Meta¶
The ways in hooks/ways/meta/.
| Way | What it covers |
|---|---|
| Codify What Worked | The user sets a standing rule (from now on, always do this), or corrects the same preference of theirs again, or a process that took several tries finally worked. Write the rule or the working method into the home that will enforce it, apply it now, and record the attempts that failed. |
| Compaction Checkpoint | |
| Deployment | How agent-ways itself deploys into the home config dir — ~/.claude as a thin projection of an XDG application (source in $XDG_DATA_HOME/agent-ways), how the agent-ways installer/update/ways manage repair work under it, how to spot a legacy pre-1.0 in-place agent-ways clone that must ways migrate instead of pull, and where the migrator lives now that 1.9.0 removed it from the binary — surfaced only when installing, updating, migrating, or reconciling agent-ways itself, or resolving an existing ~/.claude conflict during agent-ways setup |
| Developing | How we carry a piece of work through the development loop — a variable front (design, prototype, ADR, ordered by where the uncertainty lives) and a stable tail (build, review, fix, merge). Route to the /develop skill, which picks the shape and borrows the stage skills rather than reimplementing them. |
| Goals | When and why to set a /goal — running Claude in goal mode toward a completion condition, and the regime that creates |
| Governance Citation | Grounding recommendations in real governance controls — when and how to cite the standards (NIST, OWASP, ISO, SOC 2, CIS, IEEE) behind a practice |
| Introspection | opening a pull request as the moment to look back over the session, asking whether the human corrected, explained, or pushed back on something that should become a way, a skill, or a workflow |
| Knowledge | Overview of the ways system — how ways, skills, and hooks relate, domain organization, matching modes |
| Memory | Persistent memory system — MEMORY.md, topic files, what to record and when |
| Presenting Choices | deciding whether a call goes to the operator or is settled by the agent, the waterline for technical calls that set direction, and presenting genuine decisions as explicit choices rather than burying options in prose or deciding silently |
| Relational Trust | Relational trust between Claude and the human as mutual damping — borrowed resources, asymmetric consequences, governed feedback circuit, trust as a degradable spectrum not a binary gate |
| Skills | How we write Claude Code skills in this repo — skill naming and scope conventions, the global-scope caveat, and choosing a skill as the authoring surface for a procedure; defers SKILL.md mechanics to the official docs |
| Starting | Recognize the beginning of a working session and route to the /start skill — orient to where work was left off, run a structured interview on a greenfield repo, check the context gauge to confirm it really is session-start, then recommend planning with context already warm. |
| Structured Thinking | structured reasoning, thinking frameworks, cognitive scaffolding for complex decisions |
| Sub-Agents | Sub-agent delegation — when to spawn a specialized sub-agent, writing the brief so each constraint keeps its stated strength, the shape the worker reports back, and how deep delegation goes |
| Task List Checkpoint | |
| Workflows | When to reach for the Workflow tool — deterministic multi-agent orchestration (fan-out, staged pipelines, verification and synthesis across many items) and how far up the substrate ladder a task warrants climbing |
| Wrapping Up | Recognize an end-of-session signal and route to the /wrap skill — square the TaskList honestly, write a continuation prompt, and hand off to a gauge-aware directed compaction. |