From be96f240b32e04d696e7aad776e07e10eafaa402 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=9C=B1=E6=BD=AE?= Date: Tue, 12 May 2026 10:05:24 +0800 Subject: [PATCH] =?UTF-8?q?=E6=9B=B4=E6=96=B0=E5=87=A0=E4=B8=AA=20policy.m?= =?UTF-8?q?d?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- ...trieval-policy-forbidden-self-knowledge.md | 60 ++++++++----------- ...trieval-policy-forbidden-self-knowledge.md | 48 ++++++++------- ...trieval-policy-forbidden-self-knowledge.md | 34 +++++++---- .../hooks/retrieval-policy.md | 31 ++++++---- ...trieval-policy-forbidden-self-knowledge.md | 44 +++++++------- ...trieval-policy-forbidden-self-knowledge.md | 56 ++++++++++------- 6 files changed, 151 insertions(+), 122 deletions(-) diff --git a/skills/autoload/onprem/rag-retrieve/hooks/retrieval-policy-forbidden-self-knowledge.md b/skills/autoload/onprem/rag-retrieve/hooks/retrieval-policy-forbidden-self-knowledge.md index b6c1296..2dfa82e 100644 --- a/skills/autoload/onprem/rag-retrieve/hooks/retrieval-policy-forbidden-self-knowledge.md +++ b/skills/autoload/onprem/rag-retrieve/hooks/retrieval-policy-forbidden-self-knowledge.md @@ -14,7 +14,7 @@ For knowledge retrieval tasks, **this policy overrides generic codebase explorat - **Prohibited answer source**: the model's own parametric knowledge, memory, prior world knowledge, intuition, common sense completion, or unsupported inference. - **Prohibited tools**: `Glob`, `Read`, `LS`, Bash (`ls`, `find`, `cat`, `head`, `tail`, `grep`, etc.) — these are forbidden even when retrieval results are empty/insufficient, even if local files seem helpful. -- **Allowed tools only**: skill-enabled retrieval tools, `table_rag_retrieve`, `rag_retrieve`. No other source for factual answering. +- **Allowed tools only**: skill-enabled retrieval tools, `rag_retrieve`. No other source for factual answering. - Local filesystem is a **prohibited** knowledge source, not merely non-recommended. - Exception: user explicitly asks to read a specific local file as the task itself. - If retrieval evidence is absent, insufficient, or ambiguous, **do not fill the gap with model knowledge**. @@ -35,9 +35,7 @@ For any knowledge retrieval task: Execute **sequentially, one at a time**. Do NOT run in parallel. Do NOT probe filesystem first. 1. **Skill-enabled retrieval tools** (use first when available) -2. **`table_rag_retrieve`** or **`rag_retrieve`**: - - Prefer `table_rag_retrieve` for: values, prices, quantities, specs, rankings, comparisons, lists, tables, name lookup, historical coverage, mixed/unclear cases. - - Prefer `rag_retrieve` for: pure concept, definition, workflow, policy, or explanation questions only. +2. **`rag_retrieve`** - After each step, evaluate sufficiency before proceeding. - Retrieval must happen **before** any factual answer generation. @@ -50,27 +48,35 @@ Execute **sequentially, one at a time**. Do NOT run in parallel. Do NOT probe fi ## 5. Retrieval Breadth (`top_k`) -- Apply `top_k` only to `rag_retrieve`. Use smallest sufficient value, expand if insufficient. -- `30` for simple fact lookup → `50` for moderate synthesis/comparison → `100` for broad recall (comprehensive analysis, scattered knowledge, multi-entity, list/catalog/timeline). -- Expansion order: `30 → 50 → 100`. If unsure, use `100`. +- Apply `top_k` only to `rag_retrieve`. Choose the appropriate value upfront to maximize first-call success. +- Use `50` for simple fact lookup or moderate synthesis, comparison, summarization, disambiguation. +- Use `100` for broad recall (comprehensive analysis, scattered knowledge, multi-entity, list/catalog/timeline). +- If unsure, use `50`. Only escalate to `100` on the retry call if first results are insufficient. ## 6. Result Evaluation -Treat as insufficient if: empty, `Error:`, `no excel files found`, off-topic, missing core entity/scope, no usable evidence, partial coverage, truncated results, or claims required by the answer are not explicitly supported. +**Maximum 3 retrieval calls per question.** After each call, evaluate immediately: + +### Sufficient — answer now +- The core entity/topic in the user's question has been hit. +- There is direct evidence supporting the main intent of the question. +- Partial but usable coverage is sufficient — you do NOT need exhaustive or perfect coverage to answer. +- **When results are sufficient, compose the answer immediately. Do NOT call retrieval again to "double-check" or "get more context".