Skip to content
Back to search

deslop

Claude Code distribution for the deslop Agent Skill.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/mrzoyo--deslop-gpt/badge.svg)](/listings/mrzoyo--deslop-gpt)

HTML

<a href="/listings/mrzoyo--deslop-gpt"><img src="/api/listings/mrzoyo--deslop-gpt/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Codex: install v0.3.4 as a standalone Skill

Invoke the bundled installer with this GitHub Skill URL:

$skill-installer
Install the Skill from:
https://github.com/MrZoyo/deslop-GPT/tree/v0.3.4/skills/deslop

For a reviewable local checkout, symlink the runtime directory into Codex's canonical user Skill location:

git clone --branch v0.3.4 --depth 1 https://github.com/MrZoyo/deslop-GPT.git "$HOME/.local/share/deslop-GPT"
mkdir -p "$HOME/.agents/skills"
ln -s "$HOME/.local/share/deslop-GPT/skills/deslop" "$HOME/.agents/skills/deslop"

After installation, confirm that the selected Skill path points to the intended release; see the discovery checks in [Getting Started](docs/getting-started.md). The tagged v0.3.4 path is the current released, pinned standalone Skill; [`main`](https://github.com/MrZoyo/deslop-GPT/tree/main/skills/deslop) is the development branch and may contain unreleased changes.

Claude Code: install the Plugin from GitHub

Inside Claude Code, add this repository as a marketplace and install the Plugin:

/plugin marketplace add MrZoyo/deslop-GPT
/plugin install deslop@deslop

The canonical Plugin command is `/deslop:deslop`. The v0.3.4 catalog uses an explicit HTTPS Git source pinned to the release tag. For a local checkout, run `claude --plugin-dir .` from the repository root. [0.3.4 release notes](docs/0.3.4.md) describe the narrow persistence-policy change and mixed evaluation results; historical evidence remains tied to its exact payload.

One checkout, standalone discovery on both hosts

The same released runtime payload can be linked into each host's user Skill directory:

mkdir -p "$HOME/.agents/skills" "$HOME/.claude/skills"
ln -s "$HOME/.local/share/deslop-GPT/skills/deslop" "$HOME/.agents/skills/deslop"
ln -s "$HOME/.local/share/deslop-GPT/skills/deslop" "$HOME/.claude/skills/deslop"

Use only the link for the host you need, and run each `ln` command only when its destination does not already exist. A standalone Claude Code installation invokes the Skill as `/deslop`. See [Getting Started](docs/getting-started.md) for installation scope, v0.1.0 migration, updates, removal, and a safer review-first workflow. `deslop` is an independent community project, not an OpenAI or Anthropic product.

Distribution status

The shared [`skills/deslop/`](skills/deslop/) payload follows the open Agent Skills structure and is used unchanged by Codex and Claude Code. [`.claude-plugin/plugin.json`](.claude-plugin/plugin.json) and [`.claude-plugin/marketplace.json`](.claude-plugin/marketplace.json) provide Claude Code packaging. This release distributes a standalone Skill for Codex and does not include Codex Plugin metadata. The historical Plugin experiment does not describe current platform support. See the [distribution compatibility note](docs/development.md#distribution-compatibility-note).

Invoke it explicitly

| Host and distribution | Command name | | --- | --- | | Codex standalone Skill | `$deslop` | | Claude Code standalone Skill | `/deslop` | | Claude Code Plugin | `/deslop:deslop` |

Append the same mode and scope arguments to the command name for each host:

| Arguments | Effect | | --- | --- | | none | Read-only audit of the established scope | | `audit` | Explicit read-only audit | | `apply` | Apply reviewed cleanup within scope | | `tests apply` | Prioritize test signal and mutual-support test/code clusters | | `current branch apply` | Clean current work relative to its actual merge base | | `deep` | Repository-wide read-only audit | | `deep apply` | Repository-wide cleanup without redesign |

