rules-doctor
An open-source command-line tool for inspecting and diagnosing the quality and reliability of AI coding agent rule files such as CLAUDE.md, .cursorrules, and AGENTS.md. It detects issues including contradictory instructions, missing scope definitions, and ambiguous phrasing.
Timeline (last 90 days)
rules-doctor is an open-source tool for checking the quality of AI agent rules files.
rules-doctor v0.1 addresses the issue of whether rules files can be loaded correctly, including detecting broken @import references, shadowed AGENTS.md files, and bloated instructions.
rules-doctor v0.2 enhanced import analysis capabilities.
The focus of rules-doctor 0.3 has shifted from whether rules can technically be loaded to whether the content is valuable.
rules-doctor 0.3 introduces four checks: CONTRADICTORY_RULES, SCOPE_BOUNDARY_MISSING, VAGUE_DIRECTIVE, and ADR_MISSING.
The CONTRADICTORY_RULES check is designated as error level, because contradictory rules directly cause non-deterministic behavior.
The three checks SCOPE_BOUNDARY_MISSING, VAGUE_DIRECTIVE, and ADR_MISSING are designated as warning level.
rules-doctor can be installed via pip install rules-doctor and run with the command rules-doctor <directory>.
rules-doctor's checks are essentially text-based heuristics rather than a model; contradiction detection works by matching instructions of opposite polarity under the same topic, and may miss subtle conflicts or produce false positives.
rules-doctor is a zero-dependency, pure standard-library implementation and is licensed under MIT.
1 more timeline events
All Facts (11)
Compressing token count and improving instruction quality are two orthogonal goals that need to be addressed separately.
50%Unverifiedrules-doctor v0.1 addresses the issue of whether rules files can be loaded correctly, including detecting broken @import references, shadowed AGENTS.md files, and bloated instructions.
50%Unverifiedrules-doctor v0.2 enhanced import analysis capabilities.
50%UnverifiedThe focus of rules-doctor 0.3 has shifted from whether rules can technically be loaded to whether the content is valuable.
50%Unverifiedrules-doctor is an open-source tool for checking the quality of AI agent rules files.
50%UnverifiedThe CONTRADICTORY_RULES check is designated as error level, because contradictory rules directly cause non-deterministic behavior.
50%UnverifiedThe three checks SCOPE_BOUNDARY_MISSING, VAGUE_DIRECTIVE, and ADR_MISSING are designated as warning level.
50%Unverifiedrules-doctor can be installed via pip install rules-doctor and run with the command rules-doctor <directory>.
50%Unverifiedrules-doctor's checks are essentially text-based heuristics rather than a model; contradiction detection works by matching instructions of opposite polarity under the same topic, and may miss subtle conflicts or produce false positives.
50%Unverifiedrules-doctor is a zero-dependency, pure standard-library implementation and is licensed under MIT.
50%Unverifiedrules-doctor 0.3 introduces four checks: CONTRADICTORY_RULES, SCOPE_BOUNDARY_MISSING, VAGUE_DIRECTIVE, and ADR_MISSING.
50%