": [
         "API Reference",
         "Best Practices",
@@ -213,7 +213,7 @@
       }
     },
     "codex-cli": {
-      "docs_hash": "cc18a01c239497e8",
+      "docs_hash": "87a5a879a39dc62c",
       "headings": [
         "API",
         "API Reference",
@@ -291,7 +291,7 @@
         "Reference",
         "Releases",
         "Resources",
-        "Return to a saved session",
+        "Return to a saved chat",
         "Review changes before they ship",
         "Run Codex and sign in",
         "SDKs and CLI",
@@ -346,10 +346,10 @@
         "mcp_servers": false,
         "sandbox": true
       },
-      "version": "codex-cli 0.144.4"
+      "version": "codex-cli 0.145.0"
     },
     "cursor": {
-      "docs_hash": "34efb96589c7f123",
+      "docs_hash": "5387d04443d26da1",
       "headings": [],
       "markers": {
         ".cursor/rules": false,
@@ -365,7 +365,7 @@
       }
     },
     "gemini-cli": {
-      "docs_hash": "c24d346e0c5a1bb9",
+      "docs_hash": "8db25b552ffbed80",
       "headings": [
         "Breadcrumbs",
         "Directory actions",
@@ -408,7 +408,7 @@
       }
     },
     "github-copilot": {
-      "docs_hash": "58d9980c29b7edb8",
+      "docs_hash": "5d021df2f3b4d39b",
       "headings": [
         "About Copilot auto model selection",
         "About Copilot automations",
@@ -445,7 +445,7 @@
       }
     },
     "google-antigravity": {
-      "docs_hash": "9c9b420a22b35ae6",
+      "docs_hash": "03145f425dd475d1",
       "headings": [],
       "markers": {
         "AGENTS.md": false,
@@ -461,7 +461,7 @@
       }
     },
     "opencode": {
-      "docs_hash": "505b9771a6c768e6",
+      "docs_hash": "f1dd69ffcfb280fa",
       "headings": [
         "Add features",
         "Ask questions",
@@ -521,7 +521,7 @@
       }
     },
     "windsurf": {
-      "docs_hash": "2d4eb599c906bf2c",
+      "docs_hash": "bee6ba0ee1a517dd",
       "headings": [
         "Accounts",
         "Advanced",
@@ -558,7 +558,7 @@
         "AGENTS.md": true,
         "Cascade": true,
         "MCP": true,
-        "SKILL.md": true,
+        "SKILL.md": false,
         "always_on": false,
         "glob": true,
         "hooks": true,

package/benchmarks/output-filter/README.md

@@ -0,0 +1,11 @@
+# Native output-filter benchmark corpus
+
+The corpus is deterministic, synthetic, offline, and authored for ai-toolkit.
+It measures pure profile transformation separately from hook process startup.
+
+The gates are 20 ms p95 for inputs up to 100 KiB, 150 ms p95 for the 8 MiB
+hard-cap case, at least 30% reduction, and peak traced allocation no greater
+than three input sizes plus 16 MiB. The cold-process gate invokes the production
+Bash wrapper with a fresh Python process for every sample in one native session;
+its p95 limit is 75 ms. The default 100 samples keep the p95 gate stable enough
+for release validation.

package/benchmarks/output-filter/scenarios.json

@@ -0,0 +1,25 @@
+[
+  {
+    "name": "repeat-lines-100k",
+    "profile": "repeat-lines",
+    "kind": "repeat",
+    "targetBytes": 102400,
+    "lineWidth": 96,
+    "maxP95Ms": 20
+  },
+  {
+    "name": "tap-success-2k",
+    "profile": "tap-success",
+    "kind": "tap",
+    "testCount": 2000,
+    "maxP95Ms": 20
+  },
+  {
+    "name": "repeat-lines-8m",
+    "profile": "repeat-lines",
+    "kind": "repeat",
+    "targetBytes": 8388608,
+    "lineWidth": 1024,
+    "maxP95Ms": 150
+  }
+]

package/bin/ai-toolkit.js

@@ -50,6 +50,7 @@ const SCRIPT_COMMANDS = {
   'benchmark-ecosystem':  { script: 'benchmark_ecosystem.py', toolkitCwd: true },
   'evaluate':             { script: 'evaluate_skills.py',    toolkitCwd: true },
   'stats':                { script: 'stats.py' },
+  'output-filter':        { script: 'output_filter_cli.py' },
   'compile-slm':          { script: 'compile_slm.py' },
   'pack-codebase':        { script: 'pack_codebase.py' },
   'claude-app':           { script: 'claude_app.py',          toolkitCwd: true },
@@ -79,6 +80,7 @@ const COMMANDS = {
   'benchmark-ecosystem': 'Generate ecosystem benchmark snapshot (GitHub metadata + offline fallback)',
   evaluate: 'Run skill evaluation suite',
   stats: 'Show skill usage statistics (--summary for product telemetry, --reset to clear)',
+  'output-filter': 'Manage native tool-output filter (status, inspect, recover, clean)',
   create: 'Scaffold new skill from template (e.g. create skill my-lint --template=linter)',
   mcp: 'Manage MCP templates and install native editor MCP configs',
   config: 'Manage config inheritance (validate, diff, init, create-base, check)',