opencode-skills-collection 4.0.52 → 4.0.54

package/bundled-skills/llm-council/references/detailed-guide.md

@@ -0,0 +1,556 @@
+# LLM Council (Fireworks AI) - Detailed Guide
+
+> This file contains the detailed procedure and reference material extracted from `SKILL.md` for focused loading. The root skill defines activation, examples, safety constraints, and limitations.
+
+## Pre-flight Check
+
+Before running any phase, verify the Fireworks API key is set:
+
+```bash
+if [ -z "$FIREWORKS_API_KEY" ]; then
+  echo "ERROR: FIREWORKS_API_KEY is not set."
+  echo "Create a Fireworks AI account at: https://fireworks.ai/"
+  echo "Then export it in your shell profile (~/.zshrc or ~/.bashrc):"
+  echo '  read -rsp "Fireworks API key: " FIREWORKS_API_KEY; echo; export FIREWORKS_API_KEY'
+  exit 1
+fi
+echo "FIREWORKS_API_KEY is set."
+```
+
+
+## Available Models
+
+Present these options to the user via AskUserQuestion (multiselect):
+
+| Model | Fireworks ID | Provider |
+|-------|-------------|----------|
+| GLM 5 | accounts/fireworks/models/glm-5 | Z.ai |
+| DeepSeek V3.1 | accounts/fireworks/models/deepseek-v3p1 | DeepSeek |
+| DeepSeek V3.2 | accounts/fireworks/models/deepseek-v3p2 | DeepSeek |
+| MiniMax M2.1 | accounts/fireworks/models/minimax-m2p1 | MiniMax |
+| Kimi K2.5 | accounts/fireworks/models/kimi-k2p5 | Moonshot |
+| Qwen3 235B | accounts/fireworks/models/qwen3-235b-a22b | Alibaba |
+| Llama 4 Maverick | accounts/fireworks/models/llama4-maverick-instruct-basic | Meta |
+
+
+## Workflow
+
+### Step 1: Gather User Input
+
+Use AskUserQuestion to get:
+1. The query/question for the council (or accept it from the conversation)
+2. Which models to include (multiselect, recommend 3-5 models)
+3. Which model should be the Chairman (single select)
+
+Note: AskUserQuestion supports max 4 options per question. Since there are 7 models, split model selection across two questions, or show the most popular 4 and let the user type "Other" for the rest. A good default is to show 4 models in the first question and note the others are available via "Other". Rotate which models are shown based on variety.
+
+Example AskUserQuestion for model selection (show 4, mention others):
+```
+question: "Which models should participate in the LLM Council? (Also available via Other: Llama 4 Maverick, Qwen3 235B, GLM 5)"
+header: "Models"
+multiSelect: true
+options:
+  - label: "DeepSeek V3.2"
+    description: "DeepSeek's newest and most capable model"
+  - label: "MiniMax M2.1"
+    description: "MiniMax's strong open-weight model"
+  - label: "Kimi K2.5"
+    description: "Moonshot's strong open-weight model"
+  - label: "DeepSeek V3.1"
+    description: "DeepSeek's proven reasoning model"
+```
+
+Example AskUserQuestion for chairman:
+```
+question: "Which model should be the Chairman (synthesizes the final answer)?"
+header: "Chairman"
+multiSelect: false
+options:
+  - label: "DeepSeek V3.2 (Recommended)"
+    description: "Newest DeepSeek, strong at comprehensive analysis"
+  - label: "GLM 5"
+    description: "Strong reasoning for synthesis"
+  - label: "Kimi K2.5"
+    description: "Strong at structured synthesis"
+  - label: "MiniMax M2.1"
+    description: "Strong open-weight model for synthesis"
+```
+
+### Model Name to ID Mapping
+
+Use this mapping to convert user selections to Fireworks model IDs:
+
+```python
+MODEL_MAP = {
+    "GLM 5": "accounts/fireworks/models/glm-5",
+    "DeepSeek V3.1": "accounts/fireworks/models/deepseek-v3p1",
+    "DeepSeek V3.2": "accounts/fireworks/models/deepseek-v3p2",
+    "MiniMax M2.1": "accounts/fireworks/models/minimax-m2p1",
+    "Kimi K2.5": "accounts/fireworks/models/kimi-k2p5",
+    "Qwen3 235B": "accounts/fireworks/models/qwen3-235b-a22b",
+    "Llama 4 Maverick": "accounts/fireworks/models/llama4-maverick-instruct-basic",
+}
+```
+
+### Step 2: Run Phase 1 - Individual Responses
+
+After gathering input, run this script to get responses from all selected models in parallel:
+
+```bash
+QUERY="USER_QUERY_HERE"
+MODELS='["accounts/fireworks/models/glm-5", "accounts/fireworks/models/deepseek-v3p1"]'
+
+python3 << 'PYEOF'
+import os
+import json
+import requests
+import time
+from concurrent.futures import ThreadPoolExecutor, as_completed
+
+FIREWORKS_API_KEY = os.environ.get("FIREWORKS_API_KEY")
+API_URL = "https://api.fireworks.ai/inference/v1/chat/completions"
+
+QUERY = os.environ.get("QUERY", "")
+MODELS = json.loads(os.environ.get("MODELS", "[]"))
+
+# Create session directory
+timestamp = time.strftime("%Y%m%d-%H%M%S")
+SESSION_DIR = f"/tmp/llm-council/{timestamp}"
+os.makedirs(SESSION_DIR, exist_ok=True)
+
+# Save config
+config = {"query": QUERY, "models": MODELS, "timestamp": timestamp}
+with open(f"{SESSION_DIR}/config.json", "w") as f:
+    json.dump(config, f, indent=2)
+
+def call_model(model_id, query):
+    """Call a single model via Fireworks AI"""
+    try:
+        start = time.time()
+        response = requests.post(
+            API_URL,
+            headers={
+                "Authorization": f"Bearer {FIREWORKS_API_KEY}",
+                "Content-Type": "application/json"
+            },
+            json={
+                "model": model_id,
+                "messages": [
+                    {"role": "system", "content": "You are participating in an LLM council deliberation. Provide your best, most thoughtful response to the query. Be comprehensive but focused."},
+                    {"role": "user", "content": query}
+                ],
+                "max_tokens": 4000,
+                "temperature": 1
+            },
+            timeout=120
+        )
+        response.raise_for_status()
+        elapsed = time.time() - start
+        data = response.json()
+        usage = data.get("usage", {})
+        return {
+            "success": True,
+            "content": data["choices"][0]["message"]["content"],
+            "model": model_id,
+            "latency_seconds": round(elapsed, 2),
+            "tokens": {
+                "prompt": usage.get("prompt_tokens", 0),
+                "completion": usage.get("completion_tokens", 0),
+                "total": usage.get("total_tokens", 0)
+            }
+        }
+    except Exception as e:
+        return {
+            "success": False,
+            "content": f"[ERROR: {str(e)}]",
+            "model": model_id,
+            "latency_seconds": 0,
+            "tokens": {"prompt": 0, "completion": 0, "total": 0}
+        }
+
+print(f"\n{'='*60}")
+print("PHASE 1: Collecting Individual Responses")
+print(f"{'='*60}")
+print(f"Query: {QUERY[:200]}...")
+print(f"Models: {', '.join([m.split('/')[-1] for m in MODELS])}")
+print(f"Session: {SESSION_DIR}")
+print()
+
+# Parallel execution
+results = {}
+with ThreadPoolExecutor(max_workers=len(MODELS)) as executor:
+    futures = {executor.submit(call_model, m, QUERY): m for m in MODELS}
+    for future in as_completed(futures):
+        model = futures[future]
+        result = future.result()
+        results[model] = result
+        status = "OK" if result["success"] else "FAILED"
+        latency = f"{result['latency_seconds']}s" if result["success"] else "N/A"
+        print(f"  [{status}] {model.split('/')[-1]} ({latency})")
+
+# Save raw results
+with open(f"{SESSION_DIR}/phase1_responses.json", "w") as f:
+    json.dump(results, f, indent=2)
+
+print(f"\nPhase 1 complete. Results saved to: {SESSION_DIR}/phase1_responses.json")
+print(f"SESSION_DIR={SESSION_DIR}")
+PYEOF
+```
+
+### Step 3: Run Phase 2 - Cross-Model Ranking
+
+Each model reviews and ranks the anonymized responses from Phase 1:
+
+```bash
+SESSION_DIR="/tmp/llm-council/TIMESTAMP_HERE"
+
+python3 << 'PYEOF'
+import os
+import json
