Review Fix logo

Review Fix

Organization
DasDigitaleMomentum
review-fix

Apply explicitly accepted, evidence-backed defect fixes in the eligible existing session that produced the review or diagnosis, preserving secured evidence.

Overview

PublisherDasDigitaleMomentum
Repositoryopencode-processing-skills
Skill namereview-fix
Stars
61
Forks
3
Bundled files
1
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by DasDigitaleMomentum on GitHub. Read the source before you install it.

Installation

Install the Review Fix AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/DasDigitaleMomentum/opencode-processing-skills/tree/main/skills/review-fix
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/DasDigitaleMomentum/opencode-processing-skills.git /tmp/opencode-processing-skills
mkdir -p .claude/skills
cp -r /tmp/opencode-processing-skills/skills/review-fix .claude/skills/review-fix
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Review Fix in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Review Fix on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Review Fix is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Skill: Review Fix

This skill provides context-preserving defect remediation: apply explicitly accepted, evidence-backed fixes using the eligible existing session that produced the review or diagnosis. A review artifact is one supported finding source, not a prerequisite.

The point of preserving context is to reuse already-secured evidence, not collect it again. Related fixes may span files, call sites, and tests within the accepted objective. Reuse depends on correct and sufficient retained context/output, not the source workflow's mode or fix size.

When to Use

Use after:

  • review-implementation-plan identifies corrections to an implementation plan.
  • review-implementation identifies code, test, configuration, or integration corrections.
  • An analysis digest/message identifies a concrete defect and its diagnosis.
  • An Implementer digest/message identifies a concrete defect of the same completed work package.

Do not use for:

  • Initial review work.
  • Plan review corrections; plans remain conversation-owned by the primary. Resume the reviewer for clarification, then update the plan through update-plan.
  • A genuinely new objective, changed gated scope, new dependency decision, or user-requested independent perspective.
SituationRoute
Accepted defect, explicit Primary remediation instruction, eligible source sessionreview-fix in that existing session
New implementation objective/package, or fresh implementation needed under the fallback belowexecute-work-package in a fresh Implementer
New technical implementation-plan work, or fresh technical planning needed under the fallbackauthor-and-verify-implementation-plan in a fresh session
Conversation-owned plan corrections, including phase scope or acceptance changesPrimary via update-plan; reviewer may clarify, not edit the plan
Mechanical self-fix within the Maintainer's existing narrow allowanceMaintainer self-fix; no expansion of that allowance
Evidence retrievalretriever may supply evidence, but is not a remediation owner

Required Inputs

The continuation prompt must provide:

  • Finding source: review artifact path or analysis/implementation digest or message reference; multiple sources may be bundled.
  • Accepted defect set: IDs when present, otherwise stable short labels; include each defect's concrete description, expected behavior/invariant, and evidence.
  • Authority: the Primary has accepted the defects and explicitly instructed remediation within the supplied scope. This instruction is sufficient; no approval token or new Blueprint is required.
  • Session: exact source task_id, same persona and model/variant, and correct and sufficient retained review/diagnosis context/output for the assigned defects.
  • Target type: implementation-plan or implementation.
  • Scope: accepted objective, relevant boundaries, and Primary decisions. Do not require an exact file allowlist when related call sites or tests need to be discovered.
  • Verification: command or expected verification goal. Identify whether a supplied command is broad/full. The fixer may choose focused targeted tests when none were supplied.
  • Closure: one bounded remediation pass, compact digest, then stop and return unresolved defects or decisions to the Primary.

Use tpl-review-fix-prompt.md for an eligible continuation.

Session eligibility and fallback

  • If the source session's retained review/diagnosis context/output is correct and sufficient, resume its exact task_id with the same persona and model/variant through review-fix.
  • If that context/output is faulty, insufficient, or unavailable, the Primary starts a fresh session through the appropriate route above, with a pointer to the finding source as a hint. The fresh session must establish the evidence it needs; never pretend it has the old context. This is a fresh workflow, not review-fix session reuse.
  • An Implementer may be resumed after its digest only for accepted defects of the same completed package under this contract. This narrow retirement exception cannot start another phase/package or authorize indefinite continuation.
  • The Primary may bundle findings from multiple source sessions in one pass and assigns defects to the eligible sessions explicitly. Each resumption retains its own exact task_id and persona/model variant; pointers to other sources do not confer their retained context. Return each assigned digest to the Primary; bundling does not create an automatic continuation loop.

Protocol

  1. Read the finding source and reuse the secured evidence in the existing session context. Inspect current state or evidence gaps only as needed for the fix; do not repeat broad evidence collection.
  2. Confirm explicit Primary authorization, session eligibility, and that the fixes still serve the accepted objective. If eligibility fails, stop and report the fallback need to the Primary before dependent edits.
  3. Apply the necessary related corrections. For implementation-plan scope findings, remove, merge, or simplify the accepted steps/artifacts instead of adding compensating design. Size alone does not require a new work package.
  4. For implementation fixes, use targeted tests while editing and preserve any supplied broad/full final gate. For implementation-plan fixes, check only the changed steps and directly affected references; do not invent a formal verification suite.
  5. Return the compact digest below and stop. This is one remediation pass, not the start of a review loop.

Remediation posture

No Gold-Plating. No Adversarial Reviewing. No Scope Creep. Fix accepted, evidence-backed findings and the related changes required to complete them. Do not use remediation to invent improvements, reopen rejected findings, or create another review/fix cycle.

The Primary's explicit instruction naming the accepted defects and scope is the gate; acceptance alone is not an instruction to edit. Do not add unrelated improvements or silently change the objective. If a new decision is required, report it to the Primary; do not invent a new gate or switch agents autonomously.

Write Boundary

  • For implementation-plan, edit the relevant implementation-plan artifacts and related references required by the accepted findings.
  • For implementation, edit related code, configuration, integration points, and tests required by the accepted findings.
  • Do not modify a source review artifact; it remains the immutable record of the independent review.
  • Do not modify phase scope, acceptance criteria, or unrelated docs/plans.
  • Do not perform Git history operations.

Output Contract

Return only:

  • Outcome: succeeded | partial | blocked
  • Findings: fixed and unresolved defect IDs or stable short labels
  • Edits: changed files with one-line descriptions
  • Verify: command/checks and result
  • Next: stop | primary decision required; an independent re-review was not started

Independent Re-review

The remediation pass is not an independent review. A fresh review is optional and requires an explicit primary or user decision. The primary may choose it when:

  • the primary wants an independent final opinion;
  • the fix exposes uncertainty that the current fixer cannot resolve;
  • security, persistence, public API, or migration risk warrants another perspective.

Do not start or recommend a fresh review merely because findings were fixed, steps were removed, or the change spans several files. Do not chain review -> fix -> review -> fix. After one remediation pass, stop. Materially new risk/uncertainty must be escalated to the Primary, who decides whether to request an independent re-review; it never triggers one automatically.

Rules

  • Follow the session eligibility and fallback rules above; never substitute a fresh session while claiming retained context.
  • Keep the same persona and model/variant while resuming.
  • Do not re-open already rejected findings unless the primary explicitly asks for reconsideration.
  • Do not describe self-verification as an independent review.
  • Never weaken, delete, or skip tests to make verification pass.
  • Do not replace removed gold-plating with new abstraction, infrastructure, tests, or documentation unless an accepted finding and the accepted objective require it.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Review Fix AI skill do?

Apply explicitly accepted, evidence-backed defect fixes in the eligible existing session that produced the review or diagnosis, preserving secured evidence.

Why use Review Fix on TypingMind?

Because you install it once and use it with any model. Review Fix is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Review Fix in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/DasDigitaleMomentum/opencode-processing-skills/tree/main/skills/review-fix. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Review Fix?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Review Fix?

As many as you like. As long as a model supports skills, you can use Review Fix with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Review Fix AI skill free?

Yes. It is published on GitHub by DasDigitaleMomentum under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