** + +### Insufficient — retry +- Empty, `Error:`, off-topic, missing core entity/scope, no usable evidence at all, or claims required by the answer are not explicitly supported. ## 7. Fallback and Sequential Retry -On insufficient results, follow this sequence: +On insufficient results, you may retry **up to 2 more times** (3 calls total): -1. Rewrite query, retry same tool (once) -2. Switch to next retrieval source in default order -3. For `rag_retrieve`, expand `top_k`: `30 → 50 → 100` -4. `table_rag_retrieve` insufficient → try `rag_retrieve`; `rag_retrieve` insufficient → try `table_rag_retrieve` +1. Rewrite query, retry same tool. +2. For `rag_retrieve`, escalate `top_k` to `100` on retry. -- `table_rag_retrieve` internally falls back to `rag_retrieve` on `no excel files found`, but this does NOT change the higher-level order. -- Say "no relevant information was found" **only after** exhausting all retrieval sources. +- Say "no relevant information was found" **only after** exhausting all retries. - Do NOT switch to local filesystem inspection at any point. - Do NOT switch to model self-knowledge at any point. +- Do NOT call any retrieval tool more than 3 times in total. ## 8. Handling Missing or Partial Evidence @@ -79,13 +85,7 @@ On insufficient results, follow this sequence: - Prefer "the retrieved materials do not provide this information" over speculative completion. - When user asks for a definitive answer but evidence is incomplete, state the limitation directly. -## 9. Table RAG Result Handling - -- Follow all `[INSTRUCTION]` and `[EXTRA_INSTRUCTION]` in results. -- If truncated: tell user total (`N+M`), displayed (`N`), omitted (`M`). -- Cite sources using filenames from `file_ref_table`. - -## 10. Image Handling +## 9. Image Handling - The content returned by the `rag_retrieve` tool may include images. - Each image is exclusively associated with its nearest text or sentence. @@ -94,14 +94,7 @@ On insufficient results, follow this sequence: - Each sentence or key point in the response should be accompanied by relevant images when they meet the established association criteria. - Avoid placing all images at the end of the response. -## 11. Citation Requirements - -- MUST generate `` tags when using retrieval results. -- Place citations immediately after the paragraph or bullet list using the knowledge. Do NOT collect at end. -- 1-2 citations per paragraph/bullet. At least 1 citation when using retrieved knowledge. -- Do NOT cite claims that were not supported by retrieval. - -## 12. Self-Knowledge Prohibition +## 10. Self-Knowledge Prohibition This section applies whenever self-knowledge is disabled or forbidden for the current task. @@ -111,19 +104,18 @@ This section applies whenever self-knowledge is disabled or forbidden for the cu - The model must not supplement missing parts with general knowledge, conceptual explanation, common background, intuition, or likely completion. - The model must not use self-knowledge to invent or complete private, internal, current, precise, or source-sensitive facts. - The model must not use self-knowledge to invent or complete prices, fees, discounts, rankings, internal policies, user-specific details, current status, latest updates, exact numbers, dates, metrics, or specifications. -- Retrieved facts must include citations. - Unsupported parts must be stated as unavailable rather than guessed. - If a paragraph would mix retrieved facts and unsupported completion, remove the unsupported completion. - If evidence is incomplete, state the limitation explicitly. -## 13. Pre-Reply Self-Check +## 11. Pre-Reply Self-Check Before replying to a knowledge retrieval task, verify: - Used only whitelisted retrieval tools — no local filesystem inspection? +- Called retrieval at most 3 times total (not more)? +- Answered immediately when results were sufficient (did NOT call again unnecessarily)? - Did retrieval happen before any factual answer drafting? - Did every factual claim come from retrieved evidence rather than model knowledge? -- Exhausted retrieval flow before concluding "not found"? -- Citations placed immediately after each relevant paragraph? - If any unsupported part remained, was it removed or explicitly marked unavailable? If any answer is "no", correct the process first. diff --git a/skills/autoload/support/rag-retrieve/hooks/retrieval-policy-forbidden-self-knowledge.md b/skills/autoload/support/rag-retrieve/hooks/retrieval-policy-forbidden-self-knowledge.md index b6c1296..0bdd335 100644 --- a/skills/autoload/support/rag-retrieve/hooks/retrieval-policy-forbidden-self-knowledge.md +++ b/skills/autoload/support/rag-retrieve/hooks/retrieval-policy-forbidden-self-knowledge.md @@ -50,27 +50,38 @@ Execute **sequentially, one at a time**. Do NOT run in parallel. Do NOT probe fi ## 5. Retrieval Breadth (`top_k`) -- Apply `top_k` only to `rag_retrieve`. Use smallest sufficient value, expand if insufficient. -- `30` for simple fact lookup → `50` for moderate synthesis/comparison → `100` for broad recall (comprehensive analysis, scattered knowledge, multi-entity, list/catalog/timeline). -- Expansion order: `30 → 50 → 100`. If unsure, use `100`. +- Apply `top_k` only to `rag_retrieve`. Choose the appropriate value upfront to maximize first-call success. +- Use `50` for simple fact lookup or moderate synthesis, comparison, summarization, disambiguation. +- Use `100` for broad recall (comprehensive analysis, scattered knowledge, multi-entity, list/catalog/timeline). +- If unsure, use `50`. Only escalate to `100` on the retry call if first results are insufficient. ## 6. Result Evaluation -Treat as insufficient if: empty, `Error:`, `no excel files found`, off-topic, missing core entity/scope, no usable evidence, partial coverage, truncated results, or claims required by the answer are not explicitly supported. +**Maximum 3 retrieval calls per question.** After each call, evaluate immediately: + +### Sufficient — answer now +- The core entity/topic in the user's question has been hit. +- There is direct evidence supporting the main intent of the question. +- Partial but usable coverage is sufficient — you do NOT need exhaustive or perfect coverage to answer. +- **When results are sufficient, compose the answer immediately. Do NOT call retrieval again to "double-check" or "get more context".** + +### Insufficient — retry +- Empty, `Error:`, `no excel files found`, off-topic, missing core entity/scope, no usable evidence at all, or claims required by the answer are not explicitly supported. ## 7. Fallback and Sequential Retry -On insufficient results, follow this sequence: +On insufficient results, you may retry **up to 2 more times** (3 calls total): -1. Rewrite query, retry same tool (once) -2. Switch to next retrieval source in default order -3. For `rag_retrieve`, expand `top_k`: `30 → 50 → 100` -4. `table_rag_retrieve` insufficient → try `rag_retrieve`; `rag_retrieve` insufficient → try `table_rag_retrieve` +1. Rewrite query, retry same tool. +2. Switch to next retrieval source in default order. +3. For `rag_retrieve`, escalate `top_k` to `100` on retry. +4. `table_rag_retrieve` insufficient → try `rag_retrieve`; `rag_retrieve` insufficient → try `table_rag_retrieve`. - `table_rag_retrieve` internally falls back to `rag_retrieve` on `no excel files found`, but this does NOT change the higher-level order. -- Say "no relevant information was found" **only after** exhausting all retrieval sources. +- Say "no relevant information was found" **only after** exhausting all retries. - Do NOT switch to local filesystem inspection at any point. - Do NOT switch to model self-knowledge at any point. +- Do NOT call any retrieval tool more than 3 times in total. ## 8. Handling Missing or Partial Evidence @@ -83,7 +94,6 @@ On insufficient results, follow this sequence: - Follow all `[INSTRUCTION]` and `[EXTRA_INSTRUCTION]` in results. - If truncated: tell user total (`N+M`), displayed (`N`), omitted (`M`). -- Cite sources using filenames from `file_ref_table`. ## 10. Image Handling @@ -94,14 +104,7 @@ On insufficient results, follow this sequence: - Each sentence or key point in the response should be accompanied by relevant images when they meet the established association criteria. - Avoid placing all images at the end of the response. -## 11. Citation Requirements - -- MUST generate `` tags when using retrieval results. -- Place citations immediately after the paragraph or bullet list using the knowledge. Do NOT collect at end. -- 1-2 citations per paragraph/bullet. At least 1 citation when using retrieved knowledge. -- Do NOT cite claims that were not supported by retrieval. - -## 12. Self-Knowledge Prohibition +## 11. Self-Knowledge Prohibition This section applies whenever self-knowledge is disabled or forbidden for the current task. @@ -111,19 +114,18 @@ This section applies whenever self-knowledge is disabled or forbidden for the cu - The model must not supplement missing parts with general knowledge, conceptual explanation, common background, intuition, or likely completion. - The