+import requests
+import time
+from concurrent.futures import ThreadPoolExecutor, as_completed
+
+FIREWORKS_API_KEY = os.environ.get("FIREWORKS_API_KEY")
+API_URL = "https://api.fireworks.ai/inference/v1/chat/completions"
+SESSION_DIR = os.environ.get("SESSION_DIR")
+
+# Load Phase 1 results
+with open(f"{SESSION_DIR}/config.json") as f:
+    config = json.load(f)
+with open(f"{SESSION_DIR}/phase1_responses.json") as f:
+    phase1_results = json.load(f)
+
+QUERY = config["query"]
+MODELS = config["models"]
+
+# Create anonymized mapping
+labels = ["A", "B", "C", "D", "E", "F", "G"][:len(MODELS)]
+model_to_label = dict(zip(MODELS, labels))
+label_to_model = {v: k for k, v in model_to_label.items()}
+
+# Format anonymized responses
+anonymized_responses = []
+for model_id in MODELS:
+    label = model_to_label[model_id]
+    content = phase1_results[model_id]["content"]
+    anonymized_responses.append(f"=== Response {label} ===\n{content}")
+
+anonymized_text = "\n\n".join(anonymized_responses)
+
+def get_rankings(model_id, query, anonymized, own_label):
+    """Get rankings from a single model"""
+    ranking_prompt = f"""You are evaluating responses from multiple AI models to this query:
+
+QUERY: {query}
+
+Here are the anonymized responses:
+
+{anonymized}
+
+Please rank these responses from BEST to WORST. For each ranking:
+1. State the response letter (A, B, C, etc.)
+2. Give a brief reason (1-2 sentences)
+3. You may skip ranking your own response (labeled {own_label}) or rank it fairly
+
+Format your response EXACTLY as:
+RANKINGS:
+1. [Letter] - [Brief reason]
+2. [Letter] - [Brief reason]
+3. [Letter] - [Brief reason]
+..."""
+
+    try:
+        start = time.time()
+        response = requests.post(
+            API_URL,
+            headers={
+                "Authorization": f"Bearer {FIREWORKS_API_KEY}",
+                "Content-Type": "application/json"
+            },
+            json={
+                "model": model_id,
+                "messages": [
+                    {"role": "system", "content": f"You are ranking AI responses objectively. Your own response is labeled '{own_label}'."},
+                    {"role": "user", "content": ranking_prompt}
+                ],
+                "max_tokens": 1000,
+                "temperature": 1
+            },
+            timeout=90
+        )
+        response.raise_for_status()
+        elapsed = time.time() - start
+        return {
+            "success": True,
+            "content": response.json()["choices"][0]["message"]["content"],
+            "model": model_id,
+            "latency_seconds": round(elapsed, 2)
+        }
+    except Exception as e:
+        return {
+            "success": False,
+            "content": f"[ERROR: {str(e)}]",
+            "model": model_id,
+            "latency_seconds": 0
+        }
+
+print(f"\n{'='*60}")
+print("PHASE 2: Cross-Model Ranking")
+print(f"{'='*60}")
+print(f"Label mapping: {json.dumps({v: k.split('/')[-1] for k, v in model_to_label.items()})}")
+print()
+
+# Collect rankings from all models in parallel
+rankings = {}
+with ThreadPoolExecutor(max_workers=len(MODELS)) as executor:
+    futures = {
+        executor.submit(get_rankings, mid, QUERY, anonymized_text, model_to_label[mid]): mid
+        for mid in MODELS
+    }
+    for future in as_completed(futures):
+        model = futures[future]
+        result = future.result()
+        rankings[model] = result
+        status = "OK" if result["success"] else "FAILED"
+        latency = f"{result['latency_seconds']}s" if result["success"] else "N/A"
+        print(f"  [{status}] {model.split('/')[-1]} ({latency})")
+
+# Save rankings
+output = {
+    "label_mapping": label_to_model,
+    "model_to_label": model_to_label,
+    "rankings": rankings
+}
+with open(f"{SESSION_DIR}/phase2_rankings.json", "w") as f:
+    json.dump(output, f, indent=2)
+
+print(f"\nPhase 2 complete. Rankings saved to: {SESSION_DIR}/phase2_rankings.json")
+PYEOF
+```
+
+### Step 4: Run Phase 3 - Chairman Synthesis
+
+The Chairman model receives all responses and rankings, then produces the final synthesis:
+
+```bash
+SESSION_DIR="/tmp/llm-council/TIMESTAMP_HERE"
+CHAIRMAN_MODEL="accounts/fireworks/models/glm-5"
+
+python3 << 'PYEOF'
+import os
+import json
+import requests
+import time
+
+FIREWORKS_API_KEY = os.environ.get("FIREWORKS_API_KEY")
+API_URL = "https://api.fireworks.ai/inference/v1/chat/completions"
+SESSION_DIR = os.environ.get("SESSION_DIR")
+CHAIRMAN_MODEL = os.environ.get("CHAIRMAN_MODEL")
+
+# Load all previous results
+with open(f"{SESSION_DIR}/config.json") as f:
+    config = json.load(f)
+with open(f"{SESSION_DIR}/phase1_responses.json") as f:
+    phase1 = json.load(f)
+with open(f"{SESSION_DIR}/phase2_rankings.json") as f:
+    phase2 = json.load(f)
+
+QUERY = config["query"]
+label_to_model = phase2["label_mapping"]
+model_to_label = phase2["model_to_label"]
+
+# Format responses with model names revealed
+responses_text = []
+for model_id, result in phase1.items():
+    label = model_to_label.get(model_id, "?")
+    model_name = model_id.split("/")[-1]
+    responses_text.append(f"=== {label}: {model_name} ===\n{result['content']}")
+
+# Format rankings
+rankings_text = []
+for model_id, result in phase2["rankings"].items():
+    model_name = model_id.split("/")[-1]
+    rankings_text.append(f"[{model_name}'s Rankings]\n{result['content']}")
+
+synthesis_prompt = f"""You are the Chairman of an LLM Council. Your task is to synthesize the best possible answer from multiple AI responses.
+
+ORIGINAL QUERY:
+{QUERY}
+
+INDIVIDUAL RESPONSES:
+{chr(10).join(responses_text)}
+
+MODEL RANKINGS:
+{chr(10).join(rankings_text)}
+
+As Chairman, produce a FINAL SYNTHESIS that:
+1. Incorporates the strongest elements from the best-ranked responses
+2. Resolves any contradictions between responses
+3. Addresses aspects that multiple models agreed on
+4. Corrects any errors identified through cross-ranking
+5. Provides the most complete, accurate, and helpful answer
+
+Begin your synthesis:"""
+
+print(f"\n{'='*60}")
+print("PHASE 3: Chairman Synthesis")
+print(f"{'='*60}")
+print(f"Chairman: {CHAIRMAN_MODEL.split('/')[-1]}")
+print()
+
+try:
+    start = time.time()
+    response = requests.post(
+        API_URL,
+        headers={
+            "Authorization": f"Bearer {FIREWORKS_API_KEY}",
+            "Content-Type": "application/json"
+        },
+        json={
+            "model": CHAIRMAN_MODEL,
+            "messages": [
+                {"role": "system", "content": "You are the Chairman of an LLM Council. Synthesize multiple AI perspectives into a definitive, comprehensive response."},
+                {"role": "user", "content": synthesis_prompt}
+            ],
+            "max_tokens": 4000,
+            "temperature": 1
+        },
+        timeout=180
+    )
+    response.raise_for_status()
+    elapsed = time.time() - start
+    synthesis = response.json()["choices"][0]["message"]["content"]
+
+    with open(f"{SESSION_DIR}/phase3_synthesis.txt", "w") as f:
+        f.write(synthesis)
+
+    print(f"Phase 3 complete ({elapsed:.2f}s). Synthesis saved to: {SESSION_DIR}/phase3_synthesis.txt")
+
+except Exception as e:
+    print(f"ERROR: {e}")
+    synthesis = f"[ERROR: {str(e)}]"
+    with open(f"{SESSION_DIR}/phase3_synthesis.txt", "w") as f:
+        f.write(synthesis)
+
+# Update config with chairman
+config["chairman"] = CHAIRMAN_MODEL
+with open(f"{SESSION_DIR}/config.json", "w") as f:
+    json.dump(config, f, indent=2)
+PYEOF
+```
+
+### Step 5: Display Full Results
+
+Read all saved files and display the complete council deliberation:
+
+```bash
+SESSION_DIR="/tmp/llm-council/TIMESTAMP_HERE"
