guild-agents 1.4.0 → 2.0.0

package/src/templates/skills/new-feature/evals/triggers.json

@@ -1,16 +0,0 @@
-{
-  "skill": "new-feature",
-  "matcherType": "keyword",
-  "description": "Creates branch and scaffold for a new feature",
-  "threshold": 0.3,
-  "tests": [
-    { "prompt": "create a new feature branch", "shouldTrigger": true },
-    { "prompt": "scaffold a new feature", "shouldTrigger": true },
-    { "prompt": "start a new feature called user-auth", "shouldTrigger": true },
-    { "prompt": "prepare the branch for a new feature", "shouldTrigger": true },
-    { "prompt": "review my code", "shouldTrigger": false },
-    { "prompt": "create a pull request", "shouldTrigger": false },
-    { "prompt": "debug this bug", "shouldTrigger": false },
-    { "prompt": "save my session", "shouldTrigger": false }
-  ]
-}

package/src/templates/skills/review/SKILL.md

@@ -1,97 +0,0 @@
----
-name: review
-description: "Standalone code review on the current diff"
-user-invocable: true
-workflow:
-  version: 1
-  steps:
-    - id: gather-diff
-      role: system
-      intent: "Get current git diff (staged or unstaged) and run tests + lint for context."
-      commands: [git diff --staged, git diff, npm test, npm run lint]
-      produces: [diff-content, test-result, lint-result]
-    - id: review
-      role: code-reviewer
-      intent: "Review code quality, patterns, security, and tests. Classify findings by severity."
-      requires: [diff-content, test-result, lint-result]
-      produces: [review-report]
-      model-tier: reasoning
-    - id: present
-      role: system
-      intent: "Present findings organized by severity with final verdict."
-      requires: [review-report]
-      produces: [formatted-report]
-      gate: true
----
-
-# Review
-
-Runs an independent code review on the current project changes. Invokes the Code Reviewer agent to analyze quality, patterns, security, and technical debt.
-
-## When to use
-
-- Before creating a PR
-- To review your own changes before requesting review from others
-- When you want a second opinion on the code you wrote
-
-## Usage
-
-`/review`
-
-## Process
-
-### Step 1 — Get diff and verification state
-
-Get the current changes:
-
-1. First try `git diff --staged` (staged changes)
-2. If there are no staged changes, use `git diff` (unstaged changes)
-3. If there are no changes at all, report that there is nothing to review
-
-Run automated verification to give context to the reviewer:
-
-1. Run project tests (e.g., `npm test`) — capture result
-2. Run project lint (e.g., `npm run lint`) — capture result
-3. Include both results as context for the Code Reviewer
-
-Note: The Code Reviewer does not have access to Bash (only Read, Glob, Grep), so tests and lint are run here before invoking the reviewer.
-
-### Step 2 — Invoke Code Reviewer
-
-Invoke the Code Reviewer agent using Task tool with `model: "opus"` (reasoning tier):
-
-1. Read `.claude/agents/code-reviewer.md` to assume the role
-2. Read CLAUDE.md to understand the project conventions
-3. Receive the full diff + test and lint results from Step 1
-4. If tests or lint failed, this is automatically a Blocker finding
-5. Review the full diff
-6. Classify each finding by severity:
-   - **Blocker**: Must be fixed before merge
-   - **Warning**: Should be fixed, introduces technical debt
-   - **Suggestion**: Optional improvement
-
-### Step 3 — Present findings
-
-Present the report organized by severity:
-
-- Total count of findings by type
-- Detail of each finding: file, description, suggested fix
-- Final verdict: Approved / Approved with warnings / Blocked
-
-If there are blockers, suggest fixing them and running `/review` again.
-
-## Example Session
-
-```text
-User: /review
-
-Reviewing diff: 4 files changed, +127 -34
-Code Reviewer (opus) — Reviewing changes...
-
-Findings:
-- [Warning] src/api/users.js:45 — No input validation on email parameter
-- [Suggestion] src/utils/format.js:12 — Consider using Intl.DateTimeFormat
-- [Blocker] src/db/queries.js:78 — SQL injection vulnerability in raw query
-
-1 blocker, 1 warning, 1 suggestion.
-```

