Agent Claude Code
Stress-tests the design + task plan before the human reviews it; reports only problems that survived its own attempts to disprove them. Read-only.
not rated 7
changed 29d ago A 34 tokens
Agent Claude Code
Layering, separation of concerns, API contracts. Read-only.
not rated 7
changed 29d ago A 17 tokens
Agent Claude Code
One exploration pass: similar existing code, what the change touches, where it hooks in.
not rated 7
changed 29d ago A 23 tokens
Agent Claude Code
Logic, race conditions, security. Read-only.
not rated 7
changed 29d ago A 15 tokens
Agent Claude Code
Decides one pipeline gate during an autonomous run (/autopilot, or /sprint’s adjudicated queue) — APPROVE / REWORK / HALT with confidence + cited evidence. Conservative by default; escalates on doubt.
not rated 7
changed 29d ago A 50 tokens
Agent Claude Code
Static design-system audit — token compliance, scale coherence, component duplication, naming, drift vs design-system.md. Read-only.
not rated 7
changed 29d ago A 31 tokens
Agent Claude Code
Codebase health audit for /analyze. Read-only.
not rated 7
changed 29d ago A 17 tokens
Agent Claude Code
API cost + DB perf + latency for /optimize. Read-only.
not rated 7
changed 29d ago A 19 tokens
Agent Claude Code
Runs the full pipeline for one REQ in a worktree for /sprint. No sub-agents. Git per git.mode — feature branch only, never main/force/PR.
not rated 7
changed 29d ago A 42 tokens
Agent Claude Code
Conventions, naming, duplication, test coverage. Read-only.
not rated 7
changed 29d ago A 17 tokens
reflector
11Agent Claude Code
Self-review against captured lessons. Read-only.
not rated 7
changed 29d ago A 12 tokens
Agent Claude Code
Writes code for one task. No git operations.
not rated 7
changed 29d ago A 15 tokens
ui-reviewer
13Agent Claude Code
Runtime UI/UX review — runs the app in a browser to verify render, flows, and design match. Browser mechanism auto-resolves and degrades. Read-only re: source.
not rated 7
changed 29d ago A 41 tokens
analyze
14Skill Claude CodeCodex
Standalone codebase health audit.
not rated 7
changed 29d ago A 9 tokens
architect
15Skill Claude CodeCodex
Design + task breakdown + validate. Phase 2.
not rated 7
changed 29d ago A 13 tokens
autopilot
16Skill Claude CodeCodex
Autonomous pipeline — runs all five phases, routing each gate through the decision-maker; ends in one final human review.
not rated 7
changed 29d ago A 28 tokens
bugfix
17Skill Claude CodeCodex
Slimmer pipeline for bugs.
not rated 7
changed 29d ago A 9 tokens
config
18Skill Claude CodeCodex
View/change .adlc/config.yml settings via guided options; re-sync derived files.
not rated 7
changed 29d ago A 19 tokens
implement
19Skill Claude CodeCodex
Execute the task DAG. Phase 3.
not rated 7
changed 29d ago A 11 tokens
init
20Skill Claude CodeCodex
Bootstrap the .adlc/ vault in a new repo.
not rated 7
changed 29d ago A 14 tokens
optimize
21Skill Claude CodeCodex
Standalone performance/cost scan.
not rated 7
changed 29d ago A 9 tokens
proceed
22Skill Claude CodeCodex
Run all five phase skills with gates between.
not rated 7
changed 29d ago A 12 tokens
recover
23Skill Claude CodeCodex
Reconcile pipeline-state with git reality; back-fill the vault.
not rated 7
changed 29d ago A 15 tokens
review
24Skill Claude CodeCodex
Dispatch reviewers, consolidate findings. Phase 4.
not rated 7
changed 29d ago A 12 tokens
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: