opencode-skills-collection 4.1.14 → 4.1.16

package/bundled-skills/llm-evaluation/SKILL.md

@@ -20,7 +20,7 @@ Master comprehensive evaluation strategies for LLM applications, from automated
 - Clarify goals, constraints, and required inputs.
 - Apply relevant best practices and validate outcomes.
 - Provide actionable steps and verification.
-- If detailed examples are required, open `resources/implementation-playbook.md`.
+- Use the examples below; no separate implementation playbook is bundled.
 
 ## Use this skill when
 
@@ -193,7 +193,7 @@ def calculate_factuality(claim, knowledge_base):
     """Verify factual claims against knowledge base."""
     # Implementation depends on your knowledge base
     # Could use retrieval + NLI, or fact-checking API
-    pass
+    raise NotImplementedError("A verified fact-checking adapter is required")
 ```
 
 ## LLM-as-Judge Patterns
@@ -308,18 +308,18 @@ def calculate_agreement(rater1_scores, rater2_scores):
     """Calculate inter-rater agreement."""
     kappa = cohen_kappa_score(rater1_scores, rater2_scores)
 
-    interpretation = {
-        kappa < 0: "Poor",
-        kappa < 0.2: "Slight",
-        kappa < 0.4: "Fair",
-        kappa < 0.6: "Moderate",
-        kappa < 0.8: "Substantial",
-        kappa <= 1.0: "Almost Perfect"
-    }
+    # Ordered ranges: Boolean dictionary keys overwrite each other.
+    interpretation = "Undefined"
+    for upper, label in [(0, "Poor"), (0.2, "Slight"), (0.4, "Fair"),
+                         (0.6, "Moderate"), (0.8, "Substantial"),
+                         (float("inf"), "Almost Perfect")]:
+        if kappa < upper:
+            interpretation = label
+            break
 
     return {
         "kappa": kappa,
-        "interpretation": interpretation[True]
+        "interpretation": interpretation
     }
 ```
 
@@ -347,23 +347,31 @@ class ABTest:
         a_scores = self.variant_a["scores"]
         b_scores = self.variant_b["scores"]
 
+        if len(a_scores) < 2 or len(b_scores) < 2:
+            raise ValueError("At least two observations per independent group are required")
+        if not np.all(np.isfinite(a_scores)) or not np.all(np.isfinite(b_scores)):
+            raise ValueError("Require finite observations")
+        # Illustrative independent-group analysis; use a paired design for matched tasks.
         # T-test
         t_stat, p_value = stats.ttest_ind(a_scores, b_scores)
 
         # Effect size (Cohen's d)
         pooled_std = np.sqrt((np.std(a_scores)**2 + np.std(b_scores)**2) / 2)
-        cohens_d = (np.mean(b_scores) - np.mean(a_scores)) / pooled_std
+        cohens_d = (np.mean(b_scores) - np.mean(a_scores)) / pooled_std if pooled_std > 0 else None
 
         return {
             "variant_a_mean": np.mean(a_scores),
             "variant_b_mean": np.mean(b_scores),
             "difference": np.mean(b_scores) - np.mean(a_scores),
-            "relative_improvement": (np.mean(b_scores) - np.mean(a_scores)) / np.mean(a_scores),
+            "relative_improvement": ((np.mean(b_scores) - np.mean(a_scores)) / np.mean(a_scores)
+                                     if np.mean(a_scores) != 0 else None),
             "p_value": p_value,
             "statistically_significant": p_value < alpha,
             "cohens_d": cohens_d,
-            "effect_size": self.interpret_cohens_d(cohens_d),
-            "winner": "B" if np.mean(b_scores) > np.mean(a_scores) else "A"
+            "effect_size": self.interpret_cohens_d(cohens_d) if cohens_d is not None else "undefined",
+            "winner": (("B" if np.mean(b_scores) > np.mean(a_scores) else "A")
+                       if np.isfinite(p_value) and p_value < alpha and np.mean(b_scores) != np.mean(a_scores)
+                       else "inconclusive")
         }
 
