Skip to content

Latest commit

 

History

History
498 lines (309 loc) · 11.6 KB

File metadata and controls

498 lines (309 loc) · 11.6 KB

Glossary

Every term used in Knowledge & Vibes, defined


Navigation

Section What's Covered
Core Concepts Beads, North Star, Lossless, Grounding, Calibration, Drift
Requirements & Planning REQ, AC, ADR, Rigor Tier, Priorities, Spikes, Gates
Execution TDD, 3-Iteration Limit, ADaPT, Verification
Multi-Agent Coordination Agent Mail, File Reservations, Claims, Orchestrator Pattern
Tools bd, bv, ubs, cass, cm, MCP, Warp-Grep, Exa
Slash Commands /prime, /advance, /calibrate, /decompose, /ground
Phrases Common expressions and their meaning

Core Concepts

Discovery

The pre-pipeline phase where you surface every decision hiding in your idea through curiosity-driven interrogation of frontier models. Discovery happens before formal planning begins.

The plan is complete when there's nothing left to interpret.

See docs/workflow/DISCOVERY.md.

Beads

The unit of work in this system. A bead is a task with a description, status, priority, dependencies, and verification requirements. Managed with the bd command.

Beads can have sub-beads (e.g., bd-123.1, bd-123.2). Sub-beads must complete before their parent closes.

North Star Card

A one-page document that anchors the entire project. Contains: goal, success metrics, rigor tier, non-goals, and stop/ask rules.

When in doubt, check the North Star. Everything traces back to it.

Lossless

A plan or spec where nothing is left to interpretation. Another agent could implement it without asking clarifying questions.

"Lossless" means no guessing required, not "exhaustively long."

Grounding

Verifying claims against real sources before acting on them. Three types:

Type Question Tool
Codebase truth What does the actual code say? Warp-Grep
Web truth What do current docs say? Exa
History truth What worked before? CASS, cm

Research: research/008-api-hallucination.md, research/017-context-retrieval-repo-editing.md

Calibration

A hard stop between phases to check alignment. Detects drift, resolves disagreements, updates the plan.

Triggered by /calibrate. Disagreements are resolved by tests, not debate.

Research: research/003-debate-or-vote.md, research/041-debatecoder.md

Drift

When implementation gradually diverges from the original plan. Causes: long context, forgotten requirements, accumulated decisions.

Calibration catches drift before it compounds.

Research: research/004-context-length-hurts.md


Requirements & Planning

REQ (Requirement)

A numbered requirement describing what the system must do. Format: REQ-001, REQ-002.

Requirements describe outcomes, not solutions.

Example: REQ-001: Users can reset their password via email

AC (Acceptance Criteria)

Testable conditions that prove a requirement is met. Format: AC-001.1, AC-001.2.

If you can't test it, it's not an AC. It's a hope.

Example: AC-001.1: Password reset email arrives within 30 seconds

ADR (Architecture Decision Record)

A log entry documenting a significant decision. Contains:

  • What was decided
  • Alternatives considered
  • Why this option was chosen
  • What would change our mind (reversal triggers)

Rigor Tier

How much process a project needs:

Tier Use For Testing
Tier 1 (Speed) Prototypes, experiments Smoke tests only
Tier 2 (Balanced) Most projects Unit + integration tests, ADRs
Tier 3 (High Assurance) Regulated, high-stakes Full test matrix, threat modeling

P0 / P1 / P2 (Priority)

Level Meaning
P0 Must have. Project fails without it.
P1 Should have. Important but not blocking.
P2 Nice to have. Do if time permits.

Spike

A timeboxed investigation to reduce uncertainty. Run before committing to an approach.

Example: "Spike: Can we integrate with their API?" (output: yes/no + notes)

Gate

A checkpoint that must pass before proceeding:

Gate Requirement
Security gate ubs --staged clean
Tests gate All tests pass
Calibration gate Drift addressed

Execution

TDD (Test-Driven Development)

Writing tests before implementation. In this system, TDD is mandatory. Tests go in the bead description before code is written.

Research: research/054-tdd-ai-code-gen.md

3-Iteration Limit

Never attempt more than 3 repair cycles on failing code. After 3 failures: stop, decompose, or escalate.

Research: research/060-debugging-decay-index.md (DDI shows 60-80% effectiveness loss by attempt 3)

ADaPT (Adaptive Decomposition)

Only break tasks into smaller pieces when execution fails. Don't pre-decompose based on guesses.

Flow: Start coarse → Attempt execution → If fails after 3 tries → Decompose only the failing part

Research: research/038-adapt.md

Verification

Evidence that a bead is complete:

  • Tests pass
  • Security scan clean
  • Traceability updated
  • Commands and outputs recorded

"It looks right" is not verification.


