Skip to content

Loki Mode

v7.54.0 Breaking

This release includes 1 breaking change for platform teams planning a safe upgrade.

✓ No known CVEs patched
Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

ai-agents aider anthropic autonomous ci-cd claude
+13 more
cline code-review devops gemini github-action github-issues loki-mode multi-agent openai-codex openapi pull-request-review sdlc spec-driven-development

Affected surfaces

rbac

Summary

AI summary

Updates P4-3, P4-5, and documented across a mixed release.

Full changelog

Tech-debt elimination + provider honesty + regression hardening

  • Dead-stub removal (P4-3) (loki-ts/src/runner/autonomous.ts): the Bun
    runner carried logStub/noopCouncil/stubProvider fallbacks for modules that all
    exist and conform. These were unreachable dead code that silently degraded to
    wrong results if ever hit. Replaced with a fail-fast requireModule() (a
    missing load-bearing module now errors loudly at resolution time instead of
    producing a wrong build), while genuine runtime-error fallbacks are preserved.
  • Antigravity de-listed (P4-5): "Antigravity CLI" was advertised across docs
    as an upcoming Loki provider but had zero implementation, and was misattributed
    to Anthropic (it is Google's product). Removed from all provider tables/lists
    (README, SKILL, CLAUDE, wiki, docs/INSTALLATION). Honest competitor references
    in docs/COMPARISON.md are retained.
  • Bun semantic-findings consumer wired (loki-ts/src/runner/build_prompt.ts):
    the Bun semantic-test gate persisted .loki/quality/semantic-findings.txt but
    nothing on the Bun route read it (a writer-with-no-reader). build_prompt.ts now
    reads and injects those severity-tagged findings into the next-iteration prompt,
    independent of gate-failures.txt, byte-parity with the bash route. Absent/empty
    injects nothing.
  • Policy-context quoting fix (autonomy/run.sh check_policy): ${2:-{}}
    brace-eating expansion turned a non-empty JSON context into invalid JSON
    ({"a":1}}), so the policy engine received garbage and returned DENY every
    iteration when a .loki/policies.json/.yaml existed (and made the P3-3
    approval-wait unreachable on the live path). Fixed to a split default yielding
    valid {}; non-empty contexts now pass through unchanged. Users with no policy
    file are unaffected.
  • Compliance snapshot scheduler (P3-11, optional) (src/audit/compliance-scheduler.js):
    a default-disabled helper that periodically persists a compliance snapshot from
    the real audit chain (honest empty-state, never a fabricated verdict). Ships as
    a tested helper; not yet auto-invoked (documented).
  • Regression hardening: added guards for the v7.51-v7.54 features, notably
    tests/test-no-deprecated-codex-flag.sh (fails if any live codex exec --full-auto is reintroduced), plus coverage-artifact, evidence-gate-consumer,
    approval-phase-gate, and semantic-gate-bash-route regression tests.

Gates: local-ci 83/83, full pytest + bun test (1009 Bun tests) + bash/Bun parity
green, 3-reviewer council unanimous APPROVE (after a round that correctly caught a
self-contradicting code comment, now fixed).

Breaking Changes

  • Removed unreachable logStub/noopCouncil/stubProvider fallbacks in `loki-ts/src/runner/autonomous.ts`; missing load‑bearing modules now cause an error at require time instead of silent incorrect builds.

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Track Loki Mode

Get notified when new releases ship.

Sign up free

About Loki Mode

Multi-agent autonomous SDLC framework. Spec to deployed app. PRD, GitHub issue, OpenAPI/JSON/YAML, or one-line brief. 5 AI providers, 11 quality gates.

All releases →

Related context

Beta — feedback welcome: [email protected]