     @staticmethod
@@ -385,26 +393,42 @@ class ABTest:
 ### Regression Detection
 ```python
 class RegressionDetector:
-    def __init__(self, baseline_results, threshold=0.05):
+    def __init__(self, baseline_results, threshold=0.05, directions=None):
+        import math
+        if not baseline_results or not math.isfinite(threshold) or threshold < 0:
+            raise ValueError("Require baseline metrics and a finite nonnegative threshold")
         self.baseline = baseline_results
         self.threshold = threshold
+        self.directions = directions or {}
 
     def check_for_regression(self, new_results):
         """Detect if new results show regression."""
+        import math
         regressions = []
+        missing_metrics = []
 
         for metric in self.baseline.keys():
             baseline_score = self.baseline[metric]
             new_score = new_results.get(metric)
 
             if new_score is None:
+                missing_metrics.append(metric)
                 continue
+            if not math.isfinite(baseline_score) or not math.isfinite(new_score):
+                raise ValueError("Metrics must be finite")
 
             # Calculate relative change
+            if baseline_score <= 0:
+                raise ValueError("Relative regression requires a positive nonzero baseline; use an absolute threshold instead.")
             relative_change = (new_score - baseline_score) / baseline_score
 
-            # Flag if significant decrease
-            if relative_change < -self.threshold:
+            # Default is higher-is-better; explicitly mark latency/cost as lower.
+            direction = self.directions.get(metric, "higher")
+            if direction not in {"higher", "lower"}:
+                raise ValueError("Unknown metric direction")
+            is_regression = (relative_change < -self.threshold if direction == "higher"
+                             else relative_change > self.threshold)
+            if is_regression:
                 regressions.append({
                     "metric": metric,
                     "baseline": baseline_score,
@@ -413,58 +437,22 @@ class RegressionDetector:
                 })
 
         return {
-            "has_regression": len(regressions) > 0,
+            "has_regression": True if regressions else (None if missing_metrics else False),
+            "analysis_complete": not missing_metrics,
+            "missing_metrics": missing_metrics,
             "regressions": regressions
         }
 ```
 
 ## Benchmarking
 
-### Running Benchmarks
-```python
-class BenchmarkRunner:
-    def __init__(self, benchmark_dataset):
-        self.dataset = benchmark_dataset
-
-    def run_benchmark(self, model, metrics):
-        """Run model on benchmark and calculate metrics."""
-        results = {metric.name: [] for metric in metrics}
-
-        for example in self.dataset:
-            # Generate prediction
-            prediction = model.predict(example["input"])
-
-            # Calculate each metric
-            for metric in metrics:
-                score = metric.calculate(
-                    prediction=prediction,
-                    reference=example["reference"],
-                    context=example.get("context")
-                )
-                results[metric.name].append(score)
-
-        # Aggregate results
-        return {
-            metric: {
-                "mean": np.mean(scores),
-                "std": np.std(scores),
-                "min": min(scores),
-                "max": max(scores)
-            }
-            for metric, scores in results.items()
-        }
-```
+Read [the local benchmark example](references/benchmarking.md) for a bounded runner, adapter prerequisites and incomplete-data handling.
+
+## Local Resources and execution boundaries
 
-## Resources
+The local benchmark reference contains an illustrative runner, not an installed evaluation framework or dataset. Code below illustrates patterns that must be adapted to the project's actual dependencies and validated before use; the `llm_eval` interface is pseudocode, not a promised installed package. Inspect existing SDK versions instead of copying legacy provider calls unchanged.
 
