LLM Workflow

This workflow defines the execution sequence for tasks in this project.

1. Load Project Rules

Always load these files in this order before executing a task:

Also:

Load only the feature files explicitly referenced by the task or by a directly referenced feature dependency.
Do not load unrelated feature files.
Load a language-specific flavor file only when the task targets that language.
Load only the files explicitly referenced by the task and the files required by those references.
Use TODO.md only for status updates when the task explicitly requests them.
Do not infer the requested work from the first unchecked TODO item.
The phrase Execute the next TODO task explicitly requests TODO-driven execution.

Determine the requested operations, target files, constraints, and output format.
Treat TASK-DESCRIPTION as the requested work.
Treat TASK-OUTPUT as the response representation.
Treat TASK-CONTEXT as information that does not add instructions unless explicitly labeled as a constraint.
Treat TASK-FILES as scope information that does not authorize modifications by itself.
For every file operation, resolve the exact path before implementation.
For a create operation, verify that the target path is authorized by the task or by an applicable project rule.
For a modify, delete, or rename operation, verify that the referenced path exists or report that it is missing.
If an exact path cannot be resolved unambiguously, stop and ask for clarification.
Ask clarification questions before producing implementation output if any required detail is ambiguous or missing.
Internally restate the task in one concise sentence.

Treat a referenced feature file as authoritative requirements for the current task.
Do not apply unreferenced feature files.
Treat a task file as a detailed task description and apply its TASK-DESCRIPTION, TASK-OUTPUT, TASK-CONTEXT, and TASK-FILES sections according to prompts/how-to-write-tasks.md.
When a task conflicts with a referenced feature, ask for clarification unless the task explicitly overrides the feature requirement.

Create a concise internal plan before implementation.
The internal plan must identify the applicable requirements, target files, required validation, and output order.
The plan must not appear in the response unless the requested output format includes it.

For executable source-code changes, create or update tests before implementation code.
For prompt, documentation, configuration, or build-script changes, add tests only when an applicable test mechanism exists or the task requests tests.
Tests for source code belong in tests/.
Tests for scripts belong in tests/ and should validate the script behavior without placing generated output in source directories.
Prompt validation belongs in tests/ when a prompt validation mechanism exists or the task requests prompt validation.
Do not create a test runner solely to satisfy this policy unless the task requests one.

Classify every test into one tier:

Portable — runs anywhere the repository is checked out; belongs in tests/ and runs under make test.
Tool-gated — needs a prerequisite tool or dependency; skips cleanly with a message when the prerequisite is absent.
Live-state — verifies a documented model against live reality (a network, devices, a running system); requires declared access (hosts, keys, credentials); reports PASS/WARN/FAIL and exits nonzero only on failure.

Rules:

A live-state test declares its prerequisites in its header comment, including where it must run and which keys or access it needs.
A live-state test belongs in tests/ only when the repository lives in the target environment; otherwise it is run by an explicit mechanism outside make test.
A live-state test must not pass silently when it could not check anything; it reports WARN or FAIL.
Use PASS/WARN/FAIL classification: WARN for environment-dependent conditions, FAIL for broken invariants.

Modify only files within the declared task scope.
Do not perform opportunistic refactoring.
Do not reformat unrelated lines.
Do not update dependencies, generated files, or documentation unless requested or required for correctness.
Apply feature-specific requirements only when the feature is referenced by the task or by a directly referenced feature dependency.
Do not create a file merely because its directory is available.
Do not create a second file to replace or supplement an existing file unless the task explicitly requests both files.
Preserve exact casing, separators, numbering, and extensions from the resolved path.

When execution tools are available, run make test from the repository root.
Never claim that tests passed unless make test was actually executed successfully.
If make test cannot be run, report that verification was not performed when verification results are requested.
If make test fails, correct the failure within task scope and run make test again.
If the failure cannot be corrected within task scope, stop, report the failure, and do not claim completion.

Update TODO.md only when the task explicitly requests a TODO update or completes a TODO item.
Do not modify TODO.md as a side effect of unrelated work.
Mark a TODO item complete only after the requested verification succeeds.

When the task changed files and verification succeeds, commit the completed work with git.
Create one commit containing only the task’s files and any TODO update made for the task.
Use the commit-message conventions in prompts/03-conventions.md.
Do not commit generated output, logs, or unrelated files.
Skip this step when the task changed no files.

A task is complete only when every applicable item is satisfied:

Produce output only in the format specified by the task.
Do not include the internal restatement or plan unless requested.
Do not mix clarification, planning, implementation, and verification output.
When multiple files are requested, produce complete files in the specified order.