package/src/templates/skills/review/evals/evals.json

@@ -1,43 +0,0 @@
-{
-  "skill": "review",
-  "evals": [
-    {
-      "id": "rv-has-core-steps",
-      "description": "Review has gather-diff, review, and present steps",
-      "expectations": [
-        { "text": "Has gather-diff step", "assertion": "step-exists:gather-diff" },
-        { "text": "Has review step", "assertion": "step-exists:review" },
-        { "text": "Has present step", "assertion": "step-exists:present" }
-      ]
-    },
-    {
-      "id": "rv-reviewer-role",
-      "description": "Review step uses code-reviewer role",
-      "expectations": [
-        { "text": "review uses code-reviewer role", "assertion": "step-role:review:code-reviewer" }
-      ]
-    },
-    {
-      "id": "rv-reasoning-tier",
-      "description": "Code review uses reasoning tier (opus)",
-      "expectations": [
-        { "text": "review uses reasoning tier", "assertion": "step-model-tier:review:reasoning" }
-      ]
-    },
-    {
-      "id": "rv-presentation-gate",
-      "description": "Present step has gate for user review",
-      "expectations": [
-        { "text": "present has gate", "assertion": "gate-exists:present" }
-      ]
-    },
-    {
-      "id": "rv-dependencies",
-      "description": "Review requires diff content, present requires report",
-      "expectations": [
-        { "text": "review requires diff-content", "assertion": "step-requires:review:diff-content" },
-        { "text": "present requires review-report", "assertion": "step-requires:present:review-report" }
-      ]
-    }
-  ]
-}

package/src/templates/skills/review/evals/triggers.json

@@ -1,16 +0,0 @@
-{
-  "skill": "review",
-  "matcherType": "keyword",
-  "description": "Standalone code review on the current diff",
-  "threshold": 0.3,
-  "tests": [
-    { "prompt": "review my code", "shouldTrigger": true },
-    { "prompt": "do a code review on the current changes", "shouldTrigger": true },
-    { "prompt": "check my diff for issues", "shouldTrigger": true },
-    { "prompt": "review the current diff", "shouldTrigger": true },
-    { "prompt": "create a pull request", "shouldTrigger": false },
-    { "prompt": "save my session", "shouldTrigger": false },
-    { "prompt": "what phase am I in", "shouldTrigger": false },
-    { "prompt": "start a new feature", "shouldTrigger": false }
-  ]
-}