-- **references/metrics.md**: Comprehensive metric guide
-- **references/human-evaluation.md**: Annotation best practices
-- **references/benchmarking.md**: Standard benchmarks
-- **references/a-b-testing.md**: Statistical testing guide
-- **references/regression-testing.md**: CI/CD integration
-- **assets/evaluation-framework.py**: Complete evaluation harness
-- **assets/benchmark-dataset.jsonl**: Example datasets
-- **scripts/evaluate-model.py**: Automated evaluation runner
+Use sanitized, representative data and an explicit evaluation budget. Model downloads, hosted inference, paid judge calls and telemetry exports require authorization for the data and cost. Prefer already cached models; do not treat a prompt as a security boundary against instructions embedded in evaluated outputs. Calibrate judge scores against a human-labeled sample and inspect errors, bias and disagreement. Overlap scores do not prove correctness, safety or production readiness.
 
 ## Best Practices
 

package/bundled-skills/llm-evaluation/references/benchmarking.md

@@ -0,0 +1,35 @@
+# Benchmark runner example
+
+Supply the project's actual model and metric adapters and an already authorized, sanitized, materialized dataset. The pure control flow below does not supply inference, retrieval, external judges or a dataset. Bind tool versions, held-out coverage, hosted data/cost approval and a total run deadline before connecting a real model. Exceptions are incomplete runs, never a passed benchmark; retain the failed case and repair/re-run before publishing aggregate claims.
+
+```python
+import math
+import numpy as np
+
+class BenchmarkRunner:
+    def __init__(self, benchmark_dataset, max_examples=1000):
+        if not isinstance(max_examples, int) or max_examples < 1:
+            raise ValueError("Positive example limit required")
+        if not benchmark_dataset or len(benchmark_dataset) > max_examples:
+            raise ValueError("Require a nonempty bounded materialized dataset")
+        self.dataset = benchmark_dataset
+
+    def run_benchmark(self, model, metrics):
+        names = [metric.name for metric in metrics]
+        if not names or len(set(names)) != len(names):
+            raise ValueError("Require distinct declared metrics")
+        results = {name: [] for name in names}
+        for example in self.dataset:
+            prediction = model.predict(example["input"])
+            for metric in metrics:
+                score = metric.calculate(prediction=prediction,
+                    reference=example["reference"], context=example.get("context"))
+                if not isinstance(score, (int, float)) or not math.isfinite(score):
+                    raise ValueError("Missing or nonfinite metric result")
+                results[metric.name].append(score)
+        return {name: {"mean": np.mean(scores), "std": np.std(scores),
+                       "min": min(scores), "max": max(scores)}
+                for name, scores in results.items()}
+```
+
+Report dataset/model/metric identity, actual coverage, failed cases, observation count and uncertainty alongside these descriptive statistics. A successful local fake-adapter example is not a real model benchmark, and these aggregates do not establish factual accuracy, safety or legal compliance.

package/bundled-skills/logic-review/SKILL.md

@@ -9,6 +9,14 @@ date_added: 2026-07-01
 license: MIT
 license_source: https://github.com/hyhmrright/logic-lens/blob/main/LICENSE
 ---
+## Supporting material availability
+
+This catalog copy contains the inline guidance below, but does not include the following supporting paths. They are external prerequisites, not bundled files:
+
+- `<support-root>/scripts/grade-iteration.py`
+
+Obtain the required material from the [declared upstream source](https://github.com/hyhmrright/logic-lens), or from a user-provided installation of that source. Resolve each path relative to that installation, inspect its contents, and verify any applicable license and execution prerequisites. Do not run a command or claim a referenced procedure was followed until its required file exists. Do not fabricate the missing scripts or treat an unrelated file with the same name as equivalent. Independent steps fully documented here can proceed within the user’s authorized scope; report any dependent step as unavailable.
+
 
