v2.2.0

What's New

Added

• setup-ci skill — configure opencode-actions workflows (#70)
• auto-delete error comments — automatically delete error comments (403, network errors) posted by opencode to PRs after failed runs (#77). New cleanup-error-comments input (default: true) on all actions.

Fixed

• Enforce read-only mode via opencode permission config (#72)

Changed

• Add npx skills add command to README (#71)

v2.1.0

What's New

Added

• spec-coverage action for cross-referencing spec tasks against PR implementation

Fixed

• Add extra-env input and README documentation for spec-coverage
• Treat missing spec files as CRITICAL gap, not graceful exit
• Remove contradictory NO_SPEC_FOUND verdict
• Smart spec-need detection and remove gh CLI dependency
• Simplify spec-need detection to semantic judgment

Changed

• Ignore local settings

Full Changelog: v2.0.5...v2.1.0

v2.0.5

What's Changed

Added

• review action: extra-env input — pass custom environment variables to opencode runtime via multi-line KEY=VALUE pairs (#60)

- uses: sun-praise/opencode-actions/review@v2
  with:
    model: kimi-for-coding/k2p5
    extra-env: |
      KIMI_API_KEY=${{ secrets.KIMI_API_KEY }}
      E2E_TEST_USERNAME=${{ secrets.E2E_TEST_USERNAME }}

Full Changelog: v2.0.4...v2.0.5

v2.0.4

What's Changed

Fixed

• feature-missing: prevent stale analysis on new commits via prompt + version resolution + cache key prefix

Changed

• chore: ignore worktree folder

Full Changelog: v2.0.3...v2.0.4

v2.0.3

What's Changed

Fixed

• setup-opencode: restore default-version to opencode minimum version

Changed

• docs: add bump version note about default-version file
• chore: add agents.md

Full Changelog: v2.0.2...v2.0.3

v2.0.2

What's Changed

Fixed

• feature-missing: prevent contradictory verdict when no gaps found (#51)

Full Changelog: v2.0.1...v2.0.2

v2.0.1

What's Changed

• Bump minimum supported OpenCode version to 2.0.1
• feat: support reasoning-effort and enable-thinking configuration (#44)
• fix: update action refs from v1 to v2 (#42)
• ci: switch from self-hosted to ubuntu-latest (#41)

Full Changelog: v2.0.0...v2.0.1

v2.0.0

What's Changed

Added

• Timeout-based review model fallback (automatically fallback on timeout/deadline exceeded)
• Python3 rewrite of run-github-opencode with improved reliability
• feature-missing.yml workflow
• Smoke test workflow for dev branch

Changed

• Increased default model-timeout-seconds from 120 to 300
• Unified entire test suite as Python3 unittest

Fixed

• Addressed PR #10 review feedback (multiple rounds)
• Addressed PR #36 review suggestions
• Addressed CI review round 3 feedback

Full Changelog: v1.4.0...v2.0.0

v1.4.0

What's New

• feature-missing: new action that audits PR implementation against the linked issue spec to find missing or incomplete features before merge
  • Automatically reads linked issue body as the feature spec
  • Falls back to extracting requirements from PR title/body when no issue is linked
  • Classifies gaps by severity: CRITICAL, MEDIUM, LOW
  • Shares cache and inputs with existing review action

Usage

- name: Run feature missing audit
  uses: Svtter/opencode-actions/feature-missing@v1
  with:
    github-token: ${{ secrets.GITHUB_TOKEN }}
    zhipu-api-key: ${{ secrets.ZHIPU_API_KEY }}
    opencode-go-api-key: ${{ secrets.OPENCODE_GO_API_KEY }}

Full Changelog: v1.3.0...v1.4.0

v1.3.0

What's Changed

Fixed

• Remove hardcoded review prompt from github-run-opencode (#21)
• Correct spurious quotes in test command substitutions
• Remove --foreground from timeout to preserve SIGTERM guarantee
• Set FAKE_OPENCODE_VERSION for github-run-opencode smoke test

Changed

• Prefer MODEL_NAME for default review model (#16)
• Add configurable github run timeout
• Update example workflows and documentation

Added

• New review action: opinionated review wrapper with built-in defaults for opencode github run
• Enhanced smoke tests for github-run-opencode and review action