Ambiguous tools
Descriptions that are empty, vague, overlapping, or missing the words that distinguish neighboring tools.
LintLang 0.4.0 · Python 3.10+ · Apache-2.0Local · deterministic · zero LLM
Static analysis for agent language
Lint the instructions your agents execute.
LintLang catches ambiguous tools, missing limits, conflicting directives, and schema mismatches in AGENTS.md, prompts, configuration, and Python source—before a model runs.
$ uvx lintlang scan AGENTS.md
REVIEW — 2 MEDIUM
H1.6 tool:search_docs ↔ tool:search_guides
Descriptions do not state how these tools differ.Inputs
Text · YAML · JSON · Python
Files, directories, or stdin
Outcomes
PASS · REVIEW · FAIL · ERROR
Nonblocking by default
CI formats
Terminal · JSON · SARIF 2.1.0
Stable rule IDs and honest locations
Runtime
PyYAML only
No model, API, telemetry, or network call
§ 01
What it catches
Missing constraints
Retry loops, tool-using instructions, and workflows that omit limits, stop conditions, or progress checks.
Contract mismatches
Tool descriptions that disagree with schemas, conflicting output requirements, and malformed message roles.
Embedded prompts
Prompt text and uncalibrated thresholds extracted from Python source without executing the application.
§ 02
From local check to CI gate
1Run once
uvx lintlang scan AGENTS.md2Review the baseline
Fix findings or record why the language is intentional.
3Gate deliberately
lintlang scan AGENTS.md --fail-on failA normal scan reports findings without failing the process. --fail-on failblocks only HIGH and CRITICAL findings; --fail-on review also blocks MEDIUM findings. Missing or malformed requested inputs remain errors either way.
GitHub Code Scanning
Put instruction findings beside code findings.
Version 0.4.0 emits deterministic SARIF 2.1.0 with stable LintLang rule IDs, repository-relative paths, and evidence-supported line spans where the parser can justify them. Structured inputs remain file-level rather than inventing precision.
Copy the least-privilege workflow ↗$ lintlang scan AGENTS.md \
--format sarif \
--fail-on fail \
> lintlang.sarif§ 03
Public evidence
0.4.0 releaseSARIF, release integrity, and quoted-example fixes3,800+ PyPI registry downloadsDownload events through 13 August 2026, not usersRecurring public CICharacter.AI Larch executes a pinned LintLang releaseIndependent Gentoo packagingAvailable through a third-party package overlayAttributed product influenceThe Agent Lint maintainer records ideas carried into that productTechnical noteThe bounded model behind H1.6 pairwise differentia
These signals establish public ecosystem evidence, not customers, users, endorsement, detector accuracy on external projects, or proof of production fitness.
The boundary
Static analysis can narrow risk. It cannot certify an agent.
LintLang does not run models, evaluate runtime behavior, prove semantic correctness, or establish that an agent is safe in production.
Use it before prompt tests, guardrails, and production monitoring—not instead of them.
See where static linting fits in the stack →Start with one real instruction file
Catch the defect while it is still just text.
$ uvx lintlang scan AGENTS.md