 # Logic-Lens — Logic Review
 ## When to Use

package/bundled-skills/longbridge/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: longbridge
-description: "125+ agent skills for Longbridge Securities — real-time quotes, charts, fundamentals, portfolio analysis, options, and more for HK/US/A-share/SG markets. Trilingual: Simplified Chinese, Traditional Chinese, English."
+description: "Guide to using a separately installed Longbridge integration — real-time quotes, charts, fundamentals, portfolio analysis, options, and more for HK/US/A-share/SG markets. Trilingual: Simplified Chinese, Traditional Chinese, English."
 category: finance
 risk: critical
 source: official
@@ -22,9 +22,9 @@ plugin:
 
 ## Overview
 
-Longbridge is the official skill collection for Longbridge Securities, covering 125+ agent skills across real-time market data, chart analysis, company fundamentals, portfolio management, options, sector screening, and more. Supports HK, US, A-share (SH/SZ), and SG markets. All skills are trilingual (Simplified Chinese / Traditional Chinese / English).
+Longbridge is the official skill collection for Longbridge Securities, covering a separately maintained set of skills across real-time market data, chart analysis, company fundamentals, portfolio management, options, sector screening, and more. Supports HK, US, A-share (SH/SZ), and SG markets. All skills are trilingual (Simplified Chinese / Traditional Chinese / English).
 
-Source repository: [github.com/longbridge/skills](https://github.com/longbridge/skills) (~840 stars, MIT)
+Source repository: [github.com/longbridge/skills](https://github.com/longbridge/skills) (verify the selected revision and license)
 
 ## When to Use This Skill
 
@@ -34,6 +34,10 @@ Source repository: [github.com/longbridge/skills](https://github.com/longbridge/
 - Use when the user wants options analysis, sector rankings, capital flow, or news
 - Use when the user asks in Chinese (Simplified or Traditional) or English about any securities topic
 
+## Scope and data verification
+
+AAS bundles this guide only, not the upstream collection or CLI. Inspect the actual installed version/tool schemas and current data entitlements. Verify market, exchange, currency, timestamp, delay and corporate-action adjustments; broad language support does not establish symbol coverage. Use existing scoped credentials. Reading a portfolio does not authorize orders, watchlist changes, transfers or broader trading permissions. Keep account data private and do not send it to another provider without consent.
+
 ## How It Works
 
 ### Step 1: Discover the Right Subcommand
@@ -86,7 +90,7 @@ If the `longbridge` CLI binary is not installed, fall back to MCP tools. Inspect
 - Portfolio and account features require login with Trade scope.
 - Real-time data is subject to Longbridge data subscription (delayed data available without subscription).
 - Crypto symbols use `.HAS` suffix on the Longbridge platform.
-- This skill does not place orders — read-only by default unless using the account write scope.
+- The default task is read-only. A trade-capable token is not authorization to place an order; exact account, instrument, side, quantity, order terms and costs require an explicit authorized action and timeout reconciliation.
 
 ## Security & Safety Notes
 

package/bundled-skills/longbridge-content/SKILL.md

@@ -9,6 +9,9 @@ date_added: 2026-07-01
 license: MIT
 license_source: https://github.com/longbridge/skills/blob/main/LICENSE
 ---
+## Bundled upstream support
+
+Supporting files are included at skill-relative paths. See [UPSTREAM-NOTICE.md](UPSTREAM-NOTICE.md) for their pinned sources, licenses, hashes, and import changes.
 
 # Longbridge Content
 

package/bundled-skills/longbridge-content/UPSTREAM-LICENSE.txt

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Longbridge Inc.
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/bundled-skills/longbridge-content/UPSTREAM-NOTICE.md

