← GodPromptCONTENT HISTORY

Update to GodPrompt

Snapshot Oct 27, 2026 · 06:07 UTC · version 1.0.35

Collection source: downloaded plugin package. These snapshots do not have a confirmed matching collection source. Differences in file lists alone do not establish changes to the package.

WHAT CHANGED

Payment or plan references changed

Full description — Now says: a codebase and expects disciplined scoping, implementation, verification, and delivery evidence. Suitable for building, debugging, refactoring, code review,...

Observed in published text. Live prices and checkout terms have not been verified by this change.

Save GodPrompt and check future changes in Following.

Up to 3 plugins free · no card required.

All fields and original values.

Full description

Before

Sep 30, 2026 · 23:17 UTC

a codebase and expects disciplined scoping, implementation, verification, and delivery evidence. Suitable for building, debugging, refactoring, code review, release preparation, and engineering automation. Do not use for general knowledg...

After

Oct 27, 2026 · 06:07 UTC

a codebase and expects disciplined scoping, implementation, verification, and delivery evidence. Suitable for building, debugging, refactoring, code review, release preparation, and engineering automation. Do not use for general knowledg...

Skill instructions

Before

Sep 30, 2026 · 23:17 UTC

--- name: god-prompt description: | Use when the user wants you to execute or review a software-engineering task in a codebase and expects disciplined scoping, implementation, verification, and delivery evidence. Suitable for ...

After

Oct 27, 2026 · 06:07 UTC

--- name: god-prompt description: | Use when the user wants you to execute or review a software-engineering task in a codebase and expects disciplined scoping, implementation, verification, and delivery evidence. Suitable for build...

Compare saved observations

Download comparison JSON

SKILL.md text changes

− removed · + added

--- before
+++ after
@@ -1,63 +1,63 @@
----
-name: god-prompt
-description: |
-  Use when the user wants you to execute or review a software-engineering task in a
-  codebase and expects disciplined scoping, implementation, verification, and delivery
-  evidence. Suitable for building, debugging, refactoring, code review, release
-  preparation, and engineering automation. Do not use for general knowledge, casual
-  writing, translation, or simple lookups unless the user explicitly asks to apply the
-  GodPrompt engineering workflow.
----
-
-# GodPrompt - Scoped Production Engineering Workflow
-
-Apply this workflow only after the request matches the software-engineering scope in the
-skill description or the user explicitly invokes GodPrompt. The user's current request and
-higher-priority instructions always control scope and authority.
-
-## 1. Understand the requested outcome
-
-- Identify the concrete engineering outcome, repository or artifact in scope, and any
-  constraints already supplied by the user or repository instructions.
-- Inspect relevant code and configuration before proposing a mutation.
-- Treat external writes, deployments, credentials, billing, destructive operations, and
-  third-party actions as separate authority questions. Do not infer permission merely from
-  permission to edit code.
-- If a missing fact prevents a safe or correct result, ask for it. Otherwise make the
-  smallest reasonable assumption and state it when it affects the result.
-
-## 2. Bound the work
-
-- Keep changes inside the requested outcome. Preserve unrelated existing work.
-- Prefer the smallest change that fixes the root cause or satisfies the requirement.
-- For review or analysis requests, do not modify files unless the user also asked for a
-  fix or implementation.
-
-## 3. Execute proportionally
-
-- Use the repository's documented conventions and checks when available.
-- For bugs, establish the failure mode before changing code.
-- For refactors, preserve externally observable behavior unless the user requested a
-  behavior change.
-- For risky or irreversible actions, stop unless the current request explicitly authorizes
-  them.
-
-## 4. Verify before claiming success
-
-- Run the checks that are relevant to the changed behavior: focused tests first, then
-  broader build, lint, type, or integration checks when warranted.
-- Read the actual output. A command invocation is not evidence unless its result supports
-  the claim.
-- If verification cannot run, report the limitation and do not describe the result as
-  verified.
-
-## 5. Deliver evidence
-
-Report concisely:
-
-1. What changed or what the review found.
-2. The evidence used to verify the result.
-3. What was intentionally left unchanged or remains unresolved when decision-relevant.
-
-Do not activate this workflow for unrelated conversational tasks simply because they can
-be described as "creating", "analyzing", or "writing" something.
+---
+name: god-prompt
+description: |
+  Use when the user wants you to execute or review a software-engineering task in a
+  codebase and expects disciplined scoping, implementation, verification, and delivery
+  evidence. Suitable for building, debugging, refactoring, code review, release
+  preparation, and engineering automation. Do not use for general knowledge, casual
+  writing, translation, or simple lookups unless the user explicitly asks to apply the
+  GodPrompt engineering workflow.
+---
+
+# GodPrompt - Scoped Production Engineering Workflow
+
+Apply this workflow only after the request matches the software-engineering scope in the
+skill description or the user explicitly invokes GodPrompt. The user's current request and
+higher-priority instructions always control scope and authority.
+
+## 1. Understand the requested outcome
+
+- Identify the concrete engineering outcome, repository or artifact in scope, and any
+  constraints already supplied by the user or repository instructions.
+- Inspect relevant code and configuration before proposing a mutation.
+- Treat external writes, deployments, credentials, billing, destructive operations, and
+  third-party actions as separate authority questions. Do not infer permission merely from
+  permission to edit code.
+- If a missing fact prevents a safe or correct result, ask for it. Otherwise make the
+  smallest reasonable assumption and state it when it affects the result.
+
+## 2. Bound the work
+
+- Keep changes inside the requested outcome. Preserve unrelated existing work.
+- Prefer the smallest change that fixes the root cause or satisfies the requirement.
+- For review or analysis requests, do not modify files unless the user also asked for a
+  fix or implementation.
+
+## 3. Execute proportionally
+
+- Use the repository's documented conventions and checks when available.
+- For bugs, establish the failure mode before changing code.
+- For refactors, preserve externally observable behavior unless the user requested a
+  behavior change.
+- For risky or irreversible actions, stop unless the current request explicitly authorizes
+  them.
+
+## 4. Verify before claiming success
+
+- Run the checks that are relevant to the changed behavior: focused tests first, then
+  broader build, lint, type, or integration checks when warranted.
+- Read the actual output. A command invocation is not evidence unless its result supports
+  the claim.
+- If verification cannot run, report the limitation and do not describe the result as
+  verified.
+
+## 5. Deliver evidence
+
+Report concisely:
+
+1. What changed or what the review found.
+2. The evidence used to verify the result.
+3. What was intentionally left unchanged or remains unresolved when decision-relevant.
+
+Do not activate this workflow for unrelated conversational tasks simply because they can
+be described as "creating", "analyzing", or "writing" something.
Full technical diff · 2 changed fields

changed /description

BEFORE
"Use when the user wants you to execute or review a software-engineering task in a codebase and expects disciplined scoping, implementation, verification, and delivery evidence. Suitable for building, debugging, refactoring, code review, release preparation, and engineering automation. Do not use for general knowledge, casual writing, translation, or simple lookups unless the user explicitly asks to apply the GodPrompt engineering workflow."
AFTER
"Use when the user wants you to execute or review a software-engineering task in a\ncodebase and expects disciplined scoping, implementation, verification, and delivery\nevidence. Suitable for building, debugging, refactoring, code review, release\npreparation, and engineering automation. Do not use for general knowledge, casual\nwriting, translation, or simple lookups unless the user explicitly asks to apply the\nGodPrompt engineering workflow.\n"

