Skip to main content

robotframework-lint-lsp

Robot Framework lint for coding agents. The agent edits a .robot file. Safe fixes are applied right after the edit. Only the findings left come back to the agent, one short line each.

PyPI Python License: MIT

  • Language server (robotframework-lint-lsp): robocop findings plus the RFL rules below, sent after every edit. Each finding is one short line. It also gives definition, references, hover and workspace symbols for your own keywords.
  • Auto-fix hook (robotframework-lint-autofix): after each Edit or Write it applies robocop's safe fixes (VAR syntax, inline IF, one assignment style). The agent spends no turns on them.
  • Rule pack: six robocop rules from the Robot Framework 7.5 release notes that robocop doesn't check yet.

Contents

Does it help?

Measured on one task (write two suites with keywords, then run them), robocop 9.4.0 default findings left in the final files:

Model No lint robocop MCP This plugin
Haiku (25 runs per arm) 18.5 13.2 4.7
Sonnet (3 runs) 12.0 -- 2.0
Opus (3 runs) 11.0 -- 0

On Haiku and Sonnet it costs about 10 percent more than no lint. It costs less than the robocop MCP on all three models. Every run, version and chart: benchmarks (source).

Install

Needs Python 3.10 or later and Claude Code 2.1.275 or later.

  1. Install the package where your agent can run it:

    pip install robotframework-lint-lsp
    

    It pins robotframework 7.5.1, robotframework-robocop 9.4.0 and pygls 2.1.1. To keep those pins apart, install it in a venv on PATH.

  2. Add the plugin in Claude Code:

    /plugin install robotframework-lint-lsp --marketplace ai94iq/robotframework-lint-lsp
    
  3. Restart Claude Code. Edit any .robot or .resource file. The findings come back with the edit result.

To check it works, run claude --debug-file debug.log, then grep "LSP Diagnostics: Delivering" debug.log.

Lint on the command line, without the agent:

robocop check --custom-rules robotframework_lint path/to/tests

robocop list rules has no --custom-rules flag. To list the RFL rules, put custom_rules = ["robotframework_lint"] under [tool.robocop.lint] in a config file and run robocop list rules --config <file>.

What the agent gets

Feature Behaviour
Diagnostics On open, save and change (debounced 0.4 s). One line per finding, RULEID message, at most 120 characters. At most 20 per file, then one line saying how many were cut. Warnings and errors only.
Fix hints A finding robocop can fix ends with fix all: <command>; fixable Info rules are raised to Warning so the agent sees them.
Auto-fix The PostToolUse hook runs robocop's safe fixes on the edited file in an isolated temp copy, keeps your line endings, and never overwrites a newer edit.
Navigation Definition and references matched the way Robot matches names (say_hello finds Say Hello); hover shows the signature and first doc line; workspace symbols rank exact, then prefix, then substring. At most 50 results.

On start the server indexes up to 2000 .robot and .resource files under the workspace root, skipping .git, .venv, node_modules, results, links and files over 1 MB. Library keywords such as Log are not indexed.

Settings

Set them in initializationOptions of your .lsp.json:

Option Default Meaning
maxDiagnostics 20 Findings per file before the overflow line
minSeverity warning info, warning or error
select all Rule IDs to report only
ignore none Rule IDs to drop
groupFindings off One line per rule with its line numbers
robotcodeCommand off Command list for robotcode in your project venv, for example ["D:/proj/.venv/Scripts/robotcode"]. On save it adds up to 5 errors robocop cannot see: unresolved keywords, variables and imports

robotcodeCommand runs on the saved file in your workspace, not a temp copy, and robotcode imports your project's libraries to resolve them. Set it only for projects whose code you trust. Point it at the project's own venv: a venv without the project's libraries reports false import errors.

Rules

ID Name Source
RFL001 for-in-zip-without-mode (auto-fix adds mode=SHORTEST) RF 7.0
RFL002 deprecated-standard-library (Telnet) RF 7.5
RFL003 deprecated-tag-pattern (XORY, &) RF 7.5
RFL004 deprecated-translation (French Unités de test, Finnish Tagit) RF 7.5
RFL005 doc-tags-without-empty-row (Tags: in keyword docs) RF 7.5
RFL006 deprecated-argument-name (list_=, time_=) RF 7.5

RFL003 asks Robot Framework's own TagPatterns, so it reports exactly what RF 7.5.1 warns about, including all-caps tags such as -DONOTX that RF reads as DO NOT X. RFL004 reports a term only when RF parsed it as a header or setting, that is, in a French or Finnish file.

Other agents and editors

Setup for opencode, Crush, Codex, GLM and any LSP client, and how to call the hook from other hook systems: docs/agents.md.

Development

python -m venv .venv
.venv/bin/python -m pip install -e ".[dev]"   # Windows: .venv/Scripts/python
.venv/bin/python -m pytest

Releases publish to PyPI from a GitHub release through trusted publishing (.github/workflows/release.yml).

License

MIT

Metadata

Release files for robotframework-lint-lsp 1.1.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for robotframework-lint-lsp 1.1.0
File Size Uploaded Provenance Actions
robotframework_lint_lsp-1.1.0.tar.gz 31.3 kB With provenance Details for robotframework_lint_lsp-1.1.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for robotframework-lint-lsp 1.1.0
File Interpreter ABI Platform Actions
robotframework_lint_lsp-1.1.0-py3-none-any.whl Python 3 none any Details for robotframework_lint_lsp-1.1.0-py3-none-any.whl

Total release size: 56.5 kB

Release files / robotframework_lint_lsp-1.1.0.tar.gz

Download URL robotframework_lint_lsp-1.1.0.tar.gz
Size 31.3 kB
Tags Source
SHA-256 checksum
How to use checksums
69d83751fe62b776b5e8d28b12e9212a142f565088fed7411620ddd1223a5250
BLAKE2b-256 checksum
How to use checksums
4108184d8220b75ec9cc4ff421c08156c3a7af6ea11b04d982b47f3b69b7d692
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 15, 2026.

Transparency log

Release files / robotframework_lint_lsp-1.1.0-py3-none-any.whl

Download URL robotframework_lint_lsp-1.1.0-py3-none-any.whl
Size 25.1 kB
Tags Python 3
SHA-256 checksum
How to use checksums
3cef3612e51ee6576aa77ec166f572ee7981c4c3623a0cedff64ebe52c316887
BLAKE2b-256 checksum
How to use checksums
77d2ab291832864403ff5a480d66744465070c871d91d693ec6c3c354b72cdb0
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 15, 2026.

Transparency log

Security for robotframework-lint-lsp 1.1.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

Both files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for robotframework_lint_lsp-1.1.0.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 15, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

1.1.0 This release

2 release files Full provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page