model must not use self-knowledge to invent or complete private, internal, current, precise, or source-sensitive facts. - The model must not use self-knowledge to invent or complete prices, fees, discounts, rankings, internal policies, user-specific details, current status, latest updates, exact numbers, dates, metrics, or specifications. -- Retrieved facts must include citations. - Unsupported parts must be stated as unavailable rather than guessed. - If a paragraph would mix retrieved facts and unsupported completion, remove the unsupported completion. - If evidence is incomplete, state the limitation explicitly. -## 13. Pre-Reply Self-Check +## 12. Pre-Reply Self-Check Before replying to a knowledge retrieval task, verify: - Used only whitelisted retrieval tools — no local filesystem inspection? +- Called retrieval at most 3 times total (not more)? +- Answered immediately when results were sufficient (did NOT call again unnecessarily)? - Did retrieval happen before any factual answer drafting? - Did every factual claim come from retrieved evidence rather than model knowledge? -- Exhausted retrieval flow before concluding "not found"? -- Citations placed immediately after each relevant paragraph? - If any unsupported part remained, was it removed or explicitly marked unavailable? If any answer is "no", correct the process first. diff --git a/skills/developing/rag-retrieve-no-citation/hooks/retrieval-policy-forbidden-self-knowledge.md b/skills/developing/rag-retrieve-no-citation/hooks/retrieval-policy-forbidden-self-knowledge.md index 2e4d133..eb9139b 100644 --- a/skills/developing/rag-retrieve-no-citation/hooks/retrieval-policy-forbidden-self-knowledge.md +++ b/skills/developing/rag-retrieve-no-citation/hooks/retrieval-policy-forbidden-self-knowledge.md @@ -37,8 +37,8 @@ Execute **sequentially, one at a time**. Do NOT run in parallel. Do NOT probe fi 1. **Skill-enabled retrieval tools** (use first when available) 2. **`rag_retrieve`** -- After each step, evaluate sufficiency before proceeding. - Retrieval must happen **before** any factual answer generation. +- After each step, evaluate sufficiency before proceeding. ## 4. Query Preparation @@ -48,25 +48,35 @@ Execute **sequentially, one at a time**. Do NOT run in parallel. Do NOT probe fi ## 5. Retrieval Breadth (`top_k`) -- Apply `top_k` only to `rag_retrieve`. Use smallest sufficient value, expand if insufficient. -- `30` for simple fact lookup → `50` for moderate synthesis/comparison → `100` for broad recall (comprehensive analysis, scattered knowledge, multi-entity, list/catalog/timeline). -- Expansion order: `30 → 50 → 100`. If unsure, use `100`. +- Apply `top_k` only to `rag_retrieve`. Choose the appropriate value upfront to maximize first-call success. +- Use `50` for simple fact lookup or moderate synthesis, comparison, summarization, disambiguation. +- Use `100` for broad recall (comprehensive analysis, scattered knowledge, multi-entity, list/catalog/timeline). +- If unsure, use `50`. Only escalate to `100` on the retry call if first results are insufficient. ## 6. Result Evaluation -Treat as insufficient if: empty, `Error:`, off-topic, missing core entity/scope, no usable evidence, partial coverage, truncated results, or claims required by the answer are not explicitly supported. +**Maximum 3 retrieval calls per question.** After each call, evaluate immediately: + +### Sufficient — answer now +- The core entity/topic in the user's question has been hit. +- There is direct evidence supporting the main intent of the question. +- Partial but usable coverage is sufficient — you do NOT need exhaustive or perfect coverage to answer. +- **When results are sufficient, compose the answer immediately. Do NOT call retrieval again to "double-check" or "get more context".** + +### Insufficient — retry +- Empty, `Error:`, off-topic, missing core entity/scope, no usable evidence at all, or claims required by the answer are not explicitly supported. ## 7. Fallback and Sequential Retry -On insufficient results, follow this sequence: +On insufficient results, you may retry **up to 2 more times** (3 calls total): -1. Rewrite query, retry same tool (once) -2. Switch to next retrieval source in default order -3. For `rag_retrieve`, expand `top_k`: `30 → 50 → 100` +1. Rewrite query, retry same tool. +2. For `rag_retrieve`, escalate `top_k` to `100` on retry. -- Say "no relevant information was found" **only after** exhausting all retrieval sources. +- Say "no relevant information