ai-consultants 2.10.1 → 2.10.3

package/CLAUDE.md

@@ -6,7 +6,7 @@ AI Consultants is a multi-model AI deliberation system that queries up to 15 AI
 
 **Self-Exclusion**: The invoking agent is automatically excluded from the panel to prevent self-consultation. Claude Code won't query Claude, Codex CLI won't query Codex, etc.
 
-**Version**: 2.10.1
+**Version**: 2.10.3
 
 ## Distribution
 
@@ -63,6 +63,12 @@ ai-consultants/
 ├── references/
 │   ├── configuration.md      # Full configuration reference
 │   └── details.md            # Presets, strategies, best practices
+├── docs/
+│   ├── releases/             # Release notes (one per version)
+│   ├── SETUP.md              # Installation guide
+│   ├── COST_RATES.md         # Model pricing
+│   ├── SMART_ROUTING.md      # Affinity matrix
+│   └── JSON_SCHEMA.md        # Output schema
 └── templates/
     └── synthesis_prompt.md     # Synthesis prompt
 ```
@@ -452,11 +458,11 @@ All consultants now use premium models by default:
 | Cursor | composer-1.5 |
 | DeepSeek | deepseek-reasoner |
 | GLM | glm-5 |
-| Grok | grok-4-1-fast-reasoning |
+| Grok | grok-4.20-0309-reasoning |
 | Qwen3 | qwen3.5-plus |
 | Kimi | kimi-code/kimi-for-coding |
 | Aider | gpt-5.3-codex |
-| MiniMax | MiniMax-M2.5 |
+| MiniMax | MiniMax-M2.7 |
 | Kilo | auto |
 | Ollama | qwen2.5-coder:32b |
 
@@ -667,7 +673,7 @@ for f in scripts/*.sh scripts/lib/*.sh; do bash -n "$f" && echo "OK: $f"; done
 | `KIMI_MODEL` | kimi-code/kimi-for-coding | Kimi model (v2.9) |
 | `ENABLE_MINIMAX` | false | Enable MiniMax consultant (v2.10) |
 | `MINIMAX_API_KEY` | - | MiniMax API key (v2.10) |
-| `MINIMAX_MODEL` | MiniMax-M2.5 | MiniMax model (v2.10) |
+| `MINIMAX_MODEL` | MiniMax-M2.7 | MiniMax model (v2.10) |
 
 ## External Dependencies
 
@@ -722,8 +728,88 @@ For detailed information, see:
 - Use `.env.example` as template for environment configuration
 - Run `./scripts/doctor.sh` to verify configuration
 
+## Release Process
+
+Every version bump **must** include a release note in `docs/releases/v<VERSION>.md`. Use the template below.
+
+### Steps
+
+1. Update version in all files: `package.json`, `SKILL.md` (frontmatter + title), `README.md` (title + badge), `CLAUDE.md` (`**Version**`), `docs/cost_rates.json`, `docs/COST_RATES.md` (title)
+2. Add changelog entry in CLAUDE.md under `## Changelog`
+3. Create `docs/releases/v<VERSION>.md` using the template below
+4. Commit, push, and optionally create a GitHub release pointing to the release note
+
+### Release Note Template
+
+```markdown
+# Release v<VERSION>
+
+**Date:** <YYYY-MM-DD>
+**Type:** <Major | Minor | Patch> — <one-line summary>
+**Previous:** v<PREVIOUS_VERSION>
+
+## Highlights
+
+- <3-5 bullet points with the most impactful changes>
+
+## What's New
+
+### <Feature/Area 1>
+
+<Description of what changed and why. Include tables for multi-item changes.>
+
+### <Feature/Area 2>
+
+<Description.>
+
+## Breaking Changes
+
+<List breaking changes, or "None" if backwards-compatible.>
+
+## Upgrade Guide
+
+\`\`\`bash
+# If installed via git clone
+cd ~/.claude/skills/ai-consultants && git pull
+
+# If installed via curl | bash
+curl -fsSL https://raw.githubusercontent.com/matteoscurati/ai-consultants/main/scripts/install.sh | bash
+\`\`\`
+
+<Note any configuration changes required, or "No configuration changes required.">
+
+## Commits
+
+- \`<hash>\` <commit message>
+
+## Contributors
+
+- <list>
+```
+
+### Guidelines
+
+- **Audience**: Developers who use the skill. Write for someone who hasn't seen the commits.
+- **Highlights first**: Lead with impact, not implementation. "Debate rounds no longer crash on Linux" > "Added `|| true` to `((count++))` in debate_round.sh".
+- **Breaking changes prominent**: If any exist, they go in a dedicated section — never buried in a bullet list.
+- **Quantify when possible**: Token savings percentages, issue counts, file counts.
+- **Upgrade guide always present**: Even if the answer is "just pull", make it explicit.
+- **No internal jargon**: Avoid referencing issue tracker IDs or internal codenames without context.
+
 ## Changelog
 
+### v2.10.3
+- Grok premium model upgraded to `grok-4.20-0309-reasoning` (replaces `grok-4-1-fast-reasoning`)
+- GLM API endpoint migrated from `open.bigmodel.cn` to `api.z.ai/api/coding/paas/v4`
+- Removed non-functional MiniMax highspeed models (`MiniMax-M2.7-highspeed`, `MiniMax-M2.5-highspeed`)
+- Removed legacy `grok-beta` from cost catalog
+- Fixed GLM URL fallback in `common.sh` and `configure.sh` (were still using old endpoint)
+- Fixed duplicate `minimax-m2.5` entry with conflicting rates in `cost_rates.json`
+
+### v2.10.2
+- MiniMax M2.7 upgrade: premium/standard now use MiniMax-M2.7, economy uses MiniMax-M2.5
+- Model tiers: premium (MiniMax-M2.7), standard (MiniMax-M2.7), economy (MiniMax-M2.5)
+
 ### v2.10.1
 - Slash command quality improvements: file context handling, result presentation templates, error recovery guidance
 - debate_round.sh hardening: Amp/Kimi/MiniMax case entries, `((count++)) || true` fixes, `*` default case, stderr to `.err` files, ROUND_NUMBER validation
@@ -734,7 +820,7 @@ For detailed information, see:
 - MiniMax M2.5 API support via OpenAI-compatible endpoint
 - New consultant: MiniMax with "The Pragmatic Optimizer" persona (ID: 21)
 - New environment variables: `ENABLE_MINIMAX`, `MINIMAX_API_KEY`, `MINIMAX_MODEL`, `MINIMAX_API_URL`
-- Model tiers: premium (MiniMax-M2.5), standard (MiniMax-M2.1), economy (MiniMax-M2.5-highspeed)
+- Model tiers: premium (MiniMax-M2.5), standard (MiniMax-M2.1), economy (MiniMax-M2.5)
 - npx distribution: `npx ai-consultants "question"` (zero dependencies)
 - New `bin/ai-consultants` wrapper with symlink resolution and subcommand routing
 - Now supports 15 consultants total

package/README.md

@@ -1,8 +1,8 @@
-# AI Consultants v2.10.1
+# AI Consultants v2.10.3
 
 > Query multiple AI models simultaneously for expert opinions on coding questions. Get diverse perspectives, automatic synthesis, confidence-weighted recommendations, and multi-agent debate.
 
-[![Version](https://img.shields.io/badge/version-2.10.1-blue.svg)](https://github.com/matteoscurati/ai-consultants)
+[![Version](https://img.shields.io/badge/version-2.10.3-blue.svg)](https://github.com/matteoscurati/ai-consultants)
 [![License](https://img.shields.io/badge/license-MIT-green.svg)](LICENSE)
 [![Claude Code Skill](https://img.shields.io/badge/Claude%20Code-Skill-orange.svg)](https://docs.anthropic.com/en/docs/claude-code/skills)
 [![GitHub stars](https://img.shields.io/github/stars/matteoscurati/ai-consultants?style=social)](https://github.com/matteoscurati/ai-consultants)
@@ -421,9 +421,9 @@ INVOKING_AGENT=codex ./scripts/consult_all.sh "Question"    # Codex excluded
 |------------|---------------|---------|-------|
 | **Qwen3** | qwen3.5-plus | The Analyst | Data-driven analysis |
 | **GLM** | glm-5 | The Methodologist | Structured approaches |
-| **Grok** | grok-4-1-fast-reasoning | The Provocateur | Challenge conventions |
+| **Grok** | grok-4.20-0309-reasoning | The Provocateur | Challenge conventions |
 | **DeepSeek** | deepseek-reasoner | The Code Specialist | Algorithms, code generation |
-| **MiniMax** | MiniMax-M2.5 | The Pragmatic Optimizer | Performance, efficiency, pragmatism |
+| **MiniMax** | MiniMax-M2.7 | The Pragmatic Optimizer | Performance, efficiency, pragmatism |
 
 ### Local Consultants
 
@@ -478,10 +478,10 @@ Choose the right balance of quality, speed, and cost with model quality tiers.
 | Cursor | composer-1.5 | composer-1.5 | gemini-2.0-flash |
 | DeepSeek | deepseek-reasoner | deepseek-v3.2 | deepseek-chat |
 | GLM | glm-5 | glm-5 | glm-4-flash |
-| Grok | grok-4-1-fast-reasoning | grok-3 | grok-3-mini |
+| Grok | grok-4.20-0309-reasoning | grok-3 | grok-3-mini |
 | Qwen3 | qwen3.5-plus | qwen3.5-plus | qwen3-32b |
 | Aider | gpt-5.3-codex | gpt-5.3 | gpt-4o-mini |
-| MiniMax | MiniMax-M2.5 | MiniMax-M2.1 | MiniMax-M2.5-highspeed |
+| MiniMax | MiniMax-M2.7 | MiniMax-M2.7 | MiniMax-M2.5 |
 | Ollama | qwen2.5-coder:32b | llama3.3 | llama3.2 |
 
 ### Usage

package/SKILL.md

@@ -5,10 +5,10 @@ license: MIT
 compatibility: Requires bash, jq, and at least 2 AI CLI tools (gemini, codex, vibe, etc.). macOS and Linux.
 metadata:
   author: matteoscurati
-  version: 2.10.1
+  version: 2.10.3
 ---
 
-# AI Consultants v2.10.1 - AI Expert Panel
+# AI Consultants v2.10.3 - AI Expert Panel
 
 Simultaneously consult multiple AIs as "consultants" for coding questions. Each consultant has a **configurable persona** that influences their response style.
 

package/docs/COST_RATES.md

@@ -1,4 +1,4 @@
-# Cost Rates - AI Consultants v2.10.1
+# Cost Rates - AI Consultants v2.10.3
 
 This page documents the per-token rates used by the cost tracking system.
 
@@ -36,14 +36,13 @@ Since v2.5, models are organized into three tiers. Use `apply_model_tier()` or `
 | deepseek-chat | Economy | $0.001 | DeepSeek |
 | glm-5 | Premium/Standard | $0.002 | GLM |
 | glm-4-flash | Economy | $0.0005 | GLM |
-| grok-4-1-fast-reasoning | Premium | $0.005 | Grok |
+| grok-4.20-0309-reasoning | Premium | $0.005 | Grok |
 | grok-3 | Standard | $0.003 | Grok |
 | grok-3-mini | Economy | $0.001 | Grok |
 | qwen3.5-plus | Premium/Standard | $0.002 | Qwen3 |
 | qwen3-32b | Economy | $0.0005 | Qwen3 |
-| MiniMax-M2.5 | Premium | $0.002 | MiniMax |
-| MiniMax-M2.1 | Standard | $0.0015 | MiniMax |
-| MiniMax-M2.5-highspeed | Economy | $0.001 | MiniMax |
+| MiniMax-M2.7 | Premium/Standard | $0.002 | MiniMax |
+| MiniMax-M2.5 | Economy | $0.0015 | MiniMax |
 | kimi-code/kimi-for-coding | Premium/Standard | $0.002 | Kimi |
 | auto | All | $0.002 | Kilo (internal routing) |
 | default | - | $0.005 | Fallback |
@@ -70,14 +69,13 @@ Since v2.5, models are organized into three tiers. Use `apply_model_tier()` or `
 | deepseek-chat | Economy | $0.003 | DeepSeek |
 | glm-5 | Premium/Standard | $0.006 | GLM |
 | glm-4-flash | Economy | $0.0015 | GLM |
-| grok-4-1-fast-reasoning | Premium | $0.015 | Grok |
+| grok-4.20-0309-reasoning | Premium | $0.015 | Grok |
 | grok-3 | Standard | $0.009 | Grok |
 | grok-3-mini | Economy | $0.003 | Grok |
 | qwen3.5-plus | Premium/Standard | $0.006 | Qwen3 |
 | qwen3-32b | Economy | $0.0015 | Qwen3 |
-| MiniMax-M2.5 | Premium | $0.006 | MiniMax |
-| MiniMax-M2.1 | Standard | $0.004 | MiniMax |
-| MiniMax-M2.5-highspeed | Economy | $0.003 | MiniMax |
+| MiniMax-M2.7 | Premium/Standard | $0.006 | MiniMax |
+| MiniMax-M2.5 | Economy | $0.004 | MiniMax |
 | kimi-code/kimi-for-coding | Premium/Standard | $0.006 | Kimi |
 | auto | All | $0.006 | Kilo |
 | default | - | $0.015 | Fallback |

package/docs/cost_rates.json

@@ -1,5 +1,5 @@
 {
-  "version": "2.10.1",
+  "version": "2.10.3",
   "updated": "2026-03-04",
   "models": {
     "_comment_premium": "Premium tier models - March 2026",
@@ -9,10 +9,10 @@
     "mistral-large-3": {"input": 2.00, "output": 6.00},
     "deepseek-reasoner": {"input": 0.55, "output": 2.19},
     "glm-5": {"input": 0.50, "output": 2.00},
-    "grok-4-1-fast-reasoning": {"input": 5.00, "output": 15.00},
+    "grok-4.20-0309-reasoning": {"input": 5.00, "output": 15.00},
     "qwen3.5-plus": {"input": 1.60, "output": 6.40},
