Decisions

  • Pending: Which agent to build first? (researcher is likely most immediately useful)
  • Pending: Agent format — own CLAUDE.md-style instructions or just prompt templates?
  • Pending: How do agents interact with skills? (e.g., briefing-compiler used by /briefing)
  • Pending: Per-agent memory — dedicated memory files per agent or shared memory pool?
  • Pending: Persistent sessions — always-on server process or session-start preload from memory files?
  • Pending: Reviewer agent — hard gate (blocks build until approved) or advisory (warnings only)?
  • Pending: Features specialist — auto-generate full spec from one-liner or require user outline?

User Tasks


Summary

Build specialized sub-agents that handle focused tasks autonomously within Opus.

Problem / Motivation

Many Opus tasks require deep, focused work in a specific domain (research, code review, building features, compiling briefings). Currently these are done ad-hoc with generic Task tool calls. Custom agent definitions would standardize prompts, tools, and behavior for recurring specialized work.

Proposed Solution

Define and implement custom agents as reusable configurations. Each agent gets a definition file in .claude/agents/ with purpose, tools, prompt template, and usage conditions. Agents are invoked via the Task tool from skills or manual calls.

Planned Agents:

AgentPurpose
researcherGather information from web and codebase for deep exploration and analysis
reviewerQuality gate — reviews implementation plans before build, checks conventions, catches edge cases
builderPick up planned features, implement them, update status on completion
briefing-compilerCompile daily briefings from vault state, git history, and external data
featuresSpec designer — reads raw ideas and creates full FR documents with AI-generated phases, decisions, test plans, and prerequisites

Open Questions

1. First Agent Choice

Question: Which agent to build first?

OptionDescription
A) researcherMost immediately useful for exploration tasks
B) reviewerUseful for quality checks
C) builderUseful for implementation

Recommendation: Option A — researcher is the most broadly useful starting point.

Decision:

2. Agent Definition Format

Question: Should agents have their own CLAUDE.md-style instructions or just prompt templates?

OptionDescription
A) Prompt templatesSimple, lightweight, easy to iterate
B) Full instruction filesMore powerful, more maintenance

Recommendation: Option A — start simple, upgrade if needed.

Decision:


Phase Overview

PhaseDescriptionStatus
Phase 1Build first agent (researcher)
Phase 2Build remaining agents
Phase 3Persistent sessions & per-agent memory

Phase 1: First Agent (researcher) —

Goal: Establish agent definition format and build the first working agent.

File / FeatureDetailsOwnerStatus
Agent definition formatDecide on structure for agent filesmv
Build researcher agentDefinition + prompt template in .claude/agents/opus
Test researcherVerify it works via Task tool in a real taskmv
Update agents READMEDocument the built agentopus

Phase 2: Remaining Agents —

Goal: Build the remaining planned agents and integrate them with skills.

File / FeatureDetailsOwnerStatus
Build reviewer agentCode/vault review specializationmv
Build builder agentFeature implementation specializationopus
Build briefing-compilerDaily briefing specializationopus
Integrate with skillsWire agents into relevant slash commandsopus

Phase 3: Persistent Sessions & Per-Agent Memory —

Goal: Enable agents to accumulate knowledge across invocations, preserving learnings and context between sessions.

Persistent Sessions: Instead of starting cold each time, each specialist agent:

  • Keeps conversation history from previous tasks in its domain
  • Writes learnings to persistent memory files
  • Reads path-scoped rules that act as “crystallized knowledge” from previous sessions
  • On context overflow, key learnings are compressed but preserved

Per-Agent Memory: Each agent gets its own memory file (e.g., .claude/agents/memory/researcher.md) storing:

  • File paths it frequently works with
  • Patterns learned about the codebase
  • Common pitfalls in its domain
  • User preferences specific to its specialty

Reviewer Agent (Quality Gate): A dedicated reviewer agent that evaluates plans before implementation:

  • Reviews implementation plans against project conventions
  • Checks for missed edge cases, security issues, performance concerns
  • Can reject a plan and send it back for revision
  • Runs as a quality gate in the pipeline strategy (Plan Review Build)
File / FeatureDetailsOwnerStatus
.claude/agents/memory/ directoryPer-agent memory directory structureopus
Memory file formatDefine standard format for agent memory files (frontmatter + sections)opus
Session preload mechanismOn agent start, load relevant memory file into contextopus
Memory write-backAfter task completion, agent writes new learnings to its memory fileopus
Context compressionWhen memory file exceeds context budget, compress while preserving key insightsopus
Reviewer agent definitionBuild reviewer agent with quality gate behaviormv
Pipeline integrationWire reviewer into Plan Review Build flow for medium-complexity featuresmv

Prerequisites / Gap Analysis

Requirements

RequirementDescription
REQ-1.claude/agents/ directory and README

Current State

ComponentStatusDetails
.claude/agents/ directorydoneExists with README
Agent definitionsOnly planned, none built

Gap (What’s missing?)

GapEffortBlocker?
Agent definition format decisionLowNo
First agent implementationMedNo

Test

Manual tests

TestExpectedActualLast
pending-

AI-verified tests

ScenarioExpected behaviorVerification method

E2E tests

ScenarioAssertion

Integration tests

ComponentCoverage

Unit tests

ComponentTestsCoverage

History

DateEventDetails
2026-02-27CreatedMoved planned agents from .claude/agents/README.md to FR
2026-02-27RenumberedFrom FR-024 to FR-043
2026-02-28ReformattedAligned to feature-request template
2026-03-04ExpandedAdded persistent sessions, per-agent memory, reviewer agent (inspired by Nexie)
2026-03-04UpdatedAdded features specialist agent (inspired by Nexie’s spec designer)

References

  • .claude/agents/README.md — agent ideas originally listed here
  • FR-037 (Daily Briefings) — briefing-compiler agent
  • FR-045 (Complexity Routing) — routes work to agents based on complexity score
  • FR-046 (Job Registry & Priority Queue) — dispatches jobs to available agents