ctxloom v1.7.13

ctxloom v1.7.13

This release repairs and rebaselines the external-oracle benchmark introduced in v1.7.0.

• Validates all 15 merged-PR pins before benchmarking.
• Requires merged PRs on each repository's current default branch and indexes the exact merge commit.
• Audits exact independently resolved local import edges for JavaScript/TypeScript, Python, and Go.
• Reports comparable corpus token totals using the production ceil(characters / 4) estimator.
• Adds an explicit benchmark-only snapshot version-mismatch opt-in while keeping normal product snapshot loading strict.

The clean post-merge corpus run reports F1 0.43, source recall 0.63, graph reachability 0.93, and exact symbol/import coverage 1.00. Modeled code context falls from 1,367,753 estimated tokens without ctxloom to 71,046 with ctxloom, a 94.8% saving (19.3x weighted; 27.0x mean per PR).

Validated with 1,452 passing tests, Node 20/22 CI, CodeQL, security checks, production builds, benchmark CI, and a fresh-install publish smoke test.

v1.7.1

What's Changed

• fix(pr-bot): trim Marketplace description to fit the 125-char limit by @kodiii in #250
• release(v1.7.1): pr-bot Marketplace description fix by @kodiii in #251

Full Changelog: v1.7.0...v1.7.1

v1.7.0

What's Changed

• feat(graph): assignment-symbol AST + importees + ranked symbol-callers (v1.6.0 spike) by @kodiii in #175
• feat(graph): specificity-weighted ranking + path-proximity + min-score floor by @kodiii in #176
• fix(bench): gate uses sourceRecall ≥ 0.8 instead of recall ≥ 0.9 by @kodiii in #177
• feat(graph): Python call-graph extraction (v1.6.1) by @kodiii in #178
• feat(graph): Python re-export tracing through barrel init.py by @kodiii in #179
• feat(bench): wire GitOverlayStore co-change signal into prediction by @kodiii in #180
• feat(graph): semantic-similarity signal as 6th impact-radius arm (gate-class) by @kodiii in #182
• fix(bench): gate uses OR criterion (closes v1.6.0 spike gate, PASSED) by @kodiii in #183
• bench(v1.6.0): full corpus selection + first publishable summary by @kodiii in #184
• feat(bench): add graphReachability metric — separates graph quality from algorithm quality by @kodiii in #185
• feat(bench): symbol declaration coverage = 100% across all 5 languages — direct graph-quality probe by @kodiii in #186
• feat(bench): import edge coverage audit — direct resolver-correctness probe by @kodiii in #187
• feat(install): operationalize Karpathy coding principles in CