v0.6.0 Beta

What's Changed

Troubleshooting and Diagnostics Enhancements:

Added a comprehensive /cortexlink:doctor command that interactively diagnoses environment issues, plugin file integrity, hook health, agent installation/authentication, and setup state, guiding the user through targeted checks and fixes.
Expanded the README with a dedicated Troubleshooting section, providing clear steps for resolving common issues related to command visibility, hook context injection, agent detection, authentication, and delegation failures.
Setup, Cleanup, and Permissions:

Added allowed-tools frontmatter to setup (Read, Edit, Bash, Write) and cleanup (Read, Edit) commands, restricting tool usage to the minimum required set for each command.
Improved setup and cleanup command flows: now explicitly inform the user before appending configuration blocks, and provide clearer status reporting and next steps.
Command and Skill Structure Updates:

Removed the /cortexlink:auto command and its documentation, as orchestration is now automatically injected via the SessionStart hook.
Updated the help command to use direct output instructions and reflect the new command set, including the addition of /cortexlink:doctor.
Documentation and Changelog:

Updated the README with clearer prerequisites, improved installation and update instructions, and more accurate descriptions of automatic orchestration.
Added detailed changelog entries for versions 0.5.6 and 0.5.7, documenting all improvements, fixes, and architectural changes.
Internal Consistency and Formatting:

Enforced consistent line endings for hook scripts using .gitattributes.
Updated plugin version to 0.5.7 in plugin.json.
These changes collectively make CortexLink more robust, user-friendly, and easier to maintain.

Troubleshooting and Diagnostics

Added /cortexlink:doctor for interactive, symptom-driven diagnostics covering environment, plugin files, hooks, agents, and setup state.
Expanded README with a Troubleshooting section covering command visibility, hook context, agent detection, authentication, and delegation issues.
Setup, Cleanup, and Permissions

Added allowed-tools frontmatter to setup and cleanup commands for stricter tool access control.
Improved setup/cleanup flows: now inform user before appending config, and provide clearer reporting.
Command and Skill Structure

Removed /cortexlink:auto command and related documentation; orchestration is now automatic via the SessionStart hook.
Updated help command for direct output and new command set, including /cortexlink:doctor.
Documentation and Changelog

Improved README with prerequisites, installation, update, and orchestration details.
Updated CHANGELOG for versions 0.5.6 and 0.5.7, detailing all changes. [
Internal Consistency
Enforced line endings for hook scripts in .gitattributes.
Bumped plugin version to 0.5.7.

v0.5.5-beta

[0.5.5] — 2026-03-23

Changed

• Renamed project from xFlow to CortexLink
• All commands renamed from /xflow:* to /cortexlink:*
• Plugin install slug changed from xflow@agent-plugins to cortexlink@agent-plugins
• GitHub repo to be renamed from namos2502/xFlow to namos2502/CortexLink

v0.5.4-beta

[0.5.4] — 2026-03-21

Added

• Task complexity scale (Simple / Standard / Complex) in orchestration skill — classifies tasks before routing to determine spec detail and delegation style
• Effort threshold: Simple tasks stay inline or use native subagents — cross-CLI overhead is not justified for single-operation or read-only tasks
• Q&A phase for Complex tasks — control center appends a 1-turn question block to the delegation prompt before execution; agent surfaces ambiguity, control center refines, then execution proceeds
• Updated decision tree — checks task complexity before type; native subagent tier made explicit

v0.5.3-beta

[0.5.3] — 2026-03-21

Fixed

• /xflow:auto guard restructured from a soft "Step 1" bullet to a hard ⛔ STOP block — models were reading past the conditional and activating regardless; now the gate is unambiguous and precedes the activation section entirely
• /xflow:setup now ends every confirmation with a prompt to run /xflow:auto next — closes the gap where users had no clear signal of what to do after setup

v0.5.2-beta

[0.5.2] — 2026-03-21

Added

• /xflow:auto now checks both ~/.claude/CLAUDE.md and ~/.copilot/copilot-instructions.md for the ## xFlow section before activating — blocks if neither is set up, warns if only one is present

Fixed

• /xflow:auto no longer activates silently when setup hasn't been run

v0.5.1-beta

[0.5.1] — 2026-03-21

Added

/xflow:cleanup — removes the ## xFlow sections from ~/.claude/CLAUDE.md and ~/.copilot/copilot-instructions.md added during setup

