thumbgate 1.5.3 โ†’ 1.5.8

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "thumbgate-marketplace",
3
- "version": "1.5.3",
3
+ "version": "1.5.8",
4
4
  "owner": {
5
5
  "name": "Igor Ganapolsky",
6
6
  "email": "ig5973700@gmail.com"
@@ -13,7 +13,7 @@
13
13
  "source": "npm",
14
14
  "package": "thumbgate"
15
15
  },
16
- "version": "1.5.3",
16
+ "version": "1.5.8",
17
17
  "author": {
18
18
  "name": "Igor Ganapolsky"
19
19
  },
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "name": "thumbgate",
3
3
  "description": "Type ๐Ÿ‘ or ๐Ÿ‘Ž on any agent action. ThumbGate captures it, distills a lesson, and blocks the pattern from repeating. One thumbs-down = the agent physically cannot make that mistake again. 33 pre-action gates, budget enforcement, self-protection, and NIST/SOC2 compliance tags.",
4
- "version": "1.5.3",
4
+ "version": "1.5.8",
5
5
  "author": {
6
6
  "name": "Igor Ganapolsky"
7
7
  },
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "thumbgate",
3
- "version": "1.5.3",
3
+ "version": "1.5.8",
4
4
  "description": "ThumbGate โ€” ๐Ÿ‘๐Ÿ‘Ž feedback that teaches your AI agent. Thumbs down a mistake, it never happens again.",
5
5
  "homepage": "https://github.com/IgorGanapolsky/thumbgate",
6
6
  "transport": "stdio",
@@ -3,7 +3,7 @@
3
3
  - `chatgpt/openapi.yaml`: import into GPT Actions.
4
4
  - `gemini/function-declarations.json`: Gemini function-calling definitions.
5
5
  - `mcp/server-stdio.js`: underlying local MCP stdio server implementation.
6
- - `claude/.mcp.json`: example Claude Code MCP config using `npx --yes --package thumbgate@1.5.3 thumbgate serve`.
6
+ - `claude/.mcp.json`: example Claude Code MCP config using `npx --yes --package thumbgate@1.5.8 thumbgate serve`.
7
7
  - `codex/config.toml`: example Codex MCP profile section using the same version-pinned portable launcher.
8
8
  - `amp/skills/thumbgate-feedback/SKILL.md`: Amp skill template.
9
9
  - `opencode/opencode.json`: portable OpenCode MCP profile using the same version-pinned portable launcher.
@@ -2,13 +2,13 @@
2
2
  "mcpServers": {
3
3
  "thumbgate": {
4
4
  "command": "npx",
5
- "args": ["--yes", "--package", "thumbgate@1.5.3", "thumbgate", "serve"]
5
+ "args": ["--yes", "--package", "thumbgate@1.5.8", "thumbgate", "serve"]
6
6
  }
7
7
  },
8
8
  "hooks": {
9
9
  "preToolUse": {
10
10
  "command": "npx",
11
- "args": ["--yes", "--package", "thumbgate@1.5.3", "thumbgate", "gate-check"]
11
+ "args": ["--yes", "--package", "thumbgate@1.5.8", "thumbgate", "gate-check"]