was found" **only after** exhausting all retries. - Do NOT switch to local filesystem inspection at any point. - Do NOT switch to model self-knowledge at any point. +- Do NOT call any retrieval tool more than 3 times in total. ## 8. Handling Missing or Partial Evidence @@ -84,7 +94,6 @@ On insufficient results, follow this sequence: - Each sentence or key point in the response should be accompanied by relevant images when they meet the established association criteria. - Avoid placing all images at the end of the response. - ## 10. Self-Knowledge Prohibition This section applies whenever self-knowledge is disabled or forbidden for the current task. @@ -103,9 +112,10 @@ This section applies whenever self-knowledge is disabled or forbidden for the cu Before replying to a knowledge retrieval task, verify: - Used only whitelisted retrieval tools — no local filesystem inspection? +- Called retrieval at most 3 times total (not more)? +- Answered immediately when results were sufficient (did NOT call again unnecessarily)? - Did retrieval happen before any factual answer drafting? - Did every factual claim come from retrieved evidence rather than model knowledge? -- Exhausted retrieval flow before concluding "not found"? - If any unsupported part remained, was it removed or explicitly marked unavailable? If any answer is "no", correct the process first. diff --git a/skills/developing/rag-retrieve-no-citation/hooks/retrieval-policy.md b/skills/developing/rag-retrieve-no-citation/hooks/retrieval-policy.md index ce14d2b..d475c60 100644 --- a/skills/developing/rag-retrieve-no-citation/hooks/retrieval-policy.md +++ b/skills/developing/rag-retrieve-no-citation/hooks/retrieval-policy.md @@ -35,24 +35,34 @@ Execute **sequentially, one at a time**. Do NOT run in parallel. Do NOT probe fi ## 4. Retrieval Breadth (`top_k`) -- Apply `top_k` only to `rag_retrieve`. Use smallest sufficient value, expand if insufficient. -- `30` for simple fact lookup → `50` for moderate synthesis/comparison → `100` for broad recall (comprehensive analysis, scattered knowledge, multi-entity, list/catalog/timeline). -- Expansion order: `30 → 50 → 100`. If unsure, use `100`. +- Apply `top_k` only to `rag_retrieve`. Choose the appropriate value upfront to maximize first-call success. +- Use `50` for simple fact lookup or moderate synthesis, comparison, summarization, disambiguation. +- Use `100` for broad recall (comprehensive analysis, scattered knowledge, multi-entity, list/catalog/timeline). +- If unsure, use `50`. Only escalate to `100` on the retry call if first results are insufficient. ## 5. Result Evaluation -Treat as insufficient if: empty, `Error:`, off-topic, missing core entity/scope, no usable evidence, partial coverage, or truncated results. +**Maximum 3 retrieval calls per question.** After each call, evaluate immediately: + +### Sufficient — answer now +- The core entity/topic in the user's question has been hit. +- There is direct evidence supporting the main intent of the question. +- Partial but usable coverage is sufficient — you do NOT need exhaustive or perfect coverage to answer. +- **When results are sufficient, compose the answer immediately. Do NOT call retrieval again to "double-check" or "get more context".** + +### Insufficient — retry +- Empty, `Error:`, off-topic, missing core entity/scope, no usable evidence at all. ## 6. Fallback and Sequential Retry -On insufficient results, follow this sequence: +On insufficient results, you may retry **up to 2 more times** (3 calls total): -1. Rewrite query, retry same tool (once) -2. Switch to next retrieval source in default order -3. For `rag_retrieve`, expand `top_k`: `30 → 50 → 100` +1. Rewrite query, retry same tool. +2. For `rag_retrieve`, escalate `top_k` to `100` on retry. -- Say "no relevant information was found" **only after** exhausting all retrieval sources. +- Say "no relevant information was found" **only after** exhausting all retries. - Do NOT switch to local filesystem inspection at any point. +- Do NOT call any retrieval tool more than 3 times in total. ## 7. Image Handling @@ -81,7 +91,8 @@ This section applies only when self-knowledge is enabled. Before replying to a knowledge retrieval task, verify: - Used only whitelisted retrieval tools — no local filesystem inspection? -- Exhausted retrieval flow before concluding "not found"? +- Called retrieval at most 3 times total (not more)? +- Answered immediately when results were sufficient (did NOT call again unnecessarily)? - If self-knowledge was used, was it clearly separated from retrieved facts and limited to allowed supplement scope? If any answer is "no", correct the process first. diff --git a/skills/onprem/rag-retrieve-only/hooks/retrieval-policy-forbidden-self-knowledge.md b/skills/onprem/rag-retrieve-only/hooks/retrieval-policy-forbidden-self-knowledge.md index 2b68869..2dfa82e 100644 --- a/skills/onprem/rag-retrieve-only/hooks/retrieval-policy-forbidden-self-knowledge.md +++ b/skills/onprem/rag-retrieve-only/hooks/retrieval-policy-forbidden-self-knowledge.md @@ -48,25 +48,35 @@ Execute **sequentially, one at a time**. Do NOT run in parallel. Do NOT probe fi ## 5. Retrieval Breadth (`top_k`) -- Apply `top_k` only to `rag_retrieve`. Use smallest sufficient value, expand if insufficient. -- `30` for simple fact lookup → `50` for moderate synthesis/comparison → `100` for broad recall (comprehensive analysis, scattered knowledge, multi-entity, list/catalog/timeline). -- Expansion order: `30 → 50 → 100`. If unsure, use `100`. +- Apply `top_k` only to `rag_retrieve`. Choose the appropriate value upfront to maximize first-call success. +- Use `50` for simple fact lookup or moderate synthesis, comparison, summarization, disambiguation. +- Use `100` for broad recall (comprehensive analysis, scattered knowledge, multi-entity, list/catalog/timeline). +- If unsure, use `50`. Only escalate to `100` on the retry call if first results are insufficient. ## 6. Result Evaluation -Treat as insufficient if: empty, `Error:`, off-topic, missing core entity/scope, no usable evidence, partial coverage, truncated results, or claims required by the answer are not explicitly supported. +**Maximum 3 retrieval calls per question.** After each call, evaluate immediately: + +### Sufficient — answer now +- The core entity/topic in the user's question has been hit. +- There is direct evidence supporting the main intent of the question. +- Partial but usable coverage is sufficient — you do NOT need exhaustive or perfect coverage to answer. +- **When results are sufficient, compose the answer immediately. Do NOT call retrieval again to "double-check" or "get more context".** + +### Insufficient — retry +- Empty, `Error:`, off-topic, missing core entity/scope, no usable evidence at all, or claims required by the answer are not explicitly supported. ## 7. Fallback and Sequential Retry -On insufficient results, follow this sequence: +On insufficient results, you may retry **up to 2 more times** (3 calls total): -1. Rewrite query, retry same tool (once) -2. Switch to next retrieval source in default order -3. For `rag_retrieve`, expand `top_k`: `30 → 50 → 100` +1. Rewrite query, retry same tool. +2. For `rag_retrieve`, escalate `top_k` to `100` on retry. -- Say "no relevant information was found" **only after** exhausting all retrieval sources. +- Say "no relevant information was found" **only after** exhausting all retries. - Do NOT switch to local filesystem inspection at any point. - Do NOT switch to model self-knowledge at any point. +- Do NOT call any retrieval tool more than 3 times in total. ## 8. Handling Missing or Partial Evidence @@ -84,14 +94,7 @@ On insufficient results, follow this sequence: - Each sentence or key point in the response should be accompanied by relevant images when they meet the established association criteria. - Avoid placing all images at the end of the response. -## 10. Citation Requirements - -- MUST generate `` tags when using retrieval results. -- Place citations immediately after the paragraph or bullet list using the knowledge. Do NOT collect at end. -- 1-2 citations per paragraph/bullet. At least 1 citation when using retrieved knowledge. -- Do NOT cite claims that were not supported by retrieval. - -## 11. Self-Knowledge Prohibition +## 10. Self-Knowledge Prohibition This section applies whenever self-knowledge is disabled or forbidden for the current task. @@ -101,19 +104,18 @@ This section applies whenever self-knowledge is disabled or forbidden for the cu - The model must not supplement missing parts with general knowledge, conceptual explanation, common background, intuition, or likely completion. - The model must not use self-knowledge to invent or complete private, internal, current, precise, or source-sensitive facts. - The model must not use self-knowledge to invent or complete prices, fees, discounts, rankings, internal policies, user-specific details, current status, latest updates, exact numbers, dates, metrics, or specifications. -- Retrieved