package/src/templates/skills/status/SKILL.md

@@ -1,100 +0,0 @@
----
-name: status
-description: "Shows current project and session state"
-user-invocable: true
-workflow:
-  version: 1
-  steps:
-    - id: read-state
-      role: system
-      intent: "Read CLAUDE.md, PROJECT.md, and SESSION.md for project state."
-      commands: [cat CLAUDE.md, cat PROJECT.md, cat SESSION.md]
-      produces: [claude-md, project-md, session-md]
-    - id: scan-resources
-      role: system
-      intent: "List available agents and skills from .claude/ directories."
-      commands: [ls .claude/agents/, ls .claude/skills/]
-      produces: [agent-list, skill-list]
-    - id: present-status
-      role: system
-      intent: "Display project summary: name, stack, session state, agents, skills, and suggested next steps."
-      requires: [project-md, session-md, agent-list, skill-list]
-      produces: [status-display]
-      gate: true
----
-
-# Status
-
-Shows a complete summary of the current project state, the active session, and the available Guild resources.
-
-## When to use
-
-- At the start of a session to get oriented
-- To verify that Guild is correctly configured
-- To see which agents and skills are available
-
-## Usage
-
-`/status`
-
-## Process
-
-### Step 1 — Read state files
-
-Read the Guild configuration files:
-
-- `CLAUDE.md` — project instructions and conventions
-- `PROJECT.md` — identity, stack, and configured agents
-- `SESSION.md` — current session state
-
-### Step 2 — Present state
-
-Show the summary in the following format:
-
-```text
-Guild v1 — [project name]
-
-Current session:
-- Date: [date from SESSION.md]
-- Task in progress: [task or "none"]
-- State: [current state]
-
-Stack:
-- [technologies listed in PROJECT.md]
-
-Available agents:
-- [list of .md files in .claude/agents/]
-
-Available skills:
-- [list of directories in .claude/skills/]
-
-Next steps:
-- [extracted from SESSION.md]
-```
-
-### Step 3 — Suggest actions
-
-If there is no task in progress, suggest:
-
-- `/build-feature` to implement something new
-- `/new-feature` to prepare the environment for a feature
-- `/council` to debate a decision
-
-If there is a task in progress, suggest continuing with the appropriate skill based on the state.
-
-## Example Session
-
-```text
-User: /status
-
-Guild — MyProject
-Stack: Node.js 20, React 18, PostgreSQL
-
-Session: 2026-02-23
-Task: Implementing user preferences
-State: Phase 4 — Developer implementing
-
-Agents: advisor, product-owner, tech-lead, developer, code-reviewer, qa, bugfix, db-migration, platform-expert
-Skills: guild-specialize, build-feature, new-feature, council, qa-cycle, review, dev-flow,
-  status, session-start, session-end
-```

package/src/templates/skills/status/evals/evals.json

@@ -1,40 +0,0 @@
-{
-  "skill": "status",
-  "evals": [
-    {
-      "id": "st-has-core-steps",
-      "description": "Status has read-state, scan-resources, present steps",
-      "expectations": [
-        { "text": "Has read-state step", "assertion": "step-exists:read-state" },
-        { "text": "Has scan-resources step", "assertion": "step-exists:scan-resources" },
-        { "text": "Has present-status step", "assertion": "step-exists:present-status" }
-      ]
-    },
-    {
-      "id": "st-all-system",
-      "description": "All steps are system role",
-      "expectations": [
-        { "text": "read-state is system", "assertion": "step-role:read-state:system" },
-        { "text": "scan-resources is system", "assertion": "step-role:scan-resources:system" },
-        { "text": "present-status is system", "assertion": "step-role:present-status:system" }
-      ]
-    },
-    {
-      "id": "st-presentation-gate",
-      "description": "Present-status has gate",
-      "expectations": [
-        { "text": "present-status has gate", "assertion": "gate-exists:present-status" }
-      ]
-    },
-    {
-      "id": "st-dependencies",
-      "description": "Present-status requires project and session data",
-      "expectations": [
-        { "text": "present-status requires project-md", "assertion": "step-requires:present-status:project-md" },
-        { "text": "present-status requires session-md", "assertion": "step-requires:present-status:session-md" },
-        { "text": "present-status requires agent-list", "assertion": "step-requires:present-status:agent-list" },
-        { "text": "present-status requires skill-list", "assertion": "step-requires:present-status:skill-list" }
-      ]
-    }
-  ]
-}

package/src/templates/skills/status/evals/triggers.json

@@ -1,16 +0,0 @@
-{
-  "skill": "status",
-  "matcherType": "keyword",
-  "description": "Shows current project and session state",
-  "threshold": 0.3,
-  "tests": [
-    { "prompt": "show the project status", "shouldTrigger": true },
-    { "prompt": "show current session state", "shouldTrigger": true },
-    { "prompt": "what is the current project state", "shouldTrigger": true },
-    { "prompt": "how is the project going", "shouldTrigger": true },
-    { "prompt": "create a pull request", "shouldTrigger": false },
-    { "prompt": "review my code", "shouldTrigger": false },
-    { "prompt": "debug this bug", "shouldTrigger": false },
-    { "prompt": "save my session", "shouldTrigger": false }
-  ]
-}

