Instructions file CodexOpenCode
Repository instructions for qlens, including required session-start reading, links to the Anchorlaw rules, and the repository’s verification rules. A session is one period of work by the coding agent.
Instructions file CodexOpenCode
Project-specific instructions for Anchorlaw, a language-independent protocol for checking whether software claims can be verified. They describe required session setup, testing, and the project's Judge-led development process.
Skill Claude CodeCodex
A workflow for challenging a scanner rule when its warning may be a false positive, meaning it flags code that is actually acceptable.
Skill Claude CodeCodex
A reference guide for the meaning and status of “anchors”—claims linked to checks or honest statements about what is unknown. It explains the rules for using them in Python, TypeScript, and C++.
Skill Claude CodeCodex
A fallback verification procedure for generated code that cannot be compiled or run on its own. It labels the verification as full, partial, or degraded and records unresolved dependencies.
Skill Claude CodeCodex
An isolated review role that checks whether work meets its acceptance criteria and reports its findings without changing the work's status.
Skill Claude CodeCodex
An internal maintenance workflow for the Anchorlaw protocol, a language-independent set of rules and implementations. It requires tests, self-checking, evidence in protocol documents, and disciplined versioning and commits.
Skill Claude CodeCodex
A system for recording runtime failures as structured “noise cards,” which are saved notes about unexpected behavior and its cause. Unresolved cards can later be turned into regression tests, tests that prevent the same failure from returning.
Skill Claude CodeCodex
A static code-review procedure that runs Anchor Law scanners to find defensive programming patterns and labels findings as errors, warnings, or information.
Skill Claude CodeCodex
An isolated role that drafts an implementation specification from already confirmed requirements and technical rules. The draft covers naming, module structure, dependencies, framework boundaries, and interfaces for review by a judge.
Skill Claude CodeCodex
A verification command for running `anch orlaw test` and assigning each function a health state. It also handles code that cannot be compiled on its own by using a defined fallback verification method.
Skill Claude CodeCodex
An isolated implementation role that builds one assigned software module according to an approved plan, then records its work and tests for review.
Skill Claude CodeCodex
A tool for adding Anchorlaw labels to public functions: test claims backed by runtime evidence, or explicit statements about what is still unknown.
Skill Claude CodeCodex
A procedure for challenging a scanner rule when its warning may be a false positive, meaning it flags code that is not actually a problem.
Skill Claude CodeCodex
A reference guide to Anchorlaw’s anchors: labels attached to code that state what can be tested or what remains unknown. It defines valid evidence sources and the supported Python, TypeScript, and C++ forms.
Skill Claude CodeCodex
A verification procedure for cases where generated code cannot be compiled on its own. It labels verification as full, partial, or degraded and records which functions could not be compiled.
Skill Claude CodeCodex
A review role that checks whether project outputs are supported by evidence, complete, and internally consistent. It gives review findings but does not change the work's status; a human host decides whether a result is confirmed.
Skill Claude CodeCodex
A maintenance workflow for changing the Anchorlaw protocol or its own implementation. It requires passing tests, self-scanning, synchronized documentation, and disciplined commits.
Skill Claude CodeCodex
A noise-card system for recording runtime failures as structured notes that can later become regression tests, which are tests that prevent the same problem from returning.
Skill Claude CodeCodex
A static code review that scans a directory for defensive programming patterns and classifies findings as errors, warnings, or information.
Skill Claude CodeCodex
An isolated planning role that turns confirmed requirements into a draft implementation specification covering naming, modules, dependencies, interfaces, and framework boundaries.
Skill Claude CodeCodex
A command-line workflow for running anchor verification and reporting each function’s health state.
Skill Claude CodeCodex
An isolated implementation role that builds one assigned code module from an approved specification and plan, adds verification labels, and records self-test results.
Skill Claude CodeCodex
A code-annotation workflow for attaching verification records to public functions. An anchor is a note that points to a test, debugging trace, memory dump, or an honest statement that verification is not yet possible.