     "composer-1.5": {"input": 0.005, "output": 0.015},
-    "minimax-m2.5": {"input": 0.50, "output": 2.00},
+    "minimax-m2.7": {"input": 0.50, "output": 2.00},
     "kimi-code/kimi-for-coding": {"input": 0.50, "output": 2.00},
     "amp": {"input": 0.005, "output": 0.015},
 
@@ -23,7 +23,6 @@
     "mistral-medium-latest": {"input": 0.0027, "output": 0.0081},
     "deepseek-v3.2": {"input": 0.27, "output": 1.10},
     "grok-3": {"input": 3.00, "output": 9.00},
-    "minimax-m2.1": {"input": 0.30, "output": 1.50},
 
     "_comment_economy": "Economy tier models",
     "haiku-4.5": {"input": 0.001, "output": 0.005},
@@ -34,7 +33,7 @@
     "glm-4-flash": {"input": 0.001, "output": 0.003},
     "grok-3-mini": {"input": 1.00, "output": 3.00},
     "qwen3-32b": {"input": 0.40, "output": 1.60},
-    "minimax-m2.5-highspeed": {"input": 0.20, "output": 1.00},
+    "minimax-m2.5": {"input": 0.30, "output": 1.50},
 
     "_comment_legacy": "Legacy models for backwards compatibility",
     "claude-opus-4-5-20251124": {"input": 5.00, "output": 25.00},
@@ -63,9 +62,9 @@
     "qwen-turbo": {"input": 0.0008, "output": 0.002},
     "glm-4": {"input": 0.003, "output": 0.009},
     "glm-3-turbo": {"input": 0.001, "output": 0.003},
-    "grok-beta": {"input": 0.005, "output": 0.015},
     "grok-2": {"input": 0.01, "output": 0.03},
-    "deepseek-coder": {"input": 0.001, "output": 0.002}
+    "deepseek-coder": {"input": 0.001, "output": 0.002},
+    "minimax-m2.1": {"input": 0.30, "output": 1.50}
   },
   "consultant_fallbacks": {
     "_comment": "Premium defaults - Updated March 2026",
@@ -75,14 +74,14 @@
     "mistral": "mistral-large-3",
     "deepseek": "deepseek-reasoner",
     "glm": "glm-5",
-    "grok": "grok-4-1-fast-reasoning",
+    "grok": "grok-4.20-0309-reasoning",
     "qwen3": "qwen3.5-plus",
     "aider": "gpt-5.3-codex",
     "kilo": "auto",
     "cursor": "composer-1.5",
     "amp": "amp",
     "kimi": "kimi-code/kimi-for-coding",
-    "minimax": "minimax-m2.5",
+    "minimax": "minimax-m2.7",
     "ollama": "qwen2.5-coder:32b"
   },
   "model_tiers": {
@@ -94,14 +93,14 @@
       "mistral": "mistral-large-3",
       "deepseek": "deepseek-reasoner",
       "glm": "glm-5",
-      "grok": "grok-4-1-fast-reasoning",
+      "grok": "grok-4.20-0309-reasoning",
       "qwen3": "qwen3.5-plus",
       "aider": "gpt-5.3-codex",
       "kilo": "auto",
       "cursor": "composer-1.5",
       "amp": "amp",
       "kimi": "kimi-code/kimi-for-coding",
-      "minimax": "minimax-m2.5",
+      "minimax": "minimax-m2.7",
       "ollama": "qwen2.5-coder:32b"
     },
     "standard": {
@@ -118,7 +117,7 @@
       "cursor": "composer-1.5",
       "amp": "amp",
       "kimi": "kimi-code/kimi-for-coding",
-      "minimax": "minimax-m2.1",
+      "minimax": "minimax-m2.7",
       "ollama": "llama3.3"
     },
     "economy": {
@@ -135,7 +134,7 @@
       "cursor": "gemini-2.0-flash",
       "amp": "amp",
       "kimi": "kimi-code/kimi-for-coding",
-      "minimax": "minimax-m2.5-highspeed",
+      "minimax": "minimax-m2.5",
       "ollama": "llama3.2"
     }
   },

package/docs/releases/v2.10.1.md

@@ -0,0 +1,84 @@
+# Release v2.10.1
+
+**Date:** 2026-03-15
+**Type:** Patch — quality improvements, bug fixes, documentation
+**Previous:** v2.10.0
+
+## Highlights
+
+- Slash commands now guide Claude through file reading, result presentation, and error recovery
+- `debate_round.sh` hardened against 9 classes of silent failures
+- Token usage reduced ~17% on typical invocations via progressive disclosure
+
+## What's New
+
+### Slash Command Quality (P0-P2)
+
+The `/ai-consultants:consult` and `/ai-consultants:debate` commands were fully rewritten to solve three runtime problems identified by skill evaluation:
+
+- **File context handling**: When users pass file paths (e.g., `/consult "Review this" src/auth.ts`), Claude now reads each file with the Read tool and inlines the contents into the query before calling the script. This fixes the broken `$ARGUMENTS` single-string quoting issue.
+- **Result presentation**: Claude now reads `report.md` and `synthesis.json` from the output directory and presents structured results (consensus, summary table, recommendation, agreements, disagreements, risks, next steps).
+- **Error recovery**: Actionable guidance for exit code 1 (→ `doctor.sh`), exit code 124 (→ `--preset fast`), partial failures, and empty output.
+
+### debate_round.sh Hardening
+
+Nine issues fixed, including 4 critical:
+
+| Severity | Fix | Impact |
+|----------|-----|--------|
+| CRITICAL | `((count++)) \|\| true` (4 instances) | Script no longer aborts on bash 4+/5+ (Linux) |
+| CRITICAL | `*` default case in consultant routing | Future consultants no longer silently ignored |
+| CRITICAL | Amp, Kimi, MiniMax case entries added | All 15 consultants can now participate in debate |
+| HIGH | Subshell stderr → per-consultant `.err` files | Query failures now produce diagnostic logs |
+| HIGH | `$my_response` JSON validation + fallback | Corrupt responses no longer break debate prompts |
+| MEDIUM | Wait loop logs which consultant failed | Users see which consultant failed, not just count |
+| MEDIUM | ROUND_NUMBER integer validation | Non-integer input rejected with clear error |
+| MEDIUM | Config source warning instead of silent suppression | `config.sh` load failures now visible |
+| LOW | Temp file cleanup on jq failure | No more orphaned files in `/tmp` |
+
+### Token Efficiency
+
+| File | Before | After | Savings |
+|------|--------|-------|---------|
+| SKILL.md | ~2138 tokens | ~1773 tokens | -17% |
+| help.md (x3) | ~1084 tokens | ~251 tokens | -77% |
+
+Content moved to `references/details.md` for on-demand loading. Presets, strategies, bash usage, and best practices are no longer loaded on every invocation.
+
+### Documentation Audit
+
+18 issues fixed across 11 files:
+
+- **Consultant completeness**: Kimi added to README.md, SETUP.md, JSON_SCHEMA.md, SMART_ROUTING.md, COST_RATES.md, schema.json. MiniMax added where missing.
+- **Version consistency**: Headers updated from v2.8 to v2.10 in JSON_SCHEMA.md and SMART_ROUTING.md
+- **Stale defaults fixed**: OLLAMA_MODEL (`llama3.2` → `qwen2.5-coder:32b`), QWEN3_USE_API (`true` → `false`), DEBATE_CONFIDENCE_SPREAD_THRESHOLD (`2` → `3`)
+- **Path corrections**: `lib/debate_round.sh` → `debate_round.sh` in CLAUDE.md
+- **Consultant count**: `13` → `15` throughout SETUP.md
+- **Self-exclusion**: Claude's consult.md description no longer lists itself
+
+## Breaking Changes
+
+None. All changes are backwards-compatible.
+
+## Upgrade Guide
+
+```bash
+# If installed via git clone
+cd ~/.claude/skills/ai-consultants && git pull
+
+# If installed via curl | bash
+curl -fsSL https://raw.githubusercontent.com/matteoscurati/ai-consultants/main/scripts/install.sh | bash
+```
+
+No configuration changes required.
+
+## Commits
+
+- `d1c2829` feat: slash command quality improvements and debate_round.sh hardening
+- `827c203` docs: token efficiency optimization and comprehensive documentation audit
+- `0a3b057` chore: bump version to 2.10.1
+
+## Contributors
+
+- @matteoscurati
+- Claude Opus 4.6 (1M context)

package/docs/releases/v2.10.2.md

@@ -0,0 +1,40 @@
+# Release v2.10.2
+
+**Date:** 2026-03-19
+**Type:** Patch — MiniMax M2.7 model upgrade
+**Previous:** v2.10.1
+
+## Highlights
+
+- MiniMax upgraded from M2.5 to M2.7 (released 2026-03-18)
+- Premium and standard tiers now use MiniMax-M2.7