Only `apply` authorizes edits. Staging, commits, pushes, branch changes, resets, and fetching still require separate permission.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 21 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. {
  2. "$schema": "https://json.schemastore.org/claude-code-marketplace.json",
  3. "name": "deslop",
  4. "owner": {
  5. "name": "MrZoyo",
  6. "url": "https://github.com/MrZoyo"
  7. },
  8. "description": "Claude Code distribution for the deslop Agent Skill.",
  9. "plugins": [
  10. {
  11. "name": "deslop",
  12. "source": {
  13. "source": "url",
  14. "url": "https://github.com/MrZoyo/deslop-GPT.git",
  15. "ref": "v0.3.4",
  16. "sha": "c6d8245fe9a2ff8fa1cb87528906ffafc038c286"
  17. },
  18. "description": "Audit or apply test-first, evidence-backed cleanup for accumulated test, verification, and fallback bloat."
  19. }
  20. ]
  21. }
README.md · README · 241 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. <p align="center">
  2. <img src="assets/deslop-banner.svg" alt="deslop — evidence-backed code cleanup" width="100%">
  3. </p>
  4. <h1 align="center">deslop</h1>
  5. <p align="center">
  6. <strong>A deletion-first Agent Skill for agent-maintained codebases</strong>
  7. </p>
  8. <p align="center"><a href="README.zh-CN.md">简体中文</a> · <strong>English</strong></p>
  9. <p align="center">
  10. Evidence-backed cleanup that reduces accumulated machinery while preserving real behavior.
  11. </p>
  12. <p align="center">
  13. <a href="https://github.com/MrZoyo/deslop-GPT/actions/workflows/validate.yml"><img src="https://github.com/MrZoyo/deslop-GPT/actions/workflows/validate.yml/badge.svg" alt="Validate workflow"></a>
  14. <a href="LICENSE"><img src="https://img.shields.io/badge/license-MIT-2ea44f.svg?style=flat-square" alt="MIT license"></a>
  15. <a href="skills/deslop/"><img src="https://img.shields.io/badge/Agent%20Skill-Codex%20%2B%20Claude%20Code-0969da.svg?style=flat-square" alt="Codex and Claude Code compatible Agent Skill"></a>
  16. <a href="#safety-model"><img src="https://img.shields.io/badge/default-read--only-6e7781.svg?style=flat-square" alt="Read-only by default"></a>
  17. <a href="evals/real-world/cluster-gpu-monitor/README.md"><img src="https://img.shields.io/badge/field%20trial-manually%20adjudicated-8250df.svg?style=flat-square" alt="Manually adjudicated field trial"></a>
  18. </p>
  19. `deslop` audits and, when explicitly authorized, removes complexity accumulated through repeated coding-agent implementation and correction cycles. Those cycles often leave overlapping regression tests, producer-verifies-producer checks, and fallback layers that hide failures instead of handling a current contract.
  20. This is semantic subtraction, not source beautification. `deslop` is not a formatter, style humanizer, test-count minimizer, blanket ban on defensive code, or automatic permission to edit a repository. It follows justification chains to independent evidence and preserves behavior whose contract remains real or uncertain.
  21. > **Reduce test surface, not behavior surface.**
  22. ## What it targets
  23. The percentages below are design priorities, not measured prevalence.
  24. | Priority | Target | Question |
  25. | ---: | --- | --- |
  26. | ~50% | **Test-suite bloat** | Does each test protect a distinct failure domain with a current owner and an independent oracle? |
  27. | ~25% | **Verification theater** | Can the verifier fail independently from the producer, or do both share the same information and failure domain? |
  28. | ~25% | **Defensive / fallback bloat** | Does the recovery path implement a current contract, or merely mask an unexpected internal error? |
  29. Generic dead code, wrappers, abstractions, and comments are secondary. They matter only when they belong to one of these clusters or have direct high-confidence deletion evidence.
  30. ## Subtract machinery. Preserve behavior.
  31. | Remove | Preserve |
  32. | --- | --- |
  33. | Self-justifying or duplicate tests | Distinct success, rejection, error, and edge-case behavior |
  34. | Checksums, receipts, or validators with no independent consumer | Persistence and corruption checks across a real failure boundary |
  35. | Speculative or obsolete fallback chains | Supported compatibility and documented protocol behavior |
  36. | Repeated defenses inside trusted call graphs | Real handling at external and untrusted boundaries |
  37. | Wrapper/test clusters with no independent purpose | Security, transactions, concurrency, resource, and scientific invariants |
  38. Resemblance to a smell is a lead, not a verdict. Security and trust boundaries, supported callers, persisted formats, and numerical constraints are preserved by default when evidence is incomplete.