Changed

/xflow:help — simplified to plugin info and getting started guide; no longer shows internal skill structure

Fixed

/xflow:help — removed stray --- from command body that was being parsed as a second frontmatter block, causing Claude to show its default greeting instead of the help content

v0.5.0-beta

[0.5.0-beta]

Changed — Breaking

• Architecture: Complete redesign from passive flag-reference plugin to active control center orchestration system
• Skills restructured: skills/copilot-cli/ and skills/claude-cli/ replaced by skills/orchestration/ (control center protocol) and skills/agents/copilot-cli/ + skills/agents/claude-cli/ (behavioral agent skills)
• Commands trimmed: Removed ask, fix, explain, suggest, review — the orchestration skill handles routing automatically. Kept setup, auto, help.
• /xflow:auto repurposed: from "primary activation" to "explicit session override" — after /xflow:setup, xFlow is always-on via ~/.claude/CLAUDE.md
• /xflow:setup now registers xFlow in both ~/.claude/CLAUDE.md and ~/.copilot/copilot-instructions.md — always-on in Claude Code and Copilot CLI

Added

• skills/orchestration/SKILL.md — control center protocol: routing decision tree, delegation prompt template, structured report format, self-verify loop, error handling, cross-agent chaining, recursion guard
• skills/agents/copilot-cli/SKILL.md — behavioral reference: when to use, flags, invocation patterns
• skills/agents/claude-cli/SKILL.md — behavioral reference: when to use, flags, model selection, key differences from Copilot CLI
• Control Center model — your active AI agent acts as the director; fans out tasks to peer CLI agents, reviews every report before proceeding
• Agent model — each CLI agent is a full peer, not a dumb executor; uses its own platform-native tools and sub-agents internally, self-verifies, and reports back
• Fan-out tree — one level deep, width scales; agents don't chain to each other
• Structured report format — every agent returns STATUS / SUMMARY / STEPS / FILES / ISSUES (~150 words)
• Self-verify before reporting — agents must verify their own work before returning ✅
• Extensible agent architecture — skills/agents/<name>/SKILL.md pattern; drop a folder to add a new agent

NOTE: Versions before v0.5.0-beta does not have accurate Assets.

v0.4.0-beta

[0.4.0-beta] — 2026-03-19

Added

• skills/claude-cli/SKILL.md — programmatic reference for the Claude CLI; covers claude -p flags, tool permission syntax (--allowedTools, --disallowedTools, --tools), model selection, budget caps, and common invocation patterns
• /xflow:auto now loads both copilot-cli and claude-cli skill references; includes routing guidance (Copilot for GitHub tasks, Claude for general code tasks)
• /xflow:setup now detects and authenticates both Copilot CLI and Claude CLI; CLAUDE.md block updated to describe both agents
• /xflow:help now shows commands in a multi-agent context

NOTE: Version Assets is not correct to the release since there was a big change. This is mostly for release log (Correct Assets starting from v0.5.0-beta)

v0.3.0-beta

[0.3.0-beta] — 2026-03-19

Changed

• Renamed project from CoFluent to xFlow — reflecting the broader multi-agent vision
• New tagline: "One plugin. Every CLI. Stay in flow."
• All commands renamed from /cofluent:* to /xflow:*
• Plugin install slug changed from cofluent@agent-plugins to xflow@agent-plugins
• GitHub repo renamed from namos2502/CoFluent to namos2502/xFlow

NOTE: Version Assets is not correct to the release since there was a big change. This is mostly for release log (Correct Assets starting from v0.5.0-beta)

v0.2.0-beta

[0.2.0-beta] — 2026-03-18

Added

• /xflow:setup — one-time setup command that verifies Copilot CLI is installed and authenticated, then registers xFlow awareness in ~/.claude/CLAUDE.md so Claude knows about xFlow in every future session (idempotent — safe to re-run)
• ## Why xFlow section in README covering two key benefits: staying in flow and saving tokens on Claude by offloading subtasks to a separate copilot -p process

Removed

• /xflow:verify — absorbed into /xflow:setup; re-run setup anytime to re-check your installation

NOTE: Version Assets is not correct to the release since there was a big change. This is mostly for release log (Correct Assets starting from v0.5.0-beta)