+
+python3 << 'PYEOF'
+import os
+import json
+
+SESSION_DIR = os.environ.get("SESSION_DIR")
+
+# Load all data
+with open(f"{SESSION_DIR}/config.json") as f:
+    config = json.load(f)
+with open(f"{SESSION_DIR}/phase1_responses.json") as f:
+    phase1 = json.load(f)
+with open(f"{SESSION_DIR}/phase2_rankings.json") as f:
+    phase2 = json.load(f)
+with open(f"{SESSION_DIR}/phase3_synthesis.txt") as f:
+    synthesis = f.read()
+
+model_to_label = phase2["model_to_label"]
+label_to_model = phase2["label_mapping"]
+
+# Build formatted output
+output = []
+output.append("=" * 70)
+output.append("                  LLM COUNCIL DELIBERATION")
+output.append("                  Powered by Fireworks AI")
+output.append("=" * 70)
+output.append("")
+output.append(f"QUERY: {config['query']}")
+output.append(f"COUNCIL: {', '.join([m.split('/')[-1] for m in config['models']])}")
+output.append(f"CHAIRMAN: {config.get('chairman', 'N/A').split('/')[-1]}")
+output.append("")
+
+# Phase 1: Individual Responses
+output.append("-" * 70)
+output.append("                 PHASE 1: INDIVIDUAL RESPONSES")
+output.append("-" * 70)
+output.append("")
+
+for model_id, result in phase1.items():
+    model_name = model_id.split("/")[-1]
+    label = model_to_label.get(model_id, "?")
+    latency = result.get("latency_seconds", "N/A")
+    tokens = result.get("tokens", {})
+    output.append(f"[{label}] {model_name} (latency: {latency}s, tokens: {tokens.get('total', 'N/A')})")
+    output.append("-" * 40)
+    output.append(result["content"])
+    output.append("")
+
+# Phase 2: Cross-Model Rankings
+output.append("-" * 70)
+output.append("                 PHASE 2: CROSS-MODEL RANKINGS")
+output.append("-" * 70)
+output.append("")
+output.append(f"Label mapping: {json.dumps({v: k.split('/')[-1] for k, v in model_to_label.items()}, indent=2)}")
+output.append("")
+
+for model_id, result in phase2["rankings"].items():
+    model_name = model_id.split("/")[-1]
+    output.append(f"[{model_name}'s Rankings]")
+    output.append(result["content"])
+    output.append("")
+
+# Phase 3: Chairman Synthesis
+output.append("-" * 70)
+output.append("                 PHASE 3: CHAIRMAN'S SYNTHESIS")
+output.append("-" * 70)
+output.append("")
+chairman_name = config.get("chairman", "Chairman").split("/")[-1]
+output.append(f"[{chairman_name} - Chairman]")
+output.append("")
+output.append(synthesis)
+output.append("")
+output.append("=" * 70)
+output.append(f"Session files: {SESSION_DIR}/")
+
+# Save formatted output
+final_output = "\n".join(output)
+with open(f"{SESSION_DIR}/final_output.md", "w") as f:
+    f.write(final_output)
+
+print(final_output)
+print(f"\nFull output saved to: {SESSION_DIR}/final_output.md")
+PYEOF
+```
+
+
+## Important Notes
+
+1. **Session Directory**: Each run creates a unique session in `/tmp/llm-council/{timestamp}/`
+2. **Raw Data Preserved**: All API responses are saved as-is to JSON files for full transparency
+3. **Cost**: Fireworks pricing is per-token. More models and longer queries cost more. Check current pricing at https://fireworks.ai/pricing
+4. **Latency Tracking**: Each API call tracks latency so you can see Fireworks' speed in action
+5. **Token Usage**: Phase 1 responses include token counts for cost awareness
+6. **Rate Limits**: If you hit rate limits, wait briefly and retry
+7. **Model Availability**: Check https://app.fireworks.ai/ for current model status
+
+
+## Setup
+
+1. Create a Fireworks AI account at https://fireworks.ai/ and grab your API key from the dashboard
+2. Export it in your shell profile:
+   ```bash
+   read -rsp "Fireworks API key: " FIREWORKS_API_KEY
+   echo
+   export FIREWORKS_API_KEY
+   ```
+3. Restart your terminal or run `source ~/.zshrc`
+4. Invoke this skill when you want multiple open-weight AI perspectives on a question