  39. ## Quick Start
  40. ### Codex: install v0.3.4 as a standalone Skill
  41. Invoke the bundled installer with this GitHub Skill URL:
  42. ```text
  43. $skill-installer
  44. Install the Skill from:
  45. https://github.com/MrZoyo/deslop-GPT/tree/v0.3.4/skills/deslop
  46. ```
  47. For a reviewable local checkout, symlink the runtime directory into Codex's canonical user Skill location:
  48. ```bash
  49. git clone --branch v0.3.4 --depth 1 https://github.com/MrZoyo/deslop-GPT.git "$HOME/.local/share/deslop-GPT"
  50. mkdir -p "$HOME/.agents/skills"
  51. ln -s "$HOME/.local/share/deslop-GPT/skills/deslop" "$HOME/.agents/skills/deslop"
  52. ```
  53. After installation, confirm that the selected Skill path points to the intended release; see the discovery checks in [Getting Started](docs/getting-started.md). The tagged v0.3.4 path is the current released, pinned standalone Skill; [`main`](https://github.com/MrZoyo/deslop-GPT/tree/main/skills/deslop) is the development branch and may contain unreleased changes.
  54. ### Claude Code: install the Plugin from GitHub
  55. Inside Claude Code, add this repository as a marketplace and install the Plugin:
  56. ```text
  57. /plugin marketplace add MrZoyo/deslop-GPT
  58. /plugin install deslop@deslop
  59. ```
  60. The canonical Plugin command is `/deslop:deslop`. The v0.3.4 catalog uses an explicit HTTPS Git source pinned to the release tag. For a local checkout, run `claude --plugin-dir .` from the repository root. [0.3.4 release notes](docs/0.3.4.md) describe the narrow persistence-policy change and mixed evaluation results; historical evidence remains tied to its exact payload.
  61. ### One checkout, standalone discovery on both hosts
  62. The same released runtime payload can be linked into each host's user Skill directory:
  63. ```bash
  64. mkdir -p "$HOME/.agents/skills" "$HOME/.claude/skills"
  65. ln -s "$HOME/.local/share/deslop-GPT/skills/deslop" "$HOME/.agents/skills/deslop"
  66. ln -s "$HOME/.local/share/deslop-GPT/skills/deslop" "$HOME/.claude/skills/deslop"
  67. ```
  68. Use only the link for the host you need, and run each `ln` command only when its destination does not already exist. A standalone Claude Code installation invokes the Skill as `/deslop`. See [Getting Started](docs/getting-started.md) for installation scope, v0.1.0 migration, updates, removal, and a safer review-first workflow. `deslop` is an independent community project, not an OpenAI or Anthropic product.
  69. ### Distribution status
  70. The shared [`skills/deslop/`](skills/deslop/) payload follows the open Agent Skills structure and is used unchanged by Codex and Claude Code. [`.claude-plugin/plugin.json`](.claude-plugin/plugin.json) and [`.claude-plugin/marketplace.json`](.claude-plugin/marketplace.json) provide Claude Code packaging. This release distributes a standalone Skill for Codex and does not include Codex Plugin metadata. The historical Plugin experiment does not describe current platform support. See the [distribution compatibility note](docs/development.md#distribution-compatibility-note).
  71. ### Invoke it explicitly
  72. | Host and distribution | Command name |
  73. | --- | --- |
  74. | Codex standalone Skill | `$deslop` |
  75. | Claude Code standalone Skill | `/deslop` |
  76. | Claude Code Plugin | `/deslop:deslop` |
  77. Append the same mode and scope arguments to the command name for each host:
  78. | Arguments | Effect |
  79. | --- | --- |
  80. | none | Read-only audit of the established scope |
  81. | `audit` | Explicit read-only audit |
  82. | `apply` | Apply reviewed cleanup within scope |
  83. | `tests apply` | Prioritize test signal and mutual-support test/code clusters |
  84. | `current branch apply` | Clean current work relative to its actual merge base |
  85. | `deep` | Repository-wide read-only audit |
  86. | `deep apply` | Repository-wide cleanup without redesign |
  87. Only `apply` authorizes edits. Staging, commits, pushes, branch changes, resets, and fetching still require separate permission.
  88. ## Example workflow
  89. Start with evidence, not edits:
  90. ```text
  91. $deslop deep
  92. HIGH
  93. - two fallback layers handle the same internal parse failure;
  94. current callers and history show no supported legacy input
  95. - a local receipt is produced and verified by the same workflow;
  96. no external consumer or persisted trust boundary exists
  97. PRESERVE
  98. - a persisted readback detects truncated output across a write/read boundary
  99. - a compatibility branch is required by a documented external protocol
  100. ```
  101. Review each evidence chain and preservation decision. Apply only the supported scope:
  102. ```text
  103. $deslop deep apply
  104. ```
  105. The example is schematic; it does not represent a benchmark fixture or performance claim.