@@ -0,0 +1,15 @@
+# Upstream helpers for longbridge-content
+
+The files listed below were recovered from [longbridge/skills](https://github.com/longbridge/skills) at commit `66f45635ac0c7000b714ed0f34ab400b6818be74`. The repository ID is `1224163986`.
+
+They retain the upstream **MIT** license reproduced in [UPSTREAM-LICENSE.txt](UPSTREAM-LICENSE.txt). This notice applies to the listed support files; it does not replace the license metadata of other catalog files. Preserve the upstream license and this attribution when redistributing them.
+
+License wording is unchanged; text-only trailing whitespace and EOF blank lines are normalized. Original license: https://github.com/longbridge/skills/blob/66f45635ac0c7000b714ed0f34ab400b6818be74/LICENSE
+
+| Local file | Original upstream path | Original Git blob SHA | Import changes |
+| --- | --- | --- | --- |
+| `references/news.md` | `skills/longbridge-content/references/news.md` | `12499004cfa5f954020fc4612bc617da99de5370` | text-only trailing whitespace and EOF blank lines normalized |
+| `references/filing.md` | `skills/longbridge-content/references/filing.md` | `65798895c872e850917ed97e52cb74c3af19fbf6` | text-only trailing whitespace and EOF blank lines normalized |
+| `references/topic.md` | `skills/longbridge-content/references/topic.md` | `387d5fa97a4b4cdeaa79eb1e4f29ad958999c24e` | text-only trailing whitespace and EOF blank lines normalized |
+| `references/sec-filings.md` | `skills/longbridge-content/references/sec-filings.md` | `2ee84b941b993e396bcadfdc46ba1a4280f3f786` | text-only trailing whitespace and EOF blank lines normalized |
+| `references/regulatory-kb.md` | `skills/longbridge-content/references/regulatory-kb.md` | `57ba99b15b9b55a5dfb133a0fc65d983d9ad83d2` | text-only trailing whitespace and EOF blank lines normalized |

package/bundled-skills/longbridge-content/references/filing.md

@@ -0,0 +1,48 @@
+# filing
+
+```
+Regulatory filings for a symbol, or list/fetch filing content
+
+Without subcommand: lists filings for a symbol. Subcommands: list  detail Example: longbridge filing AAPL.US Example: longbridge filing list AAPL.US Example: longbridge filing detail AAPL.US 580265529766123777
+
+Usage: longbridge filing [OPTIONS] [SYMBOL] [COMMAND]
+
+Commands:
+  detail  Full Markdown content of a regulatory filing (HTML and TXT only)
+  help    Print this message or the help of the given subcommand(s)
+
+Arguments:
+  [SYMBOL]
+          Symbol in <CODE>.<MARKET> format (e.g. AAPL.US 700.HK). Omit when using a subcommand
+
+Options:
+      --count <COUNT>
+          Maximum number of filings to show (default: 20)
+
+          [default: 20]
+
+      --format <FORMAT>
+          Output format: 'pretty' for human-readable, 'json' for AI agents and scripting
+
+          [default: pretty]
+          [possible values: table, json]
+
+  -v, --verbose
+          Print verbose request info (host, elapsed) to stderr, prefixed with `*` like curl -v
+
+      --lang <LANG>
+          Language for content fetched from longbridge.com: zh-CN or en. Defaults to system LANG env var, then en
+
+  -h, --help
+          Print help (see a summary with '-h')
+```
+
+## Usage
+
+```bash
+# Run with JSON output for AI agents
+longbridge filing --format json
+
+# See all options
+longbridge filing --help
+```