Multi-Agent Coordination

Agent Mail

The coordination system for multiple AI agents. Provides:

  • Agent registration
  • Messaging between agents
  • File reservations
  • Discovery of active agents

Agent Name

A randomly generated identifier (e.g., "BlueLake", "GreenCastle"). Assigned at registration.

Names are memorable identifiers, not role descriptions.

File Reservation

A lock on files or directories that prevents other agents from editing simultaneously. Set before editing, released when done.

Prevents merge conflicts and duplicate work.

[CLAIMED] / [CLOSED]

Message subjects announcing when an agent starts or finishes a task:

Message Meaning
[CLAIMED] bd-123 - Task title Work started
[CLOSED] bd-123 - Task title Work completed

All agents see these to avoid stepping on each other.

Orchestrator-Worker Pattern

A coordination pattern where one agent (orchestrator) assigns work to others (workers) and synthesizes results.

Research: research/056-multi-agent-orchestrator.md, research/059-multi-agent-orchestrator-2025.md


Tools

bd (Beads CLI)

Command-line tool for task management.

bd create "Task name"      # Create a bead
bd ready --json            # See unblocked beads
bd update ID --status ...  # Update a bead
bd close ID --reason "..." # Close a bead
bd dep add child blocker   # Add dependency

bv (Beads Viewer)

Analyzes the bead graph and recommends next tasks.

bv --robot-next            # Single best next task
bv --robot-triage          # Full analysis
bv --robot-alerts          # Risks and stale items

Always use --robot-* flags. Bare bv hangs.

ubs (Universal Bug Scanner)

Security scanner that catches vulnerabilities before commit.

ubs --staged               # Scan staged changes
ubs --staged --fail-on-warning  # Strict mode

Mandatory before every commit.

cass (Coding Agent Session Search)

Searches past Claude Code sessions to find similar solutions.

cass search "query" --robot
cass timeline --today --json

Always use --robot or --json. Bare cass hangs.

cm (Context Memory)

Retrieves learned patterns and anti-patterns for your current task.

cm context "what I'm doing" --json

MCP (Model Context Protocol)

A standard for giving AI assistants additional capabilities. MCP servers provide tools to Claude Code.

Warp-Grep

MCP tool for fast parallel codebase search. Use to understand how things work across multiple files.

Exa

MCP tool for web search. Use to verify documentation, find best practices, research libraries.


Slash Commands

Command Purpose
/prime Start a new session. Registers with Agent Mail, checks inbox, discovers tasks.
/advance Full bead lifecycle: close current, discover, claim, reserve files, work (TDD-first), announce.
/calibrate Run a calibration checkpoint. Checks coverage, detects drift, resolves disagreements with tests.
/decompose Break a phase into beads and sub-beads. Creates the task graph for execution.
/ground Verify external libraries and APIs against current documentation.
/recall Retrieve context from past sessions (CASS search + cross-agent learning).
/explore Parallel codebase search for understanding how things work.
/verify Security scanning with UBS before commits.
/resolve Resolve disagreements between agents or approaches using test-based adjudication.
/release Run pre-ship checklist for release readiness. Verifies tests, security, documentation.

Phrases

"Give AI few decisions"

Plan as much as appropriate. Give the AI as few decisions as possible. Any decision you don't claim, you implicitly delegate.

"Nothing left to interpret"

The plan is complete when there's nothing left to interpret. If an agent would need to ask a clarifying question, you're not done planning.

"Bounded. Complete. Verified."

The three rules of lossless decomposition:

  • Bounded = scoped to what an agent can hold in working memory
  • Complete = every detail survives the decomposition
  • Verified = audited by someone who wasn't there

"Gaps become assumptions"

Gaps become assumptions. Assumptions become architecture. Bad architecture becomes a rewrite. This is why discovery matters.

"Same hours. Different outcomes."

The time gets spent either way. You either spend it upfront thinking through decisions, or later untangling decisions the model made without you.

"Truth lives outside the model"

AI output is not truth. Truth is tests that pass, code that compiles, behavior you can observe.

"Tests adjudicate, not rhetoric"

When agents disagree, they write tests that distinguish the approaches. Test results decide, not arguments.

"External feedback only"

Don't ask "are you sure?" That makes the AI second-guess correct answers. Instead, run tests and use results as feedback.

"Stop/ask rule"

A predefined trigger for when the AI should pause and ask the user. Defined in the North Star Card.

"Context sufficiency"

Checking whether you have enough information before starting. If you'd have to guess, gather more context first.

"Workflow beats prompting"

You can't prompt your way to reliability. You need a system where failures get caught before they matter.