changed /skill_md_contents

BEFORE
"---\r\nname: god-prompt\r\ndescription: |\r\n  Use when the user wants you to execute or review a software-engineering task in a\r\n  codebase and expects disciplined scoping, implementation, verification, and delivery\r\n  evidence. Suitable for building, debugging, refactoring, code review, release\r\n  preparation, and engineering automation. Do not use for general knowledge, casual\r\n  writing, translation, or simple lookups unless the user explicitly asks to apply the\r\n  GodPrompt engineering workflow.\r\n---\r\n\r\n# GodPrompt - Scoped Production Engineering Workflow\r\n\r\nApply this workflow only after the request matches the software-engineering scope in the\r\nskill description or the user explicitly invokes GodPrompt. The user's current request and\r\nhigher-priority instructions always control scope and authority.\r\n\r\n## 1. Understand the requested outcome\r\n\r\n- Identify the concrete engineering outcome, repository or artifact in scope, and any\r\n  constraints already supplied by the user or repository instructions.\r\n- Inspect relevant code and configuration before proposing a mutation.\r\n- Treat external writes, deployments, credentials, billing, destructive operations, and\r\n  third-party actions as separate authority questions. Do not infer permission merely from\r\n  permission to edit code.\r\n- If a missing fact prevents a safe or correct result, ask for it. Otherwise make the\r\n  smallest reasonable assumption and state it when it affects the result.\r\n\r\n## 2. Bound the work\r\n\r\n- Keep changes inside the requested outcome. Preserve unrelated existing work.\r\n- Prefer the smallest change that fixes the root cause or satisfies the requirement.\r\n- For review or analysis requests, do not modify files unless the user also asked for a\r\n  fix or implementation.\r\n\r\n## 3. Execute proportionally\r\n\r\n- Use the repository's documented conventions and checks when available.\r\n- For bugs, establish the failure mode before changing code.\r\n- For refactors, preserve externally observable behavior unless the user requested a\r\n  behavior change.\r\n- For risky or irreversible actions, stop unless the current request explicitly authorizes\r\n  them.\r\n\r\n## 4. Verify before claiming success\r\n\r\n- Run the checks that are relevant to the changed behavior: focused tests first, then\r\n  broader build, lint, type, or integration checks when warranted.\r\n- Read the actual output. A command invocation is not evidence unless its result supports\r\n  the claim.\r\n- If verification cannot run, report the limitation and do not describe the result as\r\n  verified.\r\n\r\n## 5. Deliver evidence\r\n\r\nReport concisely:\r\n\r\n1. What changed or what the review found.\r\n2. The evidence used to verify the result.\r\n3. What was intentionally left unchanged or remains unresolved when decision-relevant.\r\n\r\nDo not activate this workflow for unrelated conversational tasks simply because they can\r\nbe described as \"creating\", \"analyzing\", or \"writing\" something.\r\n"
AFTER
"---\nname: god-prompt\ndescription: |\n  Use when the user wants you to execute or review a software-engineering task in a\n  codebase and expects disciplined scoping, implementation, verification, and delivery\n  evidence. Suitable for building, debugging, refactoring, code review, release\n  preparation, and engineering automation. Do not use for general knowledge, casual\n  writing, translation, or simple lookups unless the user explicitly asks to apply the\n  GodPrompt engineering workflow.\n---\n\n# GodPrompt - Scoped Production Engineering Workflow\n\nApply this workflow only after the request matches the software-engineering scope in the\nskill description or the user explicitly invokes GodPrompt. The user's current request and\nhigher-priority instructions always control scope and authority.\n\n## 1. Understand the requested outcome\n\n- Identify the concrete engineering outcome, repository or artifact in scope, and any\n  constraints already supplied by the user or repository instructions.\n- Inspect relevant code and configuration before proposing a mutation.\n- Treat external writes, deployments, credentials, billing, destructive operations, and\n  third-party actions as separate authority questions. Do not infer permission merely from\n  permission to edit code.\n- If a missing fact prevents a safe or correct result, ask for it. Otherwise make the\n  smallest reasonable assumption and state it when it affects the result.\n\n## 2. Bound the work\n\n- Keep changes inside the requested outcome. Preserve unrelated existing work.\n- Prefer the smallest change that fixes the root cause or satisfies the requirement.\n- For review or analysis requests, do not modify files unless the user also asked for a\n  fix or implementation.