package/bundled-skills/longbridge-content/references/news.md

@@ -0,0 +1,118 @@
+# longbridge-news
+
+Prompt-only skill that aggregates news, filings, and community topics for a single stock — classifies them, distils a fact-only takeaway, and reports sentiment at coarse granularity. May call WebSearch as a clearly-labelled fallback.
+
+## Depth selection
+
+| Prompt cue                               | Tools                                                      |
+| ---------------------------------------- | ---------------------------------------------------------- |
+| 新闻 / news / "最近怎么了"               | `news` only                                                |
+| 搜索新闻 / search news / keyword news    | `news search <keyword>`                                    |
+| 公告 / 披露 / filing / 8-K / 中报        | `filings` only                                             |
+| 市场怎么看 / market reaction / sentiment | `news` + `topic`                                           |
+| 社区 / community / discussion            | `topic` (+ `topic detail`, `topic replies` for hot topics) |
+| 搜索社区 / search community topics       | `topic search <keyword>`                                   |
+| 全面 / 综述 / overview (default)         | `news` + `filings` + `topic` (concurrent)                  |
+
+## CLI
+
+Run `longbridge <subcommand> --help` to verify exact flags. Default omnibus example (run concurrently):
+
+```bash
+longbridge news NVDA.US --format json
+longbridge filing NVDA.US --format json
+longbridge topic NVDA.US --format json
+```
+
+Keyword search (no symbol required):
+
+```bash
+longbridge news search "AI chips" --format json        # search news by keyword
+longbridge topic search "NVDA earnings" --format json  # search community topics by keyword
+```
+
+## Workflow
+
+1. Resolve symbol; if mapping fails, ask back (this is common with small-caps).
+2. Pick depth (table above) and call CLI commands concurrently (see CLI section). If `longbridge` is not installed, fall back to MCP.
+3.
+
+4. **Classify the news array** into 6 buckets (mandatory — never dump raw titles):
+
+   | Bucket                       | Cues                                                 |
+   | ---------------------------- | ---------------------------------------------------- |
+   | **catalyst** (业绩 / 基本面) | earnings, revenue, guidance, EPS, 财报, 营收, 利润   |
+   | **regulatory** (监管 / 合规) | SEC, 证监会, fine, lawsuit, 调查, 处罚, 罚款         |
+   | **strategic** (战略 / 业务)  | acquisition, partnership, launch, 收购, 拆分, 新产品 |
+   | **financial** (资本动作)     | buyback, split, dividend, 增发, 回购, 股权激励       |
+   | **opinion** (评级 / 目标价)  | upgrade, downgrade, analyst, 评级, 目标价            |
+   | **other**                    | unclassified                                         |
+
+5. Render the structured summary (template below). Skip empty buckets but **always include the key-takeaway summary** (≤ 100 chars, fact-only).
+
+## Output template
+
+```
+{Symbol} ({code}) news digest — Source: Longbridge Securities
+
+[Past N days · M news + K filings]
+
+🟢 Earnings / fundamentals (N)
+- [date] headline — one fact (number / ratio)
+
+🟡 Strategic / business (N)
+- [date] ...
+
+🔴 Regulatory / compliance (N)
+- [date] ...
+
+📈 Analyst opinions (N)
+- [date] {rating} by {bank}; target {currency} Y (was Z)
+
+📃 Filings
+- [date] 8-K / interim / earnings preview / ...
+
+💬 Community discussion (when topic was queried)
+- N hot topics: {title} — {comment count}
+- Sentiment skew: positive / neutral / negative — coarse %
+
+[Key facts] (≤ 100 chars, neutral)
+- ...
+
+⚠️ 以上数据仅供参考，不构成投资建议。/ 以上數據僅供參考，不構成投資建議。/ For reference only. Not investment advice.
+```
+
+(Translate into the user's language.)
+
+## WebSearch fallback (optional)
+
+Use **only** when:
+
+- MCP `news` returns empty, **or**
+- the latest item is > 7 days old, **or**
+- the user asks about a breaking event the MCP dataset has not yet captured.
+
+When you do, prepend: _"Below is from a web search — not Longbridge data."_ / _"以下为网络搜索结果,非长桥数据。"_ / _"以下為網絡搜尋結果,非長橋數據。"_
+
+## Output constraints
+
+- **Must** classify into the 6 buckets — never dump raw titles.
+- **Must** include the key-takeaway summary (≤ 100 chars, fact-only).
+- **Must** end with the not-investment-advice disclaimer.
+- 给有证据的方向性判断，例如 _"市场普遍将此解读为正面 / 负面"_；避免只贴 "利好 / 利空" 标签而不附事实。
+- **Do not** quote individual community comments verbatim (cherry-pick risk); report sentiment as a coarse skew.
+- **Do not** invent news. If MCP is sparse, say so and offer WebSearch.
+
+## Compliance: hype vocabulary
+
+If `topic` / `topic_replies` content contains a high density of hype words — _"涨停板"_, _"主升浪"_, _"必涨"_, _"满仓"_, _"all in"_, _"庄家"_, _"次新妖股"_ — do not echo them. Downgrade to _"discussion contains a notable share of speculative posts"_ and move on.
+
+## Error handling
+
+| Situation                       | Reply                                                                                                       |
+| ------------------------------- | ----------------------------------------------------------------------------------------------------------- |
+| `command not found: longbridge` | Fall back to MCP; if MCP also unavailable, tell user to install longbridge-terminal.                        |
+| `news` returns empty            | "{symbol} has no recent news in Longbridge data — switching to WebSearch (note: web data, not Longbridge)." |
+| `news` > 7 days stale           | Same — note the staleness explicitly.                                                                       |
+| Symbol mapping fails            | Ask the user for the code or English ticker.                                                                |
+| stderr `not logged in`          | Tell user to run `longbridge auth login`.                                                                   |

package/bundled-skills/longbridge-content/references/regulatory-kb.md

@@ -0,0 +1,128 @@
+# longbridge-regulatory-kb
+
+A structured regulatory knowledge base covering the rules, restrictions, and mechanisms investors face across A-shares, Hong Kong, US equity markets, and cryptocurrency — answers regulatory questions without needing live data lookups.
+
+## A-Share Rules (A股规则)
+
+### Price limits (涨跌停)
+
+| Board                          | Normal stocks             | ST / \*ST stocks |
+| ------------------------------ | ------------------------- | ---------------- |
+| Shanghai / Shenzhen Main Board | ±10%                      | ±5%              |
+| STAR Market (科创板)           | ±20% (first 5 days: ±30%) | N/A              |
+| ChiNext (创业板)               | ±20% (first 5 days: ±30%) | N/A              |
+| Beijing Stock Exchange         | ±30%                      | ±30%             |
+
+New listings: first day no limit; days 2–5 follow the limits above.
+
+### Settlement (交收)
+
+- **T+1**: Sell a stock on the day after purchase. Funds from sales available next day.
+- **T+0**: Not permitted for equities (day trading is prohibited for ordinary investors).
+- ETFs and convertible bonds: T+0 selling is allowed.
+
+### Short selling (融券做空)
+
+- Available only via margin accounts through approved brokers.
+- Securities in the approved short-list only (maintained by exchanges).
+- No naked short selling — must borrow first.
+- Transfer fee: typically 0.05–0.10% daily.
+
+### Circuit breakers (熔断)
+
+- Triggered when CSI 300 falls ≥5% (15-min halt) or ≥7% (market close for the day).
+- Note: implemented 2016-01-04, suspended 2016-01-08 due to market disruption.
+
+### Delisting rules (退市新规, 2024)
+
+- **Financial**: Net profit < 0 AND revenue < 100M RMB for 2 consecutive years → risk warning; 3 years → delisting.
+- **Market cap**: < 300M RMB for 20 consecutive trading days → delisting warning.
+- **Turnover**: < 2M RMB for 20 consecutive days → delisting warning.
+- ST / *ST designation: 1 year of losses → ST; 2 consecutive → *ST; further violations → delisting review.
+
+### Stamp duty (印花税)
+
+- 0.1% on sell side only (halved from 0.1% on both sides effective Aug 2023).
+
+## HK Stock Rules (港股规则)
+
+### Price limits
+
+- **No daily price limits** on the Hong Kong Stock Exchange.
+- Extreme volatility may trigger Volatility Control Mechanism (VCM): 5-minute cooling-off if price moves ±10% in 5 minutes.
+
+### Settlement (交收)
+
+- **T+2**: Standard settlement. Sell on day T; proceeds settled T+2.
+- **T+0 selling**: Allowed — you can sell shares purchased today if settled (e.g. shares transferred to your account already, which typically requires prior holding).
+
+### Short selling (卖空)
+
+- **Designated short-selling stocks only** — SEHK publishes the list.
+- No naked short selling — must borrow.
+- Short positions must be marked as "short" at order entry.
+
+### Odd-lot trading (碎股)
+
+- Board lot = standard trading unit (varies by stock: 100, 200, 500, 1000, 2000 shares, etc.).
+- Odd lots (less than one board lot) trade in the odd-lot market at wider spreads.