package/src/templates/skills/verify/SKILL.md

@@ -1,114 +0,0 @@
----
-name: verify
-description: "Discipline skill — verification before completion. Use when about to claim work is complete, fixed, or passing, before committing or creating PRs."
-user-invocable: true
----
-
-# Verification Before Completion
-
-Claiming work is complete without verification is dishonesty, not efficiency.
-
-**Core principle:** Evidence before claims, always.
-
-## Usage
-
-`/verify`
-
-Invoke this skill before claiming any work is done, before committing, and before creating PRs.
-
-## When to use
-
-**ALWAYS before:**
-
-- Any success or completion claim
-- Committing, pushing, or creating PRs
-- Moving to the next task
-- Expressing satisfaction about work state
-
-## The Iron Law
-
-```text
-NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE
-```
-
-If you haven't run the verification command in this step, you cannot claim it passes.
-
-## The Gate Function
-
-```text
-BEFORE claiming any status:
-
-1. IDENTIFY: What command proves this claim?
-2. RUN: Execute the FULL command (fresh, complete)
-3. READ: Full output, check exit code, count failures
-4. VERIFY: Does output confirm the claim?
-   - If NO: State actual status with evidence
-   - If YES: State claim WITH evidence
-5. ONLY THEN: Make the claim
-
-Skip any step = lying, not verifying
-```
-
-## What Each Claim Requires
-
-| Claim            | Requires                        | Not Sufficient                 |
-| ---------------- | ------------------------------- | ------------------------------ |
-| Tests pass       | Test command output: 0 failures | Previous run, "should pass"    |
-| Linter clean     | Linter output: 0 errors         | Partial check, extrapolation   |
-| Build succeeds   | Build command: exit 0           | Linter passing, logs look good |
-| Bug fixed        | Test original symptom: passes   | Code changed, assumed fixed    |
-| Regression test  | Red-green cycle verified        | Test passes once               |
-| Requirements met | Line-by-line checklist          | Tests passing                  |
-
-## Red Flags - STOP
-
-- Using "should", "probably", "seems to"
-- Expressing satisfaction before verification ("Great!", "Perfect!", "Done!")
-- About to commit/push/PR without verification
-- Relying on partial verification
-- Thinking "just this once"
-- ANY wording implying success without having run verification
-
-## Common Rationalizations
-
-| Excuse                       | Reality                        |
-| ---------------------------- | ------------------------------ |
-| "Should work now"            | RUN the verification           |
-| "I'm confident"              | Confidence is not evidence     |
-| "Just this once"             | No exceptions                  |
-| "Linter passed"              | Linter is not compiler         |
-| "Partial check is enough"    | Partial proves nothing         |
-
-## Verification Patterns
-
-**Tests:**
-
-```text
-OK:  [Run test command] [See: 34/34 pass] "All tests pass"
-BAD: "Should pass now" / "Looks correct"
-```
-
-**Build:**
-
-```text
-OK:  [Run build] [See: exit 0] "Build passes"
-BAD: "Linter passed" (linter doesn't check compilation)
-```
-
-**Requirements:**
-
-```text
-OK:  Re-read plan -> Create checklist -> Verify each -> Report gaps or completion
-BAD: "Tests pass, phase complete"
-```
-
-## The Bottom Line
-
-Run the command. Read the output. THEN claim the result.
-
-No shortcuts. Non-negotiable.
-
-## Related Skills
-
-- `/tdd` — TDD ensures tests exist before claiming code works
-- `/debug` — systematic debugging when verification reveals failures

package/src/templates/skills/verify/evals/triggers.json

@@ -1,16 +0,0 @@
-{
-  "skill": "verify",
-  "matcherType": "keyword",
-  "description": "Discipline skill — verification before completion. Use when about to claim work is complete, fixed, or passing, before committing or creating PRs.",
-  "threshold": 0.3,
-  "tests": [
-    { "prompt": "verify before committing", "shouldTrigger": true },
-    { "prompt": "run verification before completion", "shouldTrigger": true },
-    { "prompt": "verify the work is complete", "shouldTrigger": true },
-    { "prompt": "make sure everything passes before the PR", "shouldTrigger": true, "keywordExpected": false },
-    { "prompt": "create a pull request", "shouldTrigger": false },
-    { "prompt": "review my code", "shouldTrigger": false },
-    { "prompt": "debug this bug", "shouldTrigger": false },
-    { "prompt": "save my session", "shouldTrigger": false }
-  ]
-}