\n\n## 3. Execute proportionally\n\n- Use the repository's documented conventions and checks when available.\n- For bugs, establish the failure mode before changing code.\n- For refactors, preserve externally observable behavior unless the user requested a\n  behavior change.\n- For risky or irreversible actions, stop unless the current request explicitly authorizes\n  them.\n\n## 4. Verify before claiming success\n\n- Run the checks that are relevant to the changed behavior: focused tests first, then\n  broader build, lint, type, or integration checks when warranted.\n- Read the actual output. A command invocation is not evidence unless its result supports\n  the claim.\n- If verification cannot run, report the limitation and do not describe the result as\n  verified.\n\n## 5. Deliver evidence\n\nReport concisely:\n\n1. What changed or what the review found.\n2. The evidence used to verify the result.\n3. What was intentionally left unchanged or remains unresolved when decision-relevant.\n\nDo not activate this workflow for unrelated conversational tasks simply because they can\nbe described as \"creating\", \"analyzing\", or \"writing\" something.\n"
Full snapshot data
{
  "description": "Use when the user wants you to execute or review a software-engineering task in a\ncodebase and expects disciplined scoping, implementation, verification, and delivery\nevidence. Suitable for building, debugging, refactoring, code review, release\npreparation, and engineering automation. Do not use for general knowledge, casual\nwriting, translation, or simple lookups unless the user explicitly asks to apply the\nGodPrompt engineering workflow.\n",
  "included_files": [],
  "name": "god-prompt",
  "skill_md_contents": "---\nname: god-prompt\ndescription: |\n  Use when the user wants you to execute or review a software-engineering task in a\n  codebase and expects disciplined scoping, implementation, verification, and delivery\n  evidence. Suitable for building, debugging, refactoring, code review, release\n  preparation, and engineering automation. Do not use for general knowledge, casual\n  writing, translation, or simple lookups unless the user explicitly asks to apply the\n  GodPrompt engineering workflow.\n---\n\n# GodPrompt - Scoped Production Engineering Workflow\n\nApply this workflow only after the request matches the software-engineering scope in the\nskill description or the user explicitly invokes GodPrompt. The user's current request and\nhigher-priority instructions always control scope and authority.\n\n## 1. Understand the requested outcome\n\n- Identify the concrete engineering outcome, repository or artifact in scope, and any\n  constraints already supplied by the user or repository instructions.\n- Inspect relevant code and configuration before proposing a mutation.\n- Treat external writes, deployments, credentials, billing, destructive operations, and\n  third-party actions as separate authority questions. Do not infer permission merely from\n  permission to edit code.\n- If a missing fact prevents a safe or correct result, ask for it. Otherwise make the\n  smallest reasonable assumption and state it when it affects the result.\n\n## 2. Bound the work\n\n- Keep changes inside the requested outcome. Preserve unrelated existing work.\n- Prefer the smallest change that fixes the root cause or satisfies the requirement.\n- For review or analysis requests, do not modify files unless the user also asked for a\n  fix or implementation.\n\n## 3. Execute proportionally\n\n- Use the repository's documented conventions and checks when available.\n- For bugs, establish the failure mode before changing code.\n- For refactors, preserve externally observable behavior unless the user requested a\n  behavior change.\n- For risky or irreversible actions, stop unless the current request explicitly authorizes\n  them.\n\n## 4. Verify before claiming success\n\n- Run the checks that are relevant to the changed behavior: focused tests first, then\n  broader build, lint, type, or integration checks when warranted.\n- Read the actual output. A command invocation is not evidence unless its result supports\n  the claim.\n- If verification cannot run, report the limitation and do not describe the result as\n  verified.\n\n## 5. Deliver evidence\n\nReport concisely:\n\n1. What changed or what the review found.\n2. The evidence used to verify the result.\n3. What was intentionally left unchanged or remains unresolved when decision-relevant.\n\nDo not activate this workflow for unrelated conversational tasks simply because they can\nbe described as \"creating\", \"analyzing\", or \"writing\" something.\n"
}

SHA-256 of public snapshot: 5fa6ca523f4a86625b24fd5eedeb6304ff7404f814a4f86d0bcdd0de39edd588