12
12
  }
13
13
  }
14
14
  }
@@ -3,9 +3,9 @@
3
3
  # ~/.codex/config.json with the ThumbGate hooks and status line.
4
4
  [mcp_servers.thumbgate]
5
5
  command = "npx"
6
- args = ["--yes", "--package", "thumbgate@1.5.3", "thumbgate", "serve"]
6
+ args = ["--yes", "--package", "thumbgate@1.5.8", "thumbgate", "serve"]
7
7
 
8
8
  # Hard PreToolUse hook for Codex
9
9
  [hooks.pre_tool_use]
10
10
  command = "npx"
11
- args = ["--yes", "--package", "thumbgate@1.5.3", "thumbgate", "gate-check"]
11
+ args = ["--yes", "--package", "thumbgate@1.5.8", "thumbgate", "gate-check"]
@@ -146,7 +146,7 @@ const {
146
146
  finalizeSession: finalizeFeedbackSession,
147
147
  } = require('../../scripts/feedback-session');
148
148
 
149
- const SERVER_INFO = { name: 'thumbgate-mcp', version: '1.5.3' };
149
+ const SERVER_INFO = { name: 'thumbgate-mcp', version: '1.5.8' };
150
150
  const COMMERCE_CATEGORIES = [
151
151
  'product_recommendation',
152
152
  'brand_compliance',
@@ -7,7 +7,7 @@
7
7
  "npx",
8
8
  "--yes",
9
9
  "--package",
10
- "thumbgate@1.5.3",
10
+ "thumbgate@1.5.8",
11
11
  "thumbgate",
12
12
  "serve"
13
13
  ],
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "thumbgate",
3
- "version": "1.5.3",
3
+ "version": "1.5.8",
4
4
  "description": "Self-improving agent governance: type thumbs-up or thumbs-down on any AI agent action. ThumbGate turns every mistake into a prevention rule and blocks the pattern from repeating. One thumbs-down, never again. 33 pre-action gates, budget enforcement, and self-protection for Claude Code, Cursor, Codex, Gemini CLI, and Amp.",
5
5
  "homepage": "https://thumbgate-production.up.railway.app",
6
6
  "repository": {
@@ -35,11 +35,14 @@
35
35
  "bin/postinstall.js",
36
36
  "config/",
37
37
  "openapi/",
38
+ "public/blog.html",
38
39
  "public/compare.html",
39
40
  "public/dashboard.html",
40
41
  "public/guide.html",
41
42
  "public/index.html",
43
+ "public/learn.html",
42
44
  "public/lessons.html",
45
+ "public/pro.html",
43
46
  "scripts/access-anomaly-detector.js",
44
47
  "scripts/agent-readiness.js",
45
48
  "scripts/agentic-data-pipeline.js",
@@ -245,7 +248,7 @@
245
248
  "trace:eval": "node scripts/decision-trace.js eval",
246
249
  "social:reply-monitor": "node scripts/social-reply-monitor.js",
247
250
  "social:reply-monitor:dry": "node scripts/social-reply-monitor.js --dry-run",
248
- "test": "npm run test:schema && npm run test:loop && npm run test:dpo && npm run test:kto && npm run test:api && npm run test:proof && npm run test:e2e && npm run test:rlaif && npm run test:attribution && npm run test:quality && npm run test:intelligence && npm run test:training-export && npm run test:deployment && npm run test:operational-integrity && npm run test:workflow && npm run test:billing && npm run test:cli && npm run test:watcher && npm run test:autoresearch && npm run test:ops && npm run test:session-analyzer && npm run test:tessl && npm run test:gates && npm run test:evoskill && npm run test:gates-hardening && npm run test:workers && npm run test:social-analytics && npm run test:memalign && npm run test:xmemory-lite && npm run test:filesystem-search && npm run test:zernio && npm run test:post-video && npm run test:post-everywhere-instagram && npm run test:obsidian-export && npm run test:lesson-db && npm run test:lesson-rotation && npm run test:memory-dedup && npm run test:feedback-quality && npm run test:sync-version && npm run test:check-congruence && npm run test:tool-registry && npm run test:feedback-to-rules && npm run test:memory-firewall && npm run test:belief-update && npm run test:hosted-config && npm run test:operational-summary && npm run test:operator-key-auth && npm run test:cloudflare-sandbox && npm run test:mcp-config && npm run test:plan-gate && npm run test:pulse && npm run test:semantic-layer && npm run test:data-pipeline && npm run test:optimize-context && npm run test:principle-extractor && npm run test:analytics-window && npm run test:funnel-analytics && npm run test:experiment-tracker && npm run test:build-metadata && npm run test:context-engine && npm run test:hf-papers && npm run test:marketing-experiment && npm run test:seo-gsd && npm run test:verify-run && npm run test:export-dpo-pairs && npm run test:export-hf-dataset && npm run test:license && npm run test:bot-detector && npm run test:postinstall && npm run test:funnel-invariants && npm run test:cli-telemetry && npm run test:pro-parity && npm run test:model-tier-router && npm run test:computer-use-firewall && npm run test:skill-exporter && npm run test:statusline && npm run test:evolution && npm run test:org-dashboard && npm run test:multi-hop-recall && npm run test:synthetic-dpo && npm run test:thumbgate-skill && npm run test:learn-hub && npm run test:feedback-fallback && npm run test:metaclaw && npm run test:server-lock && npm run test:control-tower && npm run test:pii-scanner && npm run test:data-governance && npm run test:lesson-inference && npm run test:semantic-dedup && npm run test:fs-utils && npm run test:cli-schema && npm run test:explore && npm run test:lesson-reranker && npm run test:lesson-retrieval && npm run test:cross-encoder && npm run test:reflector-agent && npm run test:feedback-session && npm run test:feedback-history-distiller && npm run test:hallucination-detector && npm run test:history-distiller && npm run test:predictive-insights && npm run test:prove-predictive-insights && npm run test:statusbar-cli && npm run test:generate-instagram-card && npm run test:instagram-thumbgate-post && npm run test:publish-instagram-thumbgate && npm run test:lesson-synthesis && npm run test:background-governance && npm run test:memory-migration && npm run test:prompt-dlp && npm run test:ephemeral-store && npm run test:agent-security && npm run test:skill-progressive && npm run test:per-step-scoring && npm run test:weekly-auto-post && npm run test:social-post-hourly && npm run test:social-quality-gate && npm run test:a2ui-engine && npm run test:gate-satisfy && npm run test:money-watcher && npm run test:budget && npm run test:quick-start && npm run test:utm && npm run test:product-feedback && npm run test:feedback-root-consolidator && npm run test:engagement-audit && npm run test:install-growth-automation && npm run test:publish-thumbgate-launch && npm run test:reconcile-thumbgate-campaign && npm run test:reddit-publisher && npm run test:schedule-thumbgate-campaign && npm run test:social-reply-monitor && npm run test:sync-launch-assets && npm run test:ai-search-visibility && npm run test:perplexity && npm run test:security-scanner && npm run test:llm-client && npm run test:managed-lesson-agent && npm run test:self-distill && npm run test:meta-agent && npm run test:harness-selector && npm run test:thumbgate-bench && npm run test:seo-guides && npm run test:enforcement-loop && npm run test:cli-agent-experience && npm run test:bot-detection && npm run test:checkout-bot-guard && npm run test:session-health && npm run test:session-episodes && npm run test:spec-gate && npm run test:decision-trace && npm run test:dashboard-insights && npm run test:prompt-eval && npm run test:demo-voiceover && npm run test:gate-coherence && npm run test:gate-eval && npm run test:high-roi && npm run test:public-static-assets && npm run test:token-savings && npm run test:workflow-gate-checkpoint && npm run test:lesson-export-import && npm run test:landing-page-claims && npm run test:dashboard-deeplink-e2e",
251
+ "test": "npm run test:schema && npm run test:loop && npm run test:dpo && npm run test:kto && npm run test:api && npm run test:proof && npm run test:e2e && npm run test:rlaif && npm run test:attribution && npm run test:quality && npm run test:intelligence && npm run test:training-export && npm run test:deployment && npm run test:operational-integrity && npm run test:workflow && npm run test:billing && npm run test:cli && npm run test:watcher && npm run test:autoresearch && npm run test:ops && npm run test:session-analyzer && npm run test:tessl && npm run test:gates && npm run test:evoskill && npm run test:gates-hardening && npm run test:workers && npm run test:social-analytics && npm run test:memalign && npm run test:xmemory-lite && npm run test:filesystem-search && npm run test:zernio && npm run test:post-video && npm run test:post-everywhere-instagram && npm run test:obsidian-export && npm run test:lesson-db && npm run test:lesson-rotation && npm run test:memory-dedup && npm run test:feedback-quality && npm run test:sync-version && npm run test:check-congruence && npm run test:tool-registry && npm run test:feedback-to-rules && npm run test:memory-firewall && npm run test:belief-update && npm run test:hosted-config && npm run test:operational-summary && npm run test:operator-key-auth && npm run test:cloudflare-sandbox && npm run test:mcp-config && npm run test:plan-gate && npm run test:pulse && npm run test:semantic-layer && npm run test:data-pipeline && npm run test:optimize-context && npm run test:principle-extractor && npm run test:analytics-window && npm run test:funnel-analytics && npm run test:experiment-tracker && npm run test:build-metadata && npm run test:context-engine && npm run test:hf-papers && npm run test:marketing-experiment && npm run test:seo-gsd && npm run test:verify-run && npm run test:export-dpo-pairs && npm run test:export-hf-dataset && npm run test:license && npm run test:bot-detector && npm run test:postinstall && npm run test:funnel-invariants && npm run test:cli-telemetry && npm run test:pro-parity && npm run test:model-tier-router && npm run test:computer-use-firewall && npm run test:skill-exporter && npm run test:statusline && npm run test:evolution && npm run test:org-dashboard && npm run test:multi-hop-recall && npm run test:synthetic-dpo && npm run test:thumbgate-skill && npm run test:learn-hub && npm run test:feedback-fallback && npm run test:metaclaw && npm run test:server-lock && npm run test:control-tower && npm run test:pii-scanner && npm run test:data-governance && npm run test:lesson-inference && npm run test:semantic-dedup && npm run test:fs-utils && npm run test:cli-schema && npm run test:explore && npm run test:lesson-reranker && npm run test:lesson-retrieval && npm run test:cross-encoder && npm run test:reflector-agent && npm run test:feedback-session && npm run test:feedback-history-distiller && npm run test:hallucination-detector && npm run test:history-distiller && npm run test:predictive-insights && npm run test:prove-predictive-insights && npm run test:statusbar-cli && npm run test:generate-instagram-card && npm run test:instagram-thumbgate-post && npm run test:publish-instagram-thumbgate && npm run test:lesson-synthesis && npm run test:background-governance && npm run test:memory-migration && npm run test:prompt-dlp && npm run test:ephemeral-store && npm run test:agent-security && npm run test:skill-progressive && npm run test:per-step-scoring && npm run test:weekly-auto-post && npm run test:social-post-hourly && npm run test:social-quality-gate && npm run test:a2ui-engine && npm run test:gate-satisfy && npm run test:money-watcher && npm run test:budget && npm run test:quick-start && npm run test:utm && npm run test:product-feedback && npm run test:feedback-root-consolidator && npm run test:engagement-audit && npm run test:install-growth-automation && npm run test:publish-thumbgate-launch && npm run test:reconcile-thumbgate-campaign && npm run test:reddit-publisher && npm run test:schedule-thumbgate-campaign && npm run test:social-reply-monitor && npm run test:sync-launch-assets && npm run test:ai-search-visibility && npm run test:perplexity && npm run test:security-scanner && npm run test:llm-client && npm run test:managed-lesson-agent && npm run test:self-distill && npm run test:meta-agent && npm run test:harness-selector && npm run test:thumbgate-bench && npm run test:seo-guides && npm run test:enforcement-loop && npm run test:cli-agent-experience && npm run test:bot-detection && npm run test:checkout-bot-guard && npm run test:session-health && npm run test:session-episodes && npm run test:spec-gate && npm run test:decision-trace && npm run test:dashboard-insights && npm run test:prompt-eval && npm run test:demo-voiceover && npm run test:gate-coherence && npm run test:gate-eval && npm run test:high-roi && npm run test:public-static-assets && npm run test:token-savings && npm run test:workflow-gate-checkpoint && npm run test:lesson-export-import && npm run test:landing-page-claims && npm run test:dashboard-deeplink-e2e && npm run test:public-package-parity && npm run test:token-savings-dashboard && npm run test:cursor-wiring",
249
252
  "test:session-health": "node --test tests/session-health-sensor.test.js",
250
253
  "test:session-episodes": "node --test tests/session-episode-store.test.js",
251
254
  "test:spec-gate": "node --test tests/spec-gate.test.js",
@@ -481,7 +484,12 @@
481
484
  "test:workflow-gate-checkpoint": "node --test tests/workflow-gate-checkpoint.test.js",
482
485
  "test:lesson-export-import": "node --test tests/lesson-export-import.test.js",
483
486
  "test:landing-page-claims": "node --test tests/landing-page-claims.test.js",
484
- "test:dashboard-deeplink-e2e": "node --test tests/dashboard-deeplink-e2e.test.js"
487
+ "test:dashboard-deeplink-e2e": "node --test tests/dashboard-deeplink-e2e.test.js",
488
+ "test:public-package-parity": "node --test tests/public-package-parity.test.js",
489
+ "prepare": "bash bin/install-hooks.sh >/dev/null 2>&1 || true",
490
+ "install:hooks": "bash bin/install-hooks.sh",
491
+ "test:token-savings-dashboard": "node --test tests/token-savings-dashboard.test.js",
492
+ "test:cursor-wiring": "node --test tests/cursor-wiring.test.js"
485
493
  },
486
494
  "keywords": [
487
495
  "mcp",
@@ -0,0 +1,474 @@
1
+ <!doctype html>
2
+ <html lang="en">
3
+ <head>
4
+ <meta charset="UTF-8" />
5
+ <meta name="viewport" content="width=device-width, initial-scale=1.0" />
6
+ <title>ThumbGate Blog โ€” Agent Governance Engineering</title>
7
+ <script defer data-domain="thumbgate-production.up.railway.app" src="https://plausible.io/js/script.js"></script>
8
+ <meta
9
+ name="description"
10
+ content="Technical breakdowns, release notes, and agent governance insights from the ThumbGate team."