facts must include citations. - Unsupported parts must be stated as unavailable rather than guessed. - If a paragraph would mix retrieved facts and unsupported completion, remove the unsupported completion. - If evidence is incomplete, state the limitation explicitly. -## 12. Pre-Reply Self-Check +## 11. Pre-Reply Self-Check Before replying to a knowledge retrieval task, verify: - Used only whitelisted retrieval tools — no local filesystem inspection? +- Called retrieval at most 3 times total (not more)? +- Answered immediately when results were sufficient (did NOT call again unnecessarily)? - Did retrieval happen before any factual answer drafting? - Did every factual claim come from retrieved evidence rather than model knowledge? -- Exhausted retrieval flow before concluding "not found"? -- Citations placed immediately after each relevant paragraph? - If any unsupported part remained, was it removed or explicitly marked unavailable? If any answer is "no", correct the process first. diff --git a/skills/support/rag-retrieve-only/hooks/retrieval-policy-forbidden-self-knowledge.md b/skills/support/rag-retrieve-only/hooks/retrieval-policy-forbidden-self-knowledge.md index 2b68869..0bdd335 100644 --- a/skills/support/rag-retrieve-only/hooks/retrieval-policy-forbidden-self-knowledge.md +++ b/skills/support/rag-retrieve-only/hooks/retrieval-policy-forbidden-self-knowledge.md @@ -14,7 +14,7 @@ For knowledge retrieval tasks, **this policy overrides generic codebase explorat - **Prohibited answer source**: the model's own parametric knowledge, memory, prior world knowledge, intuition, common sense completion, or unsupported inference. - **Prohibited tools**: `Glob`, `Read`, `LS`, Bash (`ls`, `find`, `cat`, `head`, `tail`, `grep`, etc.) — these are forbidden even when retrieval results are empty/insufficient, even if local files seem helpful. -- **Allowed tools only**: skill-enabled retrieval tools, `rag_retrieve`. No other source for factual answering. +- **Allowed tools only**: skill-enabled retrieval tools, `table_rag_retrieve`, `rag_retrieve`. No other source for factual answering. - Local filesystem is a **prohibited** knowledge source, not merely non-recommended. - Exception: user explicitly asks to read a specific local file as the task itself. - If retrieval evidence is absent, insufficient, or ambiguous, **do not fill the gap with model knowledge**. @@ -35,7 +35,9 @@ For any knowledge retrieval task: Execute **sequentially, one at a time**. Do NOT run in parallel. Do NOT probe filesystem first. 1. **Skill-enabled retrieval tools** (use first when available) -2. **`rag_retrieve`** +2. **`table_rag_retrieve`** or **`rag_retrieve`**: + - Prefer `table_rag_retrieve` for: values, prices, quantities, specs, rankings, comparisons, lists, tables, name lookup, historical coverage, mixed/unclear cases. + - Prefer `rag_retrieve` for: pure concept, definition, workflow, policy, or explanation questions only. - After each step, evaluate sufficiency before proceeding. - Retrieval must happen **before** any factual answer generation. @@ -48,25 +50,38 @@ Execute **sequentially, one at a time**. Do NOT run in parallel. Do NOT probe fi ## 5. Retrieval Breadth (`top_k`) -- Apply `top_k` only to `rag_retrieve`. Use smallest sufficient value, expand if insufficient. -- `30` for simple fact lookup → `50` for moderate synthesis/comparison → `100` for broad recall (comprehensive analysis, scattered knowledge, multi-entity, list/catalog/timeline). -- Expansion order: `30 → 50 → 100`. If unsure, use `100`. +- Apply `top_k` only to `rag_retrieve`. Choose the appropriate value upfront to maximize first-call success. +- Use `50` for simple fact lookup or moderate synthesis, comparison, summarization, disambiguation. +- Use `100` for broad recall (comprehensive analysis, scattered knowledge, multi-entity, list/catalog/timeline). +- If unsure, use `50`. Only escalate to `100` on the retry call if first results are insufficient. ## 6. Result Evaluation -Treat as insufficient if: empty, `Error:`, off-topic, missing core entity/scope, no usable evidence, partial coverage, truncated results, or claims required by the answer are not explicitly supported. +**Maximum 3 retrieval calls per question.** After each call, evaluate immediately: + +### Sufficient — answer now +- The core entity/topic in the user's question has been hit. +- There is direct evidence supporting the main intent of the question. +- Partial but usable coverage is sufficient — you do NOT need exhaustive or perfect coverage to answer. +- **When results are sufficient, compose the answer immediately. Do NOT call retrieval again to "double-check" or "get more context".** + +### Insufficient — retry +- Empty, `Error:`, `no excel files found`, off-topic, missing core entity/scope, no usable evidence at all, or claims required by the answer are not explicitly supported. ## 7. Fallback and Sequential Retry -On insufficient results, follow this sequence: +On insufficient results, you may retry **up to 2 more times** (3 calls total): -1. Rewrite query, retry same tool (once) -2. Switch to next retrieval source in default order -3. For `rag_retrieve`, expand `top_k`: `30 → 50 → 100` +1. Rewrite query, retry same tool. +2. Switch to next retrieval source in default order. +3. For `rag_retrieve`, escalate `top_k` to `100` on retry. +4. `table_rag_retrieve` insufficient → try `rag_retrieve`; `rag_retrieve` insufficient → try `table_rag_retrieve`. -- Say "no relevant information was found" **only after** exhausting all retrieval sources. +- `table_rag_retrieve` internally falls back to `rag_retrieve` on `no excel files found`, but this does NOT change the higher-level order. +- Say "no relevant information was found" **only after** exhausting all retries. - Do NOT switch to local filesystem inspection at any point. - Do NOT switch to model self-knowledge at any point. +- Do NOT call any retrieval tool more than 3 times in total. ## 8. Handling Missing or Partial Evidence @@ -75,7 +90,12 @@ On insufficient results, follow this sequence: - Prefer "the retrieved materials do not provide this information" over speculative completion. - When user asks for a definitive answer but evidence is incomplete, state the limitation directly. -## 9. Image Handling +## 9. Table RAG Result Handling + +- Follow all `[INSTRUCTION]` and `[EXTRA_INSTRUCTION]` in results. +- If truncated: tell user total (`N+M`), displayed (`N`), omitted (`M`). + +## 10. Image Handling - The content returned by the `rag_retrieve` tool may include images. - Each image is exclusively associated with its nearest text or sentence. @@ -84,13 +104,6 @@ On insufficient results, follow this sequence: - Each sentence or key point in the response should be accompanied by relevant images when they meet the established association criteria. - Avoid placing all images at the end of the response. -## 10. Citation Requirements - -- MUST generate `` tags when using retrieval results. -- Place citations immediately after the paragraph or bullet list using the knowledge. Do NOT collect at end. -- 1-2 citations per paragraph/bullet. At least 1 citation when using retrieved knowledge. -- Do NOT cite claims that were not supported by retrieval. - ## 11. Self-Knowledge Prohibition This section applies whenever self-knowledge is disabled or forbidden for the current task. @@ -101,7 +114,6 @@ This section applies whenever self-knowledge is disabled or forbidden for the cu - The model must not supplement missing parts with general knowledge, conceptual explanation, common background, intuition, or likely completion. - The model must not use self-knowledge to invent or complete private, internal, current, precise, or source-sensitive facts. - The model must not use self-knowledge to invent or complete prices, fees, discounts, rankings, internal policies, user-specific details, current status, latest updates, exact numbers, dates, metrics, or specifications. -- Retrieved facts must include citations. - Unsupported parts must be stated as unavailable rather than guessed. - If a paragraph would mix retrieved facts and unsupported completion, remove the unsupported completion. - If evidence is incomplete, state the limitation explicitly. @@ -110,10 +122,10 @@ This section applies whenever self-knowledge is disabled or forbidden for the cu Before replying to a knowledge retrieval task, verify: - Used only whitelisted retrieval tools — no local filesystem inspection? +- Called retrieval at most 3 times total (not more)? +- Answered immediately when results were sufficient (did NOT call again unnecessarily)? - Did retrieval happen before any factual answer drafting? - Did every factual claim come from retrieved evidence rather than model knowledge? -- Exhausted retrieval flow before concluding "not found"? -- Citations placed immediately after each relevant paragraph? - If any unsupported part remained, was it removed or explicitly marked unavailable? If any answer is "no", correct the process first.