Repository navigation
Releases
Releases · misiekhardcore/agents-flow
v2.1.0
[2.1.0] - 2026-06-16
Added
- Rewrite AGENTS.md and docs for tool-agnostic language
- Update templates and shared protocols for tool-agnostic language
- Add opencode frontmatter to agent files
- Add opencode-compatible frontmatter to skill files
Changed
- Release v2.1.0
- Minify all markdown files to pass format check
- Review bin tools and hooks for compatibility
- Rebrand claude-workflow → agents-flow
v2.0.1
[2.0.1] - 2026-06-16
Changed
- Release v2.0.1
Fixed
- Update token limit for NOTES.md to 2k tokens across multiple skills and documentation
v2.0.0
[2.0.0] - 2026-06-16
Added
- Add git-cliff changelog handling
- Rebuild define
- Clarify SKILL.md and agent file responsibilities, and document knowledge persistence in AGENTS.md
- Add formatter test workflow and minify markdown tests
- Enhance SKILL.md for clarity and detail in process descriptions
- Rebuild audit-issues
- Update SKILL.md for clarity and consistency - refine description, adjust model type, and streamline process steps
- Rebuild new-skill as Orchestrator- Rewrite SKILL.md as thin Orchestrator with interview flow- Read _shared/AUTHORING.md and references/interview-steps.md at point of need- Consolidate to single minimal template approach- Remove layer: field from frontmatter (handled in #177)Co-authored-by: openhands openhands@all-hands.dev
- Add AGENTS.md documentation for plugin commands and conventions
- Create missing agent files for discovery and define
- Add layer declarations to remaining v2 skills
- Split interactive skills by consumption contract (autonomous research + interactive deliberation)
- Rebuild verify and build as v2 L2 sub-skills
- Rebuild find-skills and review as v2 L2 sub-skills
- Rebuild describe and specify as v2 L2 sub-skills
- Create worktree and notes-md L3 skills, update wrap-up and compound
- Enhance SKILL.md files with detailed descriptions and user-invocable flags
- Update SKILL.md files to clarify output and handoff processes
- Add user-invocable frontmatter and caller contracts to simple L2 skills
- Decouple shared docs by moving per-skill details into owning skills- _shared/handoff-artifact.md: Trimmed to protocol-only (five-field format, shape, precedence, rules). Removed phase-specific heading rules and final-pass checklist — these are now inline in each producing skill.- _shared/seed-brief.md: Trimmed to universal protocol spec. Payload type definitions (research, fix, prior-art) and canonical example moved to skills/specialist-mode/SKILL.md.- skills/discovery/SKILL.md: Added ## Requirements section heading and final-pass checklist inline.- skills/define/SKILL.md: Cleaned handoff instructions to reference the five-field format and ## Implementation plan heading.- skills/implement/SKILL.md: Clarified section heading origins (Requirements from discovery, Implementation plan from define).- skills/specialist-mode/SKILL.md: Added Payload Types section (moved from seed-brief) and Execution Delta table (seeded vs standalone behavior).- README.md: Added seed-brief.md to the _shared/ protocol table.Co-authored-by: openhands openhands@all-hands.dev
- Add compound-on-exit protocol and assessment gate
- Rebuild as L2 specialist with dual-mode seed-brief
- Replace Lightweight/Standard/Deep scope labels with two-mechanism pattern
- Promote behavioral _shared/ files to layer-3 skills
- Author /scope-assessment as shared tier-3 skill
Changed
- Release v2.0.0
- Consolidate all agent files to root agents/ directory with workflow- prefix
- Remove outdated version references from AUTHORING.md
- Rebuild epic-autopilot
- Rebuild issue-autopilot
- Rebuild /new-skill as Orchestrator
- Rebuild /new-skill as Orchestrator
- Rebuild audit-issues as Orchestrator
- Feat/132 implement v2
- Rebuild define
- Remove stale /discovery references and post-redesign leftovers
- Rebuild discover
- Rebuild resolve-pr-feedback
- Rebuild grill-me
- Update bin/list-prune-files to remove dependency on layer: frontmatter field
- Apply dispatch primitives framework — taxonomy, skill reclassification, issue audit
- NOTES.md as orchestrator progress ledger with checkpoint/slice pattern
- Layer-3 boundary realignment: demote seed-brief, worktree, interviewing-rules to _shared/
- Switch license from MIT to PolyForm Noncommercial 1.0.0
- V2 F1: bootstrap v2 branch and AUTHORING.md for tier-3 hierarchy
Documentation
- Expand AGENTS.md with commit message formatting rules
- Add AGENTS.md with PR formatting conventions
- Update CHANGELOG with handoff-artifact and notes-md demotion
Fixed
- Add --unreleased flag to git-cliff in release workflow
- Remove stale specialist-mode reference in README protocol table
- Update argument-hint in SKILL.md to include PR-URL option
- Correct AGENTS.md path in CLAUDE.md
- Update skill count in marketplace.json description
- Clarify filing step in compound skill process
- Replace stale specialist-mode reference with implement-runner agent
- Complete v2 terminology cleanup and documentation update
v1.6.2
What's Changed
- feat(epic-autopilot): run /compound at end of stage 4 for epic-level learnings by @misiekhardcore in #124
Full Changelog: v1.6.1...v1.6.2
v1.6.1
What's Changed
- fix(issue-autopilot): move /compound review-time pass from stage 5 to stage 3 by @misiekhardcore in #122
Full Changelog: v1.6.0...v1.6.1
v1.6.0
What's Changed
- fix: enforce worktree creation for all build scopes by @misiekhardcore in #117
- docs(authoring): establish lazy-loading standard for skill reference files by @misiekhardcore in #119
- refactor(wrap-up): extract procedure to references/procedure.md by @misiekhardcore in #120
Full Changelog: v1.5.1...v1.6.0
v1.5.1
What's Changed
- slim: reduce find-skills skill to 50-line cap by @misiekhardcore in #105
- slim: reduce epic-autopilot skill to 50-line cap by @misiekhardcore in #107
- slim: reduce new-skill to 44-line cap, extract Process section by @misiekhardcore in #109
- slim: reduce build skill to 50-line cap by @misiekhardcore in #114
- slim: reduce 9 over-cap skills to 50-line limit by @misiekhardcore in #110
- fix: replace [Ref:] shorthand with explicit read instructions across skills by @misiekhardcore in #113
- slim: reduce issue-autopilot skill to 50-line cap by @misiekhardcore in #111
Full Changelog: v1.5.0...v1.5.1
v1.5.0
What's Changed
- Redesign /prune: drop Rules+Vault lanes, add Dead-state audit by @misiekhardcore in #98
Full Changelog: v1.4.0...v1.5.0
v1.4.0
What's Changed
- Add verification step to resolve-pr-feedback by @misiekhardcore in #94
- docs(AUTHORING): add external-audience writing rule by @misiekhardcore in #97
- feat(CLAUDE.md): time-box codebase reading during /define and /discovery by @misiekhardcore in #96
- feat: add /ship skill — single-issue end-to-end pipeline by @misiekhardcore in #95
Full Changelog: v1.3.0...v1.4.0
v1.3.0
What's Changed
- Improve skill authoring quality: conditional when_to_use, modularity, paired prohibitions by @misiekhardcore in #90
- feat(prune): extend Authoring Lane to scan _shared/.md and .claude/**/.md by @misiekhardcore in #91
Full Changelog: v1.2.0...v1.3.0