11
+ />
12
+ <link
13
+ rel="canonical"
14
+ href="https://thumbgate-production.up.railway.app/blog"
15
+ />
16
+ <meta
17
+ property="og:title"
18
+ content="ThumbGate Blog โ€” Agent Governance Engineering"
19
+ />
20
+ <meta
21
+ property="og:description"
22
+ content="Technical breakdowns, release notes, and agent governance insights from the ThumbGate team."
23
+ />
24
+ <meta property="og:type" content="website" />
25
+ <meta
26
+ property="og:url"
27
+ content="https://thumbgate-production.up.railway.app/blog"
28
+ />
29
+ <script type="application/ld+json">
30
+ {
31
+ "@context": "https://schema.org",
32
+ "@type": "Blog",
33
+ "name": "ThumbGate Blog",
34
+ "url": "https://thumbgate-production.up.railway.app/blog",
35
+ "publisher": { "@type": "Organization", "name": "Max Smith KDP LLC" },
36
+ "blogPost": [
37
+ {
38
+ "@type": "BlogPosting",
39
+ "headline": "Your AI agent is a supply chain attack surface. Here's how to gate it.",
40
+ "datePublished": "2026-04-10",
41
+ "keywords": "AI agent security, supply chain attack, pre-action gates, agent governance, ThumbGate"
42
+ },
43
+ {
44
+ "@type": "BlogPosting",
45
+ "headline": "The Claude Code Leak Proves Why Pre-Action Gates Matter",
46
+ "datePublished": "2026-04-01",
47
+ "keywords": "Claude Code security, Claude Code guardrails, AI agent safety, pre-action gates"
48
+ },
49
+ {
50
+ "@type": "BlogPosting",
51
+ "headline": "v0.8.5: Gate Reasoning Chains, Org Dashboard, and the Checkout Funnel That Didn't Exist",
52
+ "datePublished": "2026-03-31"
53
+ }
54
+ ]
55
+ }
56
+ </script>
57
+ <style>
58
+ :root {
59
+ --bg: #0a0a0a;
60
+ --surface: #141414;
61
+ --border: #2a2a2a;
62
+ --text: #e0e0e0;
63
+ --text-dim: #888;
64
+ --cyan: #00d4aa;
65
+ }
66
+ * {
67
+ margin: 0;
68
+ padding: 0;
69
+ box-sizing: border-box;
70
+ }
71
+ body {
72
+ font-family:
73
+ -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, sans-serif;
74
+ background: var(--bg);
75
+ color: var(--text);
76
+ line-height: 1.7;
77
+ }
78
+ .container {
79
+ max-width: 720px;
80
+ margin: 0 auto;
81
+ padding: 0 24px;
82
+ }
83
+ header {
84
+ padding: 24px 0;
85
+ border-bottom: 1px solid var(--border);
86
+ }
87
+ header a {
88
+ color: var(--cyan);
89
+ text-decoration: none;
90
+ font-weight: 600;
91
+ }
92
+ h1 {
93
+ font-size: 18px;
94
+ font-weight: 700;
95
+ }
96
+ .post {
97
+ padding: 48px 0;
98
+ border-bottom: 1px solid var(--border);
99
+ }
100
+ .post-date {
101
+ font-size: 13px;
102
+ color: var(--text-dim);
103
+ margin-bottom: 8px;
104
+ }
105
+ .post h2 {
106
+ font-size: 24px;
107
+ font-weight: 700;
108
+ margin-bottom: 16px;
109
+ letter-spacing: -0.02em;
110
+ }
111
+ .post h3 {
112
+ font-size: 18px;
113
+ font-weight: 600;
114
+ margin: 24px 0 8px;
115
+ }
116
+ .post p {
117
+ margin-bottom: 16px;
118
+ color: var(--text-dim);
119
+ }
120
+ .post ul {
121
+ margin: 0 0 16px 24px;
122
+ color: var(--text-dim);
123
+ }
124
+ .post li {
125
+ margin-bottom: 6px;
126
+ }
127
+ .post code {
128
+ background: var(--surface);
129
+ padding: 2px 6px;
130
+ border-radius: 4px;
131
+ font-size: 14px;
132
+ }
133
+ .post strong {
134
+ color: var(--text);
135
+ }
136
+ .cta {
137
+ display: inline-block;
138
+ margin-top: 16px;
139
+ padding: 10px 20px;
140
+ background: var(--cyan);
141
+ color: #000;
142
+ border-radius: 6px;
143
+ text-decoration: none;
144
+ font-weight: 600;
145
+ font-size: 14px;
146
+ }
147
+ footer {
148
+ padding: 48px 0 24px;
149
+ text-align: center;
150
+ color: var(--text-dim);
151
+ font-size: 13px;
152
+ }
153
+ footer a {
154
+ color: var(--cyan);
155
+ text-decoration: none;
156
+ }
157
+ </style>
158
+ </head>
159
+ <body>
160
+ <header>
161
+ <div
162
+ class="container"
163
+ style="
164
+ display: flex;
165
+ justify-content: space-between;
166
+ align-items: center;
167
+ "
168
+ >
169
+ <h1><a href="/">ThumbGate</a> / Blog</h1>
170
+ <a href="/">Back to home</a>
171
+ </div>
172
+ </header>
173
+
174
+ <div class="container">
175
+ <article class="post">
176
+ <div class="post-date">April 10, 2026</div>
177
+ <h2>Your AI agent is a supply chain attack surface. Here's how to gate it.</h2>
178
+
179
+ <p>
180
+ Your AI coding agent runs shell commands. It installs packages. It
181
+ modifies files, pushes commits, and calls external APIs &mdash; all
182
+ without requiring you to type a single character. That's the pitch.