  106. ## How deslop decides
  107. - **Independent evidence roots:** current requirements, real callers, public contracts, protocols, trust boundaries, persistence boundaries, or scientific invariants.
  108. - **Closed justification loops:** production code and tests do not become necessary merely by justifying each other.
  109. - **Production reachability and edge closure:** prove the current input-to-consumer path, not only isolated callers or test-injected branches.
  110. - **Production/test asymmetry:** redundant test evidence can be removed without deleting the behavior it observes.
  111. - **Fail-visible bias:** unexpected internal failures should surface unless a concrete recovery or translation contract exists.
  112. - **Subtraction without redesign:** dependencies, abstractions, wrappers, compatibility layers, and replacement scaffolding have a default budget of zero.
  113. The full decision model is documented in [Design](docs/design.md). The self-contained runtime [`SKILL.md`](skills/deslop/SKILL.md) remains authoritative for agent behavior.
  114. ## Safety model
  115. Codex enforces explicit invocation through [`allow_implicit_invocation: false`](skills/deslop/agents/openai.yaml). Claude Code does not read that OpenAI-specific metadata; the shared standards-compatible frontmatter instead tells Claude to invoke `deslop` explicitly. Claude Code may still select the Skill from its description — a [2026-09-03 control run](evals/runtime-controls/results/claude-code-haiku-20260903.md) recorded it doing exactly that — but such an invocation remains read-only unless the user includes `apply`. Default and `audit` modes are read-only, and suspicious constructs can be recorded as deliberate preservation decisions. Code is not removable merely because it looks defensive, was written by an agent, or has a test that could be deleted.
  116. Read-only verification should redirect caches or generated output when practical and disclose incidental residue. Apply authorization permits scoped edits; it does not resolve uncertainty in favor of deletion. See [Getting Started](docs/getting-started.md) for the review sequence and [Design](docs/design.md) for confidence classes and preserved boundaries.
  117. ## Evidence
  118. ### Validation status
  119. | Runtime payload | Host and path | Runs | What it establishes |
  120. | --- | --- | --- | --- |
  121. | v0.3.2 candidate exact hash | Codex collaboration subagents, `gpt-5.6-sol` with `high` reasoning, direct Skill loading | 4 selected `dev-v2` cases, 2 runs per payload and case, 16 calls | v0.3.2 passed 8/8 versus v0.3.1 at 7/8; `t02b` was 2/2 versus 1/2, while all 12 deletion-case runs passed |
  122. | v0.3.1 release payload hash | Claude Code 2.1.259 CLI, Haiku 4.5, `.claude/skills` discovery | 3 runtime controls, 1 run each, no baseline | Claude selected the Skill from its description alone and still stayed read-only; a question asking for no cleanup did not pull it in |
  123. | v0.3.1 release payload hash | Claude Code 2.1.259 CLI, Haiku 4.5, `.claude/skills` discovery | 5 `dev-v2` micro cases, 3 apply runs each, no baseline | 12 of 15 runs passed every hidden gate; `t02b` lost its supported legacy header path in 2 of 3 runs |
  124. | v0.3.1 release payload, pre-release exact hash | Codex subagents loading the Skill by path | 1 default audit plus `t02b` and `t03b` preservation cases | Narrow development regression smoke; all three left fixture content unchanged |
  125. | v0.3.1 tagged Plugin | Claude Code 2.1.259, isolated config and remote `main` catalog | 1 marketplace installation, no model call | Remote HTTPS source resolved the v0.3.1 tag to commit `a19128d`; installed version and runtime hash matched |
  126. | v0.3.0, exact release hash | Codex subagents loading the Skill by path | 3 mini-repository apply runs plus 1 audit | All three cleaned artifacts passed hidden behavior, reduction, and negative-change gates; no CLI discovery or baseline evidence |
  127. | v0.3.0, exact release hash | Claude Code 2.1.259 local Plugin, Haiku 4.5 | 1 audit plus 1 apply | Plugin loading and one valid cleanup artifact; apply stopped at its turn ceiling before the final report |
  128. | Earlier development payloads | Codex CLI 0.149.1, `gpt-5.6-sol` | rc3 micro, rc4 mini, and targeted rc5 diagnostics | Historical development evidence tied only to those payload hashes |