+- Retail investors often receive odd lots from rights issues or stock dividends.
+
+### Grey market / dark pool (暗盘)
+
+- Pre-IPO grey market: unofficial OTC trading in IPO shares before listing, typically evening before listing day.
+- Provides price discovery but is not regulated by SEHK; settlement risk exists.
+
+### Listing criteria / delisting (上市/退市)
+
+- **Profit test**: Aggregate profit ≥ HK$50M over 3 years (main board). Alternative: market cap test or revenue test.
+- **Delisting**: Prolonged trading suspension (typically > 18 months) triggers delisting review.
+
+### Stamp duty (印花税)
+
+- 0.1% on both buyer and seller (total 0.2% per round trip), effective from 2021.
+
+### Insider dealing (内幕交易条例)
+
+- Securities and Futures Ordinance (SFO) Cap. 571 governs insider dealing and market manipulation.
+- Directors must disclose dealings within 3 business days.
+
+## US Stock Rules (美股规则)
+
+Key rules: PDT ($25k minimum, 4 day-trades in 5-day window), Reg T 50% initial margin, T+1 settlement (since May 2024), S&P 500 circuit breakers at -7%/-13%/-20%, Reg SHO short-locate requirement, LULD per-stock bands ±5%.
+
+For full detail see [`references/us-rules.md`](references/us-rules.md).
+For cryptocurrency regulation and cross-border tax basics see [`references/crypto-and-tax.md`](references/crypto-and-tax.md).
+
+## CLI (supplementary)
+
+For live trading session status:
+
+```bash
+longbridge market-temp --format json   # session open/closed, market sentiment
+```
+
+For symbol-specific board lot or margin tier:
+
+```bash
+longbridge static <SYMBOL> --format json   # verify available fields with --help
+longbridge static --help
+```
+
+## Output
+
+For each regulatory question:
+
+1. Clear rule statement with the governing authority.
+2. Numeric thresholds (%, days, amounts).
+3. Consequences of violation / breach.
+4. Practical tip for the user.
+5. Caveat: rules evolve; link to official source when possible (CSRC / SFC / SEC / FINRA).
+
+## Error handling
+
+| Situation                             | 简体回复                                                           | 繁體回覆                                                           | English reply                                                                     |
+| ------------------------------------- | ------------------------------------------------------------------ | ------------------------------------------------------------------ | --------------------------------------------------------------------------------- |
+| `command not found: longbridge`       | 监管知识库无需 CLI。如需实时交易状态，请安装 longbridge-terminal。 | 監管知識庫無需 CLI。如需實時交易狀態，請安裝 longbridge-terminal。 | Regulatory KB needs no CLI. For live session status, install longbridge-terminal. |
+| Rule not covered                      | 该规则暂未收录，建议查阅 CSRC/SFC/SEC 官网。                       | 該規則暫未收錄，建議查閱 CSRC/SFC/SEC 官網。                       | Rule not yet covered; consult CSRC / SFC / SEC official website.                  |
+| User asks for specific account advice | 请咨询持牌财务顾问，本技能仅提供通用规则说明。                     | 請諮詢持牌財務顧問，本技能僅提供通用規則說明。                     | Consult a licensed financial adviser; this skill provides general rules only.     |