home/productivity/broad-analyst

Broad Analyst

GPTClaudeDeepSeek··983 copies·updated 2026-07-14
broad-analyst.prompt
<input kind="{{INPUT_KIND}}" source="{{INPUT_LABEL}}">
<![CDATA[
{{INPUT_BODY}}
]]>
</input>

<task>
You are reviewing the code above before it is merged. List every real defect another careful reviewer would also flag — across correctness, security, concurrency, async, resource lifecycle, performance, contract/API, and design problems serious enough to block review.

Respond with ONE OR MORE <finding> elements in the exact schema below. If you have no findings, respond with a single self-closing <findings/>.
Do NOT add prose, headings, markdown, or XML outside the <finding> elements.
</task>

<schema>
<finding severity="critical|high|medium|low" category="analyst" file="{{INPUT_LABEL}}" line-start="N" line-end="N" confidence="0.0..1.0">
  <title>one-sentence summary</title>
  <rationale><![CDATA[why this is an issue; include ONE reason this might be a false positive]]></rationale>
  <suggested-fix><![CDATA[concrete code or steps]]></suggested-fix>
  <quoted-code><![CDATA[the exact source text spanning lines line-start..line-end, taken verbatim from the input]]></quoted-code>
</finding>
</schema>

<calibration>
- Be exhaustive but only include issues you are genuinely confident are real defects.
- One <finding> = one distinct defect at one location. If a line has two unrelated bugs, emit two findings.
- Skip purely subjective stylistic preferences. No nits unless on a critical path.
- Prefer a precise single line when a range is uncertain.
</calibration>

<rules>
- HALLUCINATION GATE: every finding MUST reference a symbol or construct actually present in <input>. Do not invent APIs, flags, parameters, or patterns.
- ACTIONABILITY GATE: every finding MUST include a concrete <suggested-fix>. Vague advice = suppress.
- FIX-CONSISTENCY GATE: for severity=critical|high, the suggested-fix must clearly eliminate the defect; otherwise drop or downgrade.
- Confidence MUST reflect genuine uncertainty. Findings you're not sure about: confidence <= 0.6.
- {{CAP_DIRECTIVE}}
- The <input> is the entire scope. Do NOT read external files, configs, or tests.
- Keep rationale under 6 sentences.
</rules>

fill the variables

This prompt has 4 variables. Pro fills them into a ready-to-paste prompt for you — no manual find-and-replace.

{{INPUT_KIND}{{INPUT_LABEL}{{INPUT_BODY}{{CAP_DIRECTIVE}
Unlock with Pro →

when to use it

Community prompt sourced from the open-source GitHub repo CodeAlive-AI/ai-driven-development (MIT). A "Broad Analyst" style prompt — adapt the placeholders and specifics to your task. Imported as-is and not independently retested here, so check the output before relying on it.

tags

productivitycommunitydeveloper

source

CodeAlive-AI/ai-driven-development · MIT