+- Economy tier uses MiniMax-M2.5 (previous default)
+## What's New
+
+### MiniMax M2.7 Model Upgrade
+
+MiniMax released M2.7 on March 18, 2026. This patch updates all MiniMax tier assignments:
+
+| Tier | Before (v2.10.1) | After (v2.10.2) |
+|------|-------------------|------------------|
+| Premium | MiniMax-M2.5 | MiniMax-M2.7 |
+| Standard | MiniMax-M2.1 | MiniMax-M2.7 |
+| Economy | MiniMax-M2.5-highspeed | MiniMax-M2.5 |
+
+Default model (`MINIMAX_MODEL`) is now `MiniMax-M2.7`.
+
+## Breaking Changes
+
+None.
+
+## Upgrade Guide
+
+```bash
+# If installed via git clone
+cd ~/.claude/skills/ai-consultants && git pull
+
+# If installed via curl | bash
+curl -fsSL https://raw.githubusercontent.com/matteoscurati/ai-consultants/main/scripts/install.sh | bash
+```
+
+No configuration changes required. Existing `MINIMAX_MODEL` overrides continue to work.

package/docs/releases/v2.10.3.md

@@ -0,0 +1,63 @@
+# Release v2.10.3
+
+**Date:** 2026-03-21
+**Type:** Patch — Model updates, API migration, cleanup
+**Previous:** v2.10.2
+
+## Highlights
+
+- Grok premium model upgraded to `grok-4.20-0309-reasoning`
+- GLM API migrated from `open.bigmodel.cn` to `api.z.ai/api/coding/paas/v4`
+- Removed non-functional MiniMax highspeed models
+- Fixed stale GLM URL fallbacks in `common.sh` and `configure.sh`
+
+## What's New
+
+### Grok Model Upgrade
+
+Grok premium model updated from `grok-4-1-fast-reasoning` to `grok-4.20-0309-reasoning`. Standard (`grok-3`) and economy (`grok-3-mini`) tiers remain unchanged.
+
+| Tier | Before (v2.10.2) | After (v2.10.3) |
+|------|-------------------|------------------|
+| Premium | grok-4-1-fast-reasoning | grok-4.20-0309-reasoning |
+| Standard | grok-3 | grok-3 |
+| Economy | grok-3-mini | grok-3-mini |
+
+### GLM API Migration
+
+Zhipu AI migrated their API from `open.bigmodel.cn` to `api.z.ai`. The default endpoint is now `https://api.z.ai/api/coding/paas/v4/chat/completions`. Updated in:
+
+- `config.sh` (default)
+- `common.sh` (fallback in `get_api_url()`)
+- `configure.sh` (wizard URL array)
+- `.env.example` (example config)
+
+### Removed Non-Functional Models
+
+Removed models not supported by the current MiniMax token plan:
+
+- `MiniMax-M2.7-highspeed`
+- `MiniMax-M2.5-highspeed`
+- `grok-beta` (legacy, no longer recognized by xAI API)
+
+### Bug Fixes
+
+- Fixed stale GLM URL fallback in `get_api_url()` (`common.sh`) — would have routed to old endpoint if `GLM_API_URL` env var was unset
+- Fixed stale GLM URL in configure wizard (`configure.sh`) — would have written old endpoint to `.env`
+- Fixed duplicate `minimax-m2.5` entry in `cost_rates.json` with conflicting input rates ($0.30 vs $0.50)
+
+## Breaking Changes
+
+None.
+
+## Upgrade Guide
+
+```bash
+# If installed via git clone
+cd ~/.claude/skills/ai-consultants && git pull
+
+# If installed via curl | bash
+curl -fsSL https://raw.githubusercontent.com/matteoscurati/ai-consultants/main/scripts/install.sh | bash
+```
+
+No configuration changes required. Existing `GROK_MODEL` and `GLM_API_URL` overrides continue to work.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ai-consultants",
-  "version": "2.10.1",
+  "version": "2.10.3",
   "description": "Multi-model AI deliberation system - query up to 15 AI consultants for coding problems",
   "bin": {
     "ai-consultants": "bin/ai-consultants"

package/references/configuration.md

@@ -81,9 +81,9 @@ AMP_MODEL=amp
 KIMI_MODEL=kimi-code/kimi-for-coding
 QWEN3_MODEL=qwen3.5-plus
 GLM_MODEL=glm-5
-GROK_MODEL=grok-4-1-fast-reasoning
+GROK_MODEL=grok-4.20-0309-reasoning
 DEEPSEEK_MODEL=deepseek-reasoner
-MINIMAX_MODEL=MiniMax-M2.5
+MINIMAX_MODEL=MiniMax-M2.7
 OLLAMA_MODEL=qwen2.5-coder:32b
 ```
 

