Speq Code Review logo

Speq Code Review

Community
marconae
speq-code-review

Code review tag taxonomy and findings format for the code-reviewer agent during /speq-implement. Not for general code review.

Overview

Publishermarconae
Repositoryspeq-skill
Skill namespeq-code-review
Stars
54
Forks
10
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 marconae on GitHub. Read the source before you install it.

Installation

Install the Speq Code Review 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/marconae/speq-skill/tree/main/.claude/skills/speq-code-review
  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/marconae/speq-skill.git /tmp/speq-skill
mkdir -p .claude/skills
cp -r /tmp/speq-skill/.claude/skills/speq-code-review .claude/skills/speq-code-review
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Speq Code Review 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 Speq Code Review 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 Speq Code Review 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.

Code Review Taxonomy

Analyze each changed file for the categories below.

1. Guardrail Violations

Per /speq-code-guardrails:

  • [TOO_MANY_ARGUMENTS]: more than 3 arguments
  • [SIDE_EFFECT]: function has a side effect its name and contract do not announce (a query that mutates, a command that touches unrelated state)
  • [BOOLEAN_FLAG_PARAMETER]: boolean flag parameter
  • [MAGIC_NUMBER]: magic number without a named constant (standing in for a failure, it is [SENTINEL_ERROR_VALUE], not this tag)
  • [MISSING_DOC_COMMENT]: missing doc comment on a public interface
  • [INLINE_COMMENT]: inline comment present (TODOs and other work-tracking comments are [WORK_TRACKING_COMMENT], not this tag)
  • [SELECTOR_ARGUMENT]: an argument (of any type, not just boolean) that picks which branch a function takes
  • [OUTPUT_PARAMETER]: a value returned via a mutated argument instead of the return value
  • [MIXED_ABSTRACTION_LEVEL]: a function mixes high-level orchestration with low-level detail
  • [COMMAND_QUERY_MIX]: a single call both mutates something and hands back an answer
  • [VAGUE_NAME]: a name that states no responsibility (Manager, Processor, Handler, Data, Info, Util)
  • [IMPLEMENTATION_IN_NAME]: a name that bakes in a transport, vendor, or format instead of the abstraction

2. Dead Code

  • [UNUSED_FUNCTION]: unused function or method
  • [UNREACHABLE_CODE]: unreachable code path
  • [UNUSED_IMPORT]: import not used
  • [UNUSED_VARIABLE]: variable assigned but never read

3. Test Quality

  • [OBSOLETE_TEST]: tests removed functionality
  • [DUPLICATE_TEST]: duplicate test coverage
  • [ASSERTION_FREE_TEST]: test always passes, no assertions
  • [VAGUE_TEST_NAME]: test name does not state the condition and expected behavior
  • [NONDETERMINISTIC_TEST]: test depends on real clock, network, filesystem, or unseeded randomness
  • [IMPLEMENTATION_COUPLED_TEST]: test asserts internal state instead of observable behavior
  • [UNTESTED_ERROR_PATH]: a failure path with no test
  • [MISSING_BOUNDARY_TEST]: no test for empty, single, maximum, off-by-one, or transition input
  • [SKIPPED_TEST]: test is skipped or ignored rather than fixed or deleted
  • [SUPPRESSED_WARNING]: a lint or compiler warning is silenced instead of resolved

4. Bad Comments

  • [REDUNDANT_COMMENT]: describes "what" not "why"
  • [OUTDATED_COMMENT]: does not match the code
  • [COMMENTED_OUT_CODE]: commented-out code block
  • [WORK_TRACKING_COMMENT]: TODO, FIXME, ticket refs

5. Optimization Opportunities

The Evidence Rule applies here: raise a finding in this category only with a measurement.