package/src/utils/accounting.js

@@ -1,139 +0,0 @@
-/**
- * accounting.js — Token usage recording, persistence, and aggregation.
- *
- * Persists usage data to .claude/guild/usage.json.
- */
-
-import { existsSync, readFileSync, writeFileSync, mkdirSync } from 'fs';
-import { join, dirname } from 'path';
-import { estimateCost } from './pricing.js';
-
-const USAGE_PATH = join('.claude', 'guild', 'usage.json');
-
-export function emptyUsage() {
-  return {
-    version: 1,
-    lastUpdated: new Date().toISOString(),
-    entries: [],
-    totals: {
-      totalTokens: 0,
-      totalInputTokens: 0,
-      totalOutputTokens: 0,
-      totalCostUSD: 0,
-      tokensByModel: {},
-      tokensByTier: {},
-      tokensByWorkflow: {},
-      workflowCount: 0,
-    },
-  };
-}
-
-export function createEntry({ workflow, agent, tier, model, inputTokens, outputTokens }) {
-  const totalTokens = inputTokens + outputTokens;
-  return {
-    timestamp: new Date().toISOString(),
-    workflow,
-    agent,
-    tier,
-    model,
-    inputTokens,
-    outputTokens,
-    totalTokens,
-    estimatedCostUSD: estimateCost(model, inputTokens, outputTokens),
-  };
-}
-
-export function loadUsage(root) {
-  const filePath = join(root, USAGE_PATH);
-  if (!existsSync(filePath)) return emptyUsage();
-  try {
-    return JSON.parse(readFileSync(filePath, 'utf8'));
-  } catch {
-    return emptyUsage();
-  }
-}
-
-export function saveUsage(root, usage) {
-  const filePath = join(root, USAGE_PATH);
-  mkdirSync(dirname(filePath), { recursive: true });
-  usage.lastUpdated = new Date().toISOString();
-  writeFileSync(filePath, JSON.stringify(usage, null, 2) + '\n');
-}
-
-function updateTotals(totals, entry) {
-  totals.totalTokens += entry.totalTokens;
-  totals.totalInputTokens += entry.inputTokens;
-  totals.totalOutputTokens += entry.outputTokens;
-  totals.totalCostUSD += entry.estimatedCostUSD;
-  totals.tokensByModel[entry.model] = (totals.tokensByModel[entry.model] || 0) + entry.totalTokens;
-  totals.tokensByTier[entry.tier] = (totals.tokensByTier[entry.tier] || 0) + entry.totalTokens;
-  totals.tokensByWorkflow[entry.workflow] = (totals.tokensByWorkflow[entry.workflow] || 0) + entry.totalTokens;
-  totals.workflowCount += 1;
-}
-
-export function recordStep(root, params) {
-  const usage = loadUsage(root);
-  const entry = createEntry(params);
-  usage.entries.push(entry);
-  updateTotals(usage.totals, entry);
-  saveUsage(root, usage);
-}
-
-const PROFILES = {
-  max: { reasoning: 'claude-opus-4-6', execution: 'claude-sonnet-4-5', routine: 'claude-haiku-4-5' },
-  pro: { reasoning: 'claude-sonnet-4-5', execution: 'claude-sonnet-4-5', routine: 'claude-haiku-4-5' },
-  'all-opus': { reasoning: 'claude-opus-4-6', execution: 'claude-opus-4-6', routine: 'claude-opus-4-6' },
-};
-
-export function aggregate(root, period) {
-  const usage = loadUsage(root);
-  const now = new Date();
-  let cutoff;
-
-  switch (period) {
-    case 'today':
-      cutoff = new Date(now.getFullYear(), now.getMonth(), now.getDate());
-      break;
-    case 'week':
-      cutoff = new Date(now);
-      cutoff.setDate(cutoff.getDate() - 7);
-      break;
-    case 'month':
-      cutoff = new Date(now);
-      cutoff.setDate(cutoff.getDate() - 30);
-      break;
-    default:
-      cutoff = new Date(0);
-  }
-
-  const filtered = usage.entries.filter(e => new Date(e.timestamp) >= cutoff);
-
-  const totals = {
-    totalTokens: 0,
-    totalInputTokens: 0,
-    totalOutputTokens: 0,
-    totalCostUSD: 0,
-    tokensByModel: {},
-    tokensByTier: {},
-    tokensByWorkflow: {},
-    workflowCount: 0,
-  };
-
-  for (const entry of filtered) {
-    updateTotals(totals, entry);
-  }
-
-  return totals;
-}
-
-export function estimateWithProfile(entries, profileName) {
-  const profile = PROFILES[profileName];
-  if (!profile) return 0;
-
-  let cost = 0;
-  for (const entry of entries) {
-    const model = profile[entry.tier] || entry.model;
-    cost += estimateCost(model, entry.inputTokens, entry.outputTokens);
-  }
-  return cost;
-}

