Agent
Mission: Break a stalled loop. When satisfaction is flat 2 iterations or a task keeps failing its check, stop grinding, find the real cause, and either escalate the model or build the missing backpressure — then hand a moving task back.
Agent
One of wgm's four docs-audit personas — reviews documentation for clarity and onboarding quality from a junior developer's vantage point, reporting findings only.
Agent
One of wgm's four docs-audit personas — reviews documentation for status accuracy, risk visibility, and traceability from a project manager's vantage point, reporting findings only.
Agent
One of wgm's four docs-audit personas — reviews documentation for architectural fit, strategic consistency, and constitution conformance from a principal developer's vantage point, reporting findings only.
Agent
One of wgm's four docs-audit personas — reviews documentation for correctness, completeness, and maintainability from a senior developer's vantage point, reporting findings only.
Agent
Consolidates wgm's four docs-audit personas into one paper-trail report — normalizes feedback, preserves dissent, and classifies every action item strictly as Agent action or Operator action.
Agent
Runs wgm's Grill alignment interview — one question at a time, each with a recommended answer, self-answering from the codebase until goal, success criteria, and constraints are locked.
Agent
Aggregates Hive Growth Loop lessons — anonymizes first, checks consent, de-dups open learning issues, and publishes upstream only when consented.
Agent
Implements exactly one wgm IMPLEMENTATIONPLAN.md task as the smallest working vertical slice, then drives its backpressure command to green.
Agent
Stage 2 of wgm's two-stage review — a high-signal rubber-duck that catches bugs, logic errors, and weak validation, with a binary PASS / CHANGES-REQUESTED verdict.
Agent
Stage 1 of wgm's two-stage review — checks a task's diff against its spec, acceptance criteria, and the constitution, with a binary PASS / CHANGES-REQUESTED verdict.
Agent
Mission: Judge how well a finished slice satisfies its holdout acceptance scenarios — scoring 0–100 by tier — without ever letting a satisfaction score override a failing deterministic check.
Instructions file CodexOpenCode
Instructions for agent-frontier/wgm, covering agents.md, build & run, validate (backpressure), operational notes and codebase patterns.
Skill Claude CodeCodex
Autonomous build skill that turns a rough request into working software through a relentless requirements interview (grill), a persistent plan, and a Ralph-style build loop (analyze → implement → validate → review → record). Use when the user runs /wgm, or asks to build, implement, prototype, or ship a feature or app…
Skill Claude CodeCodex
Companion assessment skill that grills a user on a codebase until their understanding is proven rather than assumed. Asks one question at a time, grades every answer against the real code with citations, scores 0-100 by difficulty tier, and drills the weak areas it finds. Use when the user runs /quiz-me, or asks to be…
Skill Claude CodeCodex
Companion review skill that stress-tests a request, spec, plan, diff, or system against the operators and environment that will actually run it, not an idealized one. Runs a disciplined lifecycle (Triage -> Context -> Bottleneck decomposition -> Simplify -> Field test -> Verdict) and emits exactly one of three…
Skill Claude CodeCodex
Companion learning skill that makes a codebase legible fast. Surveys a repository, builds a cited map, runs a guided tour in dependency order, and finishes with a validated first change so the learner has actually moved the code. Use when the user runs /teach-me, or asks to learn, understand, onboard onto, get up to…