  • [PERFORMANCE_ISSUE]: obvious performance issue
  • [UNNECESSARY_ALLOCATION]: unnecessary allocation in a loop
  • [DUPLICATE_OPERATION]: operation that repeats work already done
  • [UNMEASURED_OPTIMIZATION]: a change justified as a performance optimization with no measurement behind it

6. YAGNI / Over-Engineering

  • [STANDARD_LIBRARY_DUPLICATE]: logic that reimplements something the language's standard library already provides
  • [SHRINKABLE]: same logic expressible in meaningfully fewer lines
  • [DEAD_FLEXIBILITY]: a feature flag, extension point, or parameter that is never varied
  • [UNNEEDED_DEPENDENCY]: a dependency added for something the standard library or an already-installed dependency already covers
  • [SPECULATIVE_ABSTRACTION]: an interface, generic type, or configuration value with exactly one implementation or caller, and not a seam over I/O, nondeterminism, or a third party

7. Error Handling

  • [SENTINEL_ERROR_VALUE]: a magic value or in-band signal stands in for an error instead of the language's own error mechanism
  • [CONTEXTLESS_ERROR]: an error that does not state what was attempted, the input that failed, or the constraint violated
  • [SWALLOWED_ERROR]: an error is discarded instead of handled or propagated
  • [BROAD_CATCH]: a catch broader than the specific error it handles
  • [LEAKED_PROVIDER_ERROR]: a third-party error type crosses a module boundary unwrapped
  • [ERROR_AS_CONTROL_FLOW]: an error mechanism used for expected, non-exceptional flow

8. Design Depth

  • [SHALLOW_MODULE]: learning the interface takes almost as much effort as the implementation behind it would, or classitis (many small modules named for a role, not a responsibility
  • [INFORMATION_COUPLING]: a single design choice (a format, a protocol, an execution-order split) shows up in more than one module and would need editing in both if it changed
  • [TACTICAL_SHORTCUT]: a shortcut taken with no follow-up to invest in the design
  • [MISSING_DESIGN_INTENT]: a public/interface comment states purpose but not the design intent or rationale a non-obvious abstraction needs
  • [BOUNDARY_VIOLATION]: business logic names a delivery mechanism, storage engine, or framework directly
  • [IO_IN_BUSINESS_LOGIC]: I/O performed directly inside business logic instead of through an injected abstraction
  • [AMBIENT_STATE_READ]: environment or global state read in place instead of injected
  • [LEAKED_BOUNDARY_TYPE]: a framework, storage, or third-party type crosses a module boundary
  • [DEPENDENCY_CYCLE]: a cycle in the module dependency graph
  • [SELF_CONSTRUCTED_DEPENDENCY]: a module constructs its own concrete dependency instead of receiving it
  • [PROVIDER_SHAPED_ABSTRACTION]: an abstraction shaped around a provider's API instead of the consumer's own vocabulary
  • [FEATURE_ENVY]: a function reaches into another module's data more than its own

Output Format

Write the findings document to specs/_plans/<plan-name>/review-findings.md per references/review-findings-template.md. Then return exactly one line and nothing else:

CODE REVIEW: <n> findings — standard: <n>, expert: <n> — specs/_plans/<plan-name>/review-findings.md

Routing

You partition the findings. The orchestrator never sees them individually. Place each finding under ## Standard fixes or ## Expert fixes in the findings document. The partition decides which single agent applies the whole fix pass: any Expert finding routes both sections to implementer-expert-agent. With no Expert finding, implementer-agent applies ## Standard fixes.

Every tag across all categories is eligible for either section. Route a finding to ## Expert fixes when its fix has cross-file, concurrency, or subtle-correctness implications: removing a dependency or abstraction with several call sites, correcting a dependency-direction or boundary violation, or any change whose failure mode is a passing test over wrong behavior. Everything else goes to ## Standard fixes. You hold the context for this call. Decide it here, do not defer 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 Speq Code Review AI skill do?

Code review tag taxonomy and findings format for the code-reviewer agent during /speq-implement. Not for general code review.

Why use Speq Code Review on TypingMind?

Because you install it once and use it with any model. Speq Code Review 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 Speq Code Review in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/marconae/speq-skill/tree/main/.claude/skills/speq-code-review. 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 Speq Code Review?

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 Speq Code Review?

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

Is the Speq Code Review AI skill free?

Yes. It is published on GitHub by marconae 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 👇