Drewgent configuration for opencode — agent profiles, skills, scripts, and tooling
humanerd-drew/opencode-drewgent
Agent
Data analysis agent. Queries kanban DB, git log, knowledge.db for patterns, trends, and insights. Produces structured reports. Read-only on production data.
humanerd-drew/opencode-drewgent
Agent
Documentation and record-keeping agent. Writes changelogs, updates docs, saves kanban completion summaries. Does NOT implement features.
humanerd-drew/opencode-drewgent
Agent
CMO-style content agent. Observes recent work activity (sessions, kanban, git), identifies narrative-worthy material, produces blog + X thread + LinkedIn drafts. Maintains narrative arc continuity.
humanerd-drew/opencode-drewgent
Agent
Content editing and QA agent. Reviews drafts for tone, voice, clarity, and language quality. Final quality gate before publishing.
humanerd-drew/opencode-drewgent
Agent
Read-only research and exploration agent. Gathers context, traces call chains, reports patterns. Cannot modify files. ESCALATE capability for hard problems.
humanerd-drew/opencode-drewgent
Agent
Implementation agent. Writes code, creates files, applies patches. Follows project conventions. Verifies with lint/test after changes.
humanerd-drew/opencode-drewgent
Agent
Work orchestration and pipeline management. Assigns tasks across agent profiles, tracks progress, resolves blockers. Delegates — does NOT implement.
humanerd-drew/opencode-drewgent
Agent
Task decomposition and planning agent. Breaks complex goals into atomic, actionable tasks with dependency ordering. Produces structured plans only.
humanerd-drew/opencode-drewgent
Agent
Critical code review for architecturally significant changes. Architecture integrity, abstraction boundaries, performance, migration strategy. Second set of eyes after standard reviewer.
humanerd-drew/opencode-drewgent
Agent
Code review agent. Reviews changes for logic errors, edge cases, style, security, testing, and over-engineering. Does NOT modify files.
humanerd-drew/opencode-drewgent
Agent
Security-focused code review. Audits for injection, crypto misuse, auth bypass, privilege escalation, secret leakage. Blocks release on CRITICAL findings.
humanerd-drew/opencode-drewgent
Agent
Site Reliability Engineering agent. Manages services, deployment pipelines, cron jobs, and incident response. Read-only first.
humanerd-drew/opencode-drewgent
Agent
Test agent. Writes and runs tests for implemented code. Verifies correctness before review. Max 2 fix cycles before reporting unrecoverable failure.