Library

Every public guide, in one place.

A complete catalog grouped by the kind of decision or evidence it supports.

21 entries

Runbooks

Field guides

Interface contracts describe states

Specify loading, ready, no-results, error, transitions, and evidence before implementation.

Read →

Reproducible web acquisition

Separate fetched, rendered, and extracted evidence.

Read →

Agent extension review

Inspect declared capabilities before installation.

Read →

Plan contract mutation lab

Test plan scope, evidence, dependencies, and rollback.

Read →

Application surface coverage

Track applicability, implementation, and verification.

Read →

Repository research records

Keep research traceable and separate from adoption.

Read →

Omarchy source inspection

Pin and hash official source without installing it.

Read →

Model gateway attribution

Retain fallback attempt lineage and estimate provenance.

Read →

Browser interface contract translation

Compare declared CLI and MCP intent without claiming runtime equivalence.

Read →

Spec change lineage

Detect stale plans, tasks, and checks after a requirement changes.

Read →

PR feedback calibration

Separate mechanical review-output checks from human adjudication.

Read →

Replayable coding-agent task packet

Separate declared work, replayed file effects, recorded evidence, and human review.

Read →

A chart is not its source

Keep source facts, transformations, deliverables, and review evidence connected.

Read →

Start with the question, not the monitoring stack

Choose an observation layer from the decision a signal needs to support.

Read →

A backup becomes evidence after a restore

Plan a clean restore record across scope, integrity, recovery material, extraction, and review.

Read →

One source, two documentation surfaces

Review semantic parity across human and agent projections without requiring identical presentation.

Read →

Decision records

Repository research