package/src/utils/dispatch-protocol.js

@@ -1,74 +0,0 @@
-/**
- * dispatch-protocol.js — Constants and type definitions for the Guild dispatch protocol.
- *
- * Defines the vocabulary shared by dispatch utilities, workflow parser,
- * model routing, and trace modules. Zero dependencies.
- */
-
-/**
- * Valid model tier values. Each tier maps to a class of model capability.
- * @type {readonly ['reasoning', 'execution', 'routine']}
- */
-export const MODEL_TIERS = ['reasoning', 'execution', 'routine'];
-
-/**
- * Valid failure strategy base values for workflow steps.
- * - abort: halt the workflow on failure (default)
- * - continue: skip this step and proceed
- * Additionally, `goto:<step-id>` is valid for redirecting to another step.
- * @type {readonly ['abort', 'continue']}
- */
-export const FAILURE_STRATEGIES = ['abort', 'continue'];
-
-/**
- * Default failure strategy when none is specified.
- * @type {string}
- */
-export const DEFAULT_FAILURE_STRATEGY = 'abort';
-
-/**
- * Default tier assignment for each Guild agent role.
- * Used as fallback when neither the workflow step nor the agent frontmatter
- * specifies a tier.
- * @type {Record<string, string>}
- */
-export const DEFAULT_AGENT_TIERS = {
-  'advisor': 'reasoning',
-  'product-owner': 'reasoning',
-  'tech-lead': 'reasoning',
-  'code-reviewer': 'reasoning',
-  'developer': 'execution',
-  'bugfix': 'execution',
-  'db-migration': 'execution',
-  'qa': 'execution',
-  'platform-expert': 'execution',
-  'learnings-extractor': 'routine',
-};
-
-/**
- * Built-in model profiles mapping tiers to concrete model IDs.
- * @type {Record<string, Record<string, string>>}
- */
-export const DEFAULT_MODEL_PROFILES = {
-  max: {
-    reasoning: 'claude-opus-4-6',
-    execution: 'claude-sonnet-4-6',
-    routine: 'claude-haiku-4-5',
-  },
-  pro: {
-    reasoning: 'claude-sonnet-4-6',
-    execution: 'claude-sonnet-4-6',
-    routine: 'claude-haiku-4-5',
-  },
-};
-
-/**
- * Fallback chain for tier resolution. When a tier's model is unavailable,
- * fall back to the next tier. `null` means no further fallback.
- * @type {Record<string, string|null>}
- */
-export const FALLBACK_CHAIN = {
-  reasoning: 'execution',
-  execution: 'routine',
-  routine: null,
-};