  129. Version-bound forward smokes are published under [`evals/release-smoke/`](evals/release-smoke/). The 2026-09-04 cross-version smoke is the v0.3.2 release gate; it uses known cases and two runs per payload, so it supports only the narrow release decision recorded above. The earlier Claude Code CLI runs remain under [`evals/runtime-controls/results/`](evals/runtime-controls/results/) and [`evals/dev-v2-focused/results/`](evals/dev-v2-focused/results/). Those Haiku runs exposed the `t02b` ambiguity but remain secondary host diagnostics, not target-model release evidence. None of these records is held-out model-effect evidence. The older rc3 micro pilot measured 63.1% more total tokens and 16.5% more wall time with its then-current Skill; that one-run result does not predict v0.3.x cost, but it supports using `deslop` for deliberate accumulated-slop work rather than routine tiny diffs.
  130. ### Focused development evaluation
  131. [`dev-v2-focused`](evals/dev-v2-focused/README.md) tests preservation and simplification decisions across paired micro cases and three end-to-end miniature repositories. Behavior gates run before reduction metrics. Micro and mini-repository results remain separate, and the repository publishes no project-level performance score.
  132. The follow-up [`dev-v3-evidence-edges`](evals/dev-v3-evidence-edges/README.md) draft records 19 anonymized field observations and contains nine executable pairs. Draft2 adds a public-API/manifest pair and strengthens correction-history and coverage calibration; draft3 adds a fingerprint-contract pair with a zero-test polarity state, and draft4 reviews retirement evidence and grading recognition. The 0.3.4 notes report a separate exposed development comparison using the edge-coverage-v1 revision; it is not holdout evidence or a claim of general superiority.
  133. The [0.3.4 release](docs/0.3.4.md) adds lifecycle counterexamples, independently designed CSV/SQLite cases, and a separate coverage-recognition revision. Historical inputs and scores remain unchanged.
  134. See [Evaluation](docs/evaluation.md) for interpretation limits and [`evals/README.md`](evals/README.md) for the canonical protocol.
  135. ### Real-world field trials
  136. | Case | Method | Status |
  137. | --- | --- | --- |
  138. | [`cluster-gpu-monitor`](evals/real-world/cluster-gpu-monitor/README.md) | Real repository; read-only audit, human adjudication, then two reviewed cleanup batches | Frozen historical evidence |
  139. The first field trial records both accepted cleanups and deliberate preservation decisions with public before/after provenance. It had no independent baseline run from the same frozen state, so it is not a controlled A/B comparison and does not establish general superiority, 100% precision, or production-proven correctness.
  140. Future cases can be added without becoming Skill-tuning inputs; see [Field Trials](docs/field-trials.md).
  141. ## Documentation
  142. | Document | Purpose |
  143. | --- | --- |
  144. | [Documentation index](docs/README.md) | Choose a user, design, evidence, or development path |
  145. | [Getting Started](docs/getting-started.md) | Installation, invocation modes, scopes, updates, and safe workflows |
  146. | [Design](docs/design.md) | Evidence roots, closed loops, preservation, and subtraction principles |
  147. | [Evaluation](docs/evaluation.md) | Focused corpus, hard gates, run discipline, and interpretation limits |
  148. | [Field Trials](docs/field-trials.md) | Real-world methodology, provenance, isolation, and case registry |
  149. | [Development](docs/development.md) | Repository layout, validation, contribution, and release boundaries |
  150. ## Repository structure
  151. ```text
  152. .claude-plugin/ Claude Code Plugin and marketplace metadata
  153. skills/deslop/ Self-contained runtime Skill payload
  154. docs/ User, design, evidence, and development guides
  155. evals/dev-v2-focused/ Active focused development evaluation
  156. evals/dev-v3-evidence-edges/ Follow-up evidence-edge draft
  157. evals/runtime-controls/ Authorization and host/runtime controls
  158. evals/release-smoke/ Version-bound forward-smoke records
  159. evals/real-world/ Manually adjudicated real-world evidence
  160. evals/archive/ Retired historical evaluation material
  161. scripts/ Validation and evaluation tooling
  162. assets/ README and project presentation assets
  163. ```
  164. ## Project status and contributing
  165. Public releases use semantic versioning, beginning with v0.1.0. A `0.x` release is usable but still evolving; it is not a `stable`, `production-ready`, or 1.0-quality claim. Immutable Git tags identify released runtime and distribution states. Benchmark candidates retain their separate evaluation tags.
  166. The most useful contribution is an evidence-backed case with a nearby preservation counterexample and an independent behavioral oracle—not an isolated snippet that merely looks verbose. Read [Development](docs/development.md) before proposing a Skill policy or evaluation change.
  167. ## License
  168. [MIT](LICENSE)

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.