183
+ That's also the attack surface.
184
+ </p>
185
+
186
+ <h3>The gap is pre-action enforcement</h3>
187
+ <p>
188
+ Static analysis catches known-bad patterns in code you've already
189
+ written. Dependency scanners audit lock files <em>after</em> packages
190
+ are installed. By the time your scanner flags a problem, the agent
191
+ already ran the command.
192
+ </p>
193
+ <p>
194
+ These tools operate on the <em>output</em> of agent actions. You need
195
+ something that operates on the <em>input</em> &mdash; before execution.
196
+ </p>
197
+
198
+ <h3>Pre-Action Gates via PreToolUse hooks</h3>
199
+ <p>
200
+ ThumbGate implements pre-action gates via <code>PreToolUse</code> hooks
201
+ &mdash; interception points that run before every tool invocation. No
202
+ action reaches execution without passing through the gate. Not Bash
203
+ commands, not file edits, not web fetches.
204
+ </p>
205
+ <p>
206
+ What makes this more than a static blocklist is the
207
+ <strong>feedback-to-enforcement pipeline</strong>. When something goes
208
+ wrong, you record a thumbs-down with context. That failure feeds a
209
+ promotion engine. One failure becomes a warning. Three confirmed
210
+ failures of the same pattern become a hard block.
211
+ </p>
212
+
213
+ <h3>Real examples</h3>
214
+ <ul>
215
+ <li>
216
+ <strong>Force-push to main</strong> &mdash; Gate fires, push never
217
+ happens. Agent is redirected to create a branch and open a PR.
218
+ </li>
219
+ <li>
220
+ <strong>Unknown dependency install</strong> &mdash; Flagged for human
221
+ review. Agent pauses until you approve.
222
+ </li>
223
+ <li>
224
+ <strong>Destructive shell command</strong> &mdash; Blocked by a
225
+ prevention rule learned from a prior incident.
226
+ </li>
227
+ </ul>
228
+
229
+ <h3>Five-minute setup</h3>
230
+ <p>
231
+ <code>npx thumbgate init</code> installs the PreToolUse hook and
232
+ generates a starter gate config. Gates are just JSON &mdash; commit
233
+ them, review them, share them across your team.
234
+ </p>
235
+ <p>
236
+ <strong>Human judgment leads. AI supports. ThumbGate enforces it.</strong>
237
+ </p>
238
+
239
+ <a class="cta" href="/guide">Full setup guide &rarr;</a>
240
+ </article>
241
+
242
+ <article class="post">
243
+ <div class="post-date">April 1, 2026</div>
244
+ <h2>Dual-Signal Feedback: Why "What Failed" Isn't Enough</h2>
245
+
246
+ <p>
247
+ Standard thumbs-down tells you <em>something</em> went wrong. But was
248
+ it a bad decision (wrong tool) or bad execution (right tool, wrong
249
+ parameters)?