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

@@ -257,6 +257,12 @@ Criterios:
 - Applying recommendations without understanding your specific context
 - Not providing enough project context for accurate analysis
 
+## Example
+
+**User request:**
+
+> Use @llm-ops for this task: LLM Operations -- RAG, embeddings, vector databases, fine-tuning, prompt engineering avancado, custos de LLM, evals de qualidade e arquiteturas de IA para producao.
+
 ## Limitations
 - Use this skill only when the task clearly matches the scope described above.
 - Do not treat the output as a substitute for environment-specific validation, testing, or expert review.

package/bundled-skills/local-legal-seo-audit/SKILL.md

@@ -346,6 +346,12 @@ Same format
 - **ai-seo**: For optimizing legal content for AI search experiences and featured snippets.
 - **page-cro**: For improving conversion rate on practice area pages and contact forms.
 
+## Example
+
+**User request:**
+
+> Audit or improve local SEO for a law firm, attorney, forensic expert, or similar legal/professional services website.
+
 ## Limitations
 - Use this skill only when the task clearly matches the scope described above.
 - Do not treat the output as a substitute for environment-specific validation, testing, or expert review.

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

@@ -1,6 +1,6 @@
 ---
 name: logic-diff
-description: Compare two code versions for semantic equivalence via semi-formal tracing of both versions side-by-side. Trigger when the user shares a refactor, rewrite, migration, or A/B implementation and wants to confirm behavior is unchanged — "did I break anything", "is this equivalent", "are...
+description: "Compare two code versions for semantic equivalence via semi-formal tracing of both versions side-by-side."
 risk: safe
 source: https://github.com/hyhmrright/logic-lens/tree/main/skills/logic-diff
 source_repo: hyhmrright/logic-lens

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

@@ -1,6 +1,6 @@
 ---
 name: logic-explain
-description: Explain what a specific piece of code actually does for a given input by producing a step-by-step execution trace (interprocedural, with name resolution and type transitions). Trigger when the user is confused about behavior or asks why code produces X instead of Y — "walk me through...
+description: "Explain what a specific piece of code actually does for a given input by producing a step-by-step execution trace (interprocedural, with name resolution and type transitions)."
 risk: safe
 source: https://github.com/hyhmrright/logic-lens/tree/main/skills/logic-explain
 source_repo: hyhmrright/logic-lens
@@ -43,6 +43,12 @@ Note: `logic-risks.md` is intentionally skipped — logic-explain does not produ
 
 **Note:** Execution Explain is descriptive, not evaluative. Omit the Logic Score / Fault Confidence / Verdict line from the report header.
 
+## Example
+
+**User request:**
+
+> Explain what a specific piece of code actually does for a given input by producing a step-by-step execution trace (interprocedural, with name resolution and type transitions).
+
 ## Limitations
 
 - Use this skill only when the task clearly matches its upstream source and local project context.

package/bundled-skills/logic-fix-all/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: logic-fix-all
-description: 'Autonomous repository-wide audit-and-fix pipeline: health → review → locate/explain → fix → diff-verify → iterate until clean. Starts with a mandatory consent prompt (token-intensive); after consent runs hands-free. Trigger when the user wants ALL logic issues found and fixed — "fix...'
+description: "Autonomous repository-wide audit-and-fix pipeline: health → review → locate/explain → fix → diff-verify → iterate until clean. Starts with a mandatory consent prompt (token-intensive); after consent runs hands-free."
 risk: critical
 source: https://github.com/hyhmrright/logic-lens/tree/main/skills/logic-fix-all
 source_repo: hyhmrright/logic-lens

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

@@ -1,6 +1,6 @@
 ---
 name: logic-locate
-description: Locate the root cause of a CONFIRMED failure via backward-then-forward semi-formal tracing. Trigger when the user provides a stack trace, failing assertion, error message, or specific wrong-value observation — "find the bug", "this test is failing", "track down this crash", "why is...
+description: "Locate the root cause of a CONFIRMED failure via backward-then-forward semi-formal tracing."
 risk: safe
 source: https://github.com/hyhmrright/logic-lens/tree/main/skills/logic-locate
 source_repo: hyhmrright/logic-lens

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

@@ -1,6 +1,6 @@
 ---
 name: logic-review
-description: Find logic bugs in a single file or function via semi-formal execution tracing (Premises → Trace → Divergence → Trigger → Remedy). Trigger when a user shares code and suspects something is wrong without naming a concrete failure — phrases like "review this", "does this look right",...
+description: "Find logic bugs in a single file or function via semi-formal execution tracing (Premises → Trace → Divergence → Trigger → Remedy)."
 risk: critical
 source: https://github.com/hyhmrright/logic-lens/tree/main/skills/logic-review
 source_repo: hyhmrright/logic-lens