package/scripts/config.sh

@@ -119,7 +119,7 @@ QWEN3_FORMAT="${QWEN3_FORMAT:-qwen}"
 
 GLM_MODEL="${GLM_MODEL:-glm-5}"
 GLM_TIMEOUT_SECONDS="${GLM_TIMEOUT:-180}"
-GLM_API_URL="${GLM_API_URL:-https://open.bigmodel.cn/api/paas/v4/chat/completions}"
+GLM_API_URL="${GLM_API_URL:-https://api.z.ai/api/coding/paas/v4/chat/completions}"
 GLM_FORMAT="${GLM_FORMAT:-openai}"
 # API key: Set GLM_API_KEY environment variable
 
@@ -127,7 +127,7 @@ GLM_FORMAT="${GLM_FORMAT:-openai}"
 # GROK CONFIGURATION - The Provocateur (API-based)
 # =============================================================================
 
-GROK_MODEL="${GROK_MODEL:-grok-4-1-fast-reasoning}"
+GROK_MODEL="${GROK_MODEL:-grok-4.20-0309-reasoning}"
 GROK_TIMEOUT_SECONDS="${GROK_TIMEOUT:-180}"
 GROK_API_URL="${GROK_API_URL:-https://api.x.ai/v1/chat/completions}"
 GROK_FORMAT="${GROK_FORMAT:-openai}"
@@ -147,7 +147,7 @@ DEEPSEEK_FORMAT="${DEEPSEEK_FORMAT:-openai}"
 # MINIMAX CONFIGURATION - The Pragmatic Optimizer (API-based, v2.10)
 # =============================================================================
 
-MINIMAX_MODEL="${MINIMAX_MODEL:-MiniMax-M2.5}"
+MINIMAX_MODEL="${MINIMAX_MODEL:-MiniMax-M2.7}"
 MINIMAX_TIMEOUT_SECONDS="${MINIMAX_TIMEOUT:-180}"
 MINIMAX_API_URL="${MINIMAX_API_URL:-https://api.minimax.io/v1/chat/completions}"
 MINIMAX_FORMAT="${MINIMAX_FORMAT:-openai}"
@@ -532,13 +532,13 @@ get_model_for_tier() {
                 cursor)   echo "composer-1.5" ;;
                 deepseek) echo "deepseek-reasoner" ;;
                 glm)      echo "glm-5" ;;
-                grok)     echo "grok-4-1-fast-reasoning" ;;
+                grok)     echo "grok-4.20-0309-reasoning" ;;
                 qwen3)    echo "qwen3.5-plus" ;;
                 aider)    echo "gpt-5.3-codex" ;;
                 amp)      echo "amp" ;;
                 ollama)   echo "qwen2.5-coder:32b" ;;
                 kimi)     echo "kimi-code/kimi-for-coding" ;;
-                minimax)  echo "MiniMax-M2.5" ;;
+                minimax)  echo "MiniMax-M2.7" ;;
                 kilo)     echo "auto" ;;
                 *)        echo "" ;;
             esac
@@ -558,7 +558,7 @@ get_model_for_tier() {
                 amp)      echo "amp" ;;  # Same model (no tiers)
                 ollama)   echo "llama3.3" ;;
                 kimi)     echo "kimi-code/kimi-for-coding" ;;
-                minimax)  echo "MiniMax-M2.1" ;;
+                minimax)  echo "MiniMax-M2.7" ;;
                 kilo)     echo "auto" ;;
                 *)        echo "" ;;
             esac
@@ -578,7 +578,7 @@ get_model_for_tier() {
                 amp)      echo "amp" ;;  # Same model (no tiers)
                 ollama)   echo "llama3.2" ;;
                 kimi)     echo "kimi-code/kimi-for-coding" ;;
-                minimax)  echo "MiniMax-M2.5-highspeed" ;;
+                minimax)  echo "MiniMax-M2.5" ;;
                 kilo)     echo "auto" ;;
                 *)        echo "" ;;
             esac

package/scripts/configure.sh

@@ -47,8 +47,8 @@ CLI_AGENT_PERSONAS=("The Architect" "The Pragmatist" "The Devil's Advocate" "The
 # API Agents: name|key_var|api_url|model|persona
 API_AGENT_NAMES=("Qwen3" "GLM" "Grok" "DeepSeek")
 API_AGENT_KEY_VARS=("QWEN3_API_KEY" "GLM_API_KEY" "GROK_API_KEY" "DEEPSEEK_API_KEY")
-API_AGENT_URLS=("https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation" "https://open.bigmodel.cn/api/paas/v4/chat/completions" "https://api.x.ai/v1/chat/completions" "https://api.deepseek.com/v1/chat/completions")
-API_AGENT_MODELS=("qwen-max" "glm-4" "grok-4-1-fast-reasoning" "deepseek-reasoner")
+API_AGENT_URLS=("https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation" "https://api.z.ai/api/coding/paas/v4/chat/completions" "https://api.x.ai/v1/chat/completions" "https://api.deepseek.com/v1/chat/completions")
+API_AGENT_MODELS=("qwen-max" "glm-4" "grok-4.20-0309-reasoning" "deepseek-reasoner")
 API_AGENT_PERSONAS=("The Analyst" "The Methodologist" "The Provocateur" "The Code Specialist")
 
 # State arrays (parallel to CLI_AGENT_NAMES)
@@ -771,7 +771,7 @@ QWEN3_TIMEOUT=180
 GLM_MODEL=glm-4
 GLM_TIMEOUT=180
 
-GROK_MODEL=grok-4-1-fast-reasoning
+GROK_MODEL=grok-4.20-0309-reasoning
 GROK_TIMEOUT=180
 
 DEEPSEEK_MODEL=deepseek-reasoner

package/scripts/lib/common.sh

@@ -354,7 +354,7 @@ get_api_url() {
         CLAUDE)     echo "${CLAUDE_API_URL:-https://api.anthropic.com/v1/messages}" ;;
         MISTRAL)    echo "${MISTRAL_API_URL:-https://api.mistral.ai/v1/chat/completions}" ;;
         QWEN3)      echo "${QWEN3_API_URL:-https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation}" ;;
-        GLM)        echo "${GLM_API_URL:-https://open.bigmodel.cn/api/paas/v4/chat/completions}" ;;
+        GLM)        echo "${GLM_API_URL:-https://api.z.ai/api/coding/paas/v4/chat/completions}" ;;
         GROK)       echo "${GROK_API_URL:-https://api.x.ai/v1/chat/completions}" ;;
         DEEPSEEK)   echo "${DEEPSEEK_API_URL:-https://api.deepseek.com/v1/chat/completions}" ;;
         MINIMAX)    echo "${MINIMAX_API_URL:-https://api.minimax.io/v1/chat/completions}" ;;

package/scripts/lib/costs.sh

@@ -149,7 +149,7 @@ get_input_cost_per_1k() {
         glm-4)            echo "0.003" ;;
         glm-3-turbo)      echo "0.001" ;;
         # Grok models (xAI)
-        grok-4-1-fast-reasoning)        echo "0.005" ;;
+        grok-4.20-0309-reasoning)        echo "0.005" ;;
         grok-2)           echo "0.01" ;;
         # Default
         *)                echo "0.005" ;;
@@ -196,7 +196,7 @@ get_output_cost_per_1k() {
         glm-4)            echo "0.009" ;;
         glm-3-turbo)      echo "0.003" ;;
         # Grok models (xAI)
-        grok-4-1-fast-reasoning)        echo "0.015" ;;
+        grok-4.20-0309-reasoning)        echo "0.015" ;;
         grok-2)           echo "0.03" ;;
         # Default
         *)                echo "0.015" ;;
@@ -493,7 +493,7 @@ get_economic_model() {
         grok)     echo "grok-3-mini" ;;
         deepseek) echo "deepseek-chat" ;;
 
-        minimax)  echo "MiniMax-M2.5-highspeed" ;;
+        minimax)  echo "MiniMax-M2.5" ;;
         *)        echo "" ;;
     esac
 }

package/scripts/query_grok.sh

@@ -5,7 +5,7 @@
 #
 # Environment variables:
 #   GROK_API_KEY    - API key for xAI (required)
-#   GROK_MODEL      - Model to use (default: grok-4-1-fast-reasoning)
+#   GROK_MODEL      - Model to use (default: grok-4.20-0309-reasoning)
 #   GROK_TIMEOUT    - Timeout in seconds (default: 180)
 #   ENABLE_PERSONA  - Enable "The Provocateur" persona (default: true)
 

package/scripts/query_minimax.sh

@@ -5,7 +5,7 @@
 #
 # Environment variables:
 #   MINIMAX_API_KEY    - API key for MiniMax (required)
-#   MINIMAX_MODEL      - Model to use (default: MiniMax-M2.5)
+#   MINIMAX_MODEL      - Model to use (default: MiniMax-M2.7)
 #   MINIMAX_TIMEOUT    - Timeout in seconds (default: 180)
 #   ENABLE_PERSONA     - Enable "The Pragmatic Optimizer" persona (default: true)