250
+ </p>
251
+
252
+ <p>
253
+ Inspired by
254
+ <a
255
+ href="https://huggingface.co/papers/2603.28767"
256
+ style="color: var(--cyan)"
257
+ >Gen-Searcher's dual reward system</a
258
+ >, ThumbGate now supports an optional <code>failureType</code> field
259
+ on <code>capture_feedback</code>:
260
+ </p>
261
+
262
+ <ul>
263
+ <li>
264
+ <strong><code>"decision"</code></strong> โ€” the agent chose the wrong
265
+ action entirely
266
+ </li>
267
+ <li>
268
+ <strong><code>"execution"</code></strong> โ€” right action, bad
269
+ parameters or output
270
+ </li>
271
+ </ul>
272
+
273
+ <p>
274
+ Thompson Sampling creates separate sub-arms (e.g.,
275
+ <code>git:decision</code> and <code>git:execution</code>) so
276
+ reliability scores diverge per dimension. An agent might be great at
277
+ choosing git commands but bad at parameterizing them โ€” now you can see
278
+ that distinction.
279
+ </p>
280
+
281
+ <p>
282
+ Backward compatible. Existing feedback without
283
+ <code>failureType</code> works unchanged.
284
+ </p>
285
+
286
+ <a class="cta" href="https://www.npmjs.com/package/thumbgate"
287
+ >Try it now</a
288
+ >
289
+ </article>
290
+
291
+ <article class="post">
292
+ <div class="post-date">April 1, 2026</div>
293
+ <h2>The Claude Code Leak Proves Why Pre-Action Gates Matter</h2>
294
+
295
+ <p>
296
+ Anthropic accidentally shipped 512,000 lines of Claude Code source
297
+ inside an npm package. A missing <code>.npmignore</code> exposed the
298
+ full agent architecture: tool-call loops, permission models, retry
299
+ logic, 44 unreleased feature flags.
300
+ </p>
301
+
302
+ <p>
303
+ Within 24 hours, a clean rewrite called Claw-code hit 100K GitHub
304
+ stars โ€” the fastest-growing repo in GitHub history.
305
+ </p>
306
+
307
+ <h3>What the leak revealed about agent security</h3>
308
+ <p>
309
+ Claude Code has a sophisticated permission model and tool-calling
310
+ pipeline. What it does <strong>not</strong> have is feedback-driven
311
+ enforcement โ€” the ability to learn from past mistakes and physically
312
+ block the agent from repeating them.
313
+ </p>
314
+
315
+ <p>
316
+ That's exactly what ThumbGate does. Every Claude Code user โ€” and every
317
+ Claw-code user โ€” can add pre-action gates today:
318
+ </p>
319
+
320
+ <ul>
321
+ <li>
322
+ <strong>Thumbs-down a mistake</strong> โ€” it auto-generates a
323
+ prevention rule
324
+ </li>
325
+ <li>
326
+ <strong>Gates enforce</strong> โ€” PreToolUse hooks block the action
327
+ before execution
328
+ </li>
329
+ <li>
330
+ <strong>Reasoning chains explain</strong> โ€” every block tells you
331
+ WHY
332
+ </li>
333
+ <li>
334
+ <strong>Thompson Sampling adapts</strong> โ€” confidence tiers prevent
335
+ false blocks
336
+ </li>
337
+ </ul>
338
+
339
+ <h3>Install in 30 seconds</h3>
340
+ <p>
341
+ <code>npx thumbgate init</code> works with Claude Code,
342
+ Claw-code, Cursor, Codex, Gemini, Amp, and any MCP-compatible agent.
343
+ </p>
344
+
345
+ <p>
346
+ The leak proves agents are powerful but fallible software. Memory
347
+ without enforcement is a suggestion.
348
+ <strong>ThumbGate is a guarantee.</strong>
349
+ </p>
350
+
351
+ <a class="cta" href="https://www.npmjs.com/package/thumbgate"
352
+ >Install ThumbGate</a
353
+ >
354
+ </article>
355
+
356
+ <article class="post">
357
+ <div class="post-date">March 31, 2026</div>
358
+ <h2>
359
+ v0.8.5: Gate Reasoning Chains, Org Dashboard, and the Checkout Funnel
360
+ That Didn't Exist
361
+ </h2>
362
+
363
+ <p>
364
+ ThumbGate v0.8.5 is our biggest release yet. Here's what shipped and
365
+ why.
366
+ </p>
367
+
368
+ <h3>The problem we didn't see</h3>
369
+ <p>
370
+ ~1,700 developers install ThumbGate via npm every month.
371
+ <strong>Zero of them ever saw a checkout button.</strong> They find
372
+ the GitHub README, run <code>npx thumbgate init</code>, use
373
+ it for free, and never visit the landing page. The checkout flow
374
+ nobody reaches is irrelevant. We were optimizing a storefront in a
375
+ building with no door.
376
+ </p>
377
+
378
+ <h3>Gate reasoning chains</h3>
379
+ <p>
380
+ Every gate block and warning now explains <strong>WHY</strong> it
381
+ fired. When ThumbGate blocks a <code>git push --force</code>, the
382
+ response includes:
383
+ </p>
384
+ <ul>
385
+ <li>Which pattern matched and what it matched against</li>
386
+ <li>Gate identity: ID, action, layer, severity</li>
387
+ <li>Source: manual policy rule vs auto-promoted from feedback</li>
388
+ <li>
389
+ How to bypass: <code>satisfy_gate("pr_threads_checked")</code>
390
+ </li>
391
+ <li>Historical fire count: "blocked 23x, warned 15x"</li>
392
+ </ul>
393
+ <p>
394
+ This was inspired by the neuro-symbolic explainability trend in
395
+ production AI systems. Gates are the symbolic rules; Thompson Sampling
396
+ provides the statistical confidence. The reasoning chain bridges both.
397
+ </p>
398
+
399
+ <h3>Multi-agent org dashboard</h3>
400
+ <p>
401
+ "I'm not going to have 10,000 agents running in the environment that I
402
+ don't know what they're doing." โ€” CIO.com, March 2026
403
+ </p>
404
+ <p>
405
+ The new <code>org_dashboard</code> MCP tool aggregates gate decisions
406
+ across all registered agent sessions. CIOs and team leads see: total
407
+ active agents, org-wide adherence rate, top blocked gates, and risk
408
+ agents (those with the lowest adherence). Free tier shows 3 agents;
409
+ Pro shows the full org.
410
+ </p>
411
+
412
+ <h3>Multi-hop agentic retrieval</h3>
413
+ <p>
414
+ Inspired by Chroma's Context-1,
415
+ <code>constructMultiHopPack</code> iteratively retrieves context,
416
+ prunes weak chunks, refines the query with expansion terms, and checks
417
+ coverage โ€” stopping when the coverage threshold (60%) is met or max
418
+ hops are reached. Each hop is logged.
419
+ </p>
420
+
421
+ <h3>Thompson Sampling calibration</h3>
422
+ <p>
423
+ <code>MIN_SAMPLES_THRESHOLD</code> (5) prevents low-sample
424
+ overconfidence. <code>getCalibration()</code> reports per-category
425
+ confidence tiers: none (0 samples), low (1-4), medium (5-19), high
426
+ (20+). Callers know when to trust the statistical arm vs fall back to
427
+ rules.
428
+ </p>
429
+
430
+ <h3>The funnel fix</h3>
431
+ <p>
432
+ Four touchpoints now put the checkout URL where 100% of npm users
433
+ actually are:
434
+ </p>
435
+ <ul>
436
+ <li>
437
+ <strong>Post-install banner</strong> โ€” prints after
438
+ <code>npm install</code> (stderr, CI-safe)
439
+ </li>
440
+ <li>
441
+ <strong>Free-tier rate limits</strong> โ€” power features capped,
442
+ upgrade URL in error
443
+ </li>
444
+ <li>
445
+ <strong>MCP enforceLimit</strong> โ€” agents surface the checkout URL
446
+ when limits hit
447
+ </li>
448
+ <li>
449
+ <strong>CLI upgrade nudge</strong> โ€” after <code>init</code>,
450
+ <code>capture</code>, <code>stats</code>
451
+ </li>
452
+ </ul>
453
+ <p>
454
+ 13 funnel invariant CI tests prevent this blindspot from ever
455
+ regressing.
456
+ </p>
457
+
458
+ <a class="cta" href="https://www.npmjs.com/package/thumbgate"
459
+ >Install v0.8.5 on npm</a
460
+ >
461
+ </article>
462
+ </div>
463
+
464
+ <footer>
465
+ <div class="container">
466
+ <a href="/">Home</a> ยท
467
+ <a href="https://github.com/IgorGanapolsky/ThumbGate">GitHub</a> ยท
468
+ <a href="https://x.com/IgorGanapolsky">X</a> ยท
469
+ <a href="https://www.linkedin.com/in/igorganapolsky">LinkedIn</a>
470
+ <br /><br />ยฉ 2026 Max Smith KDP LLC ยท MIT License
471
+ </div>
472
+ </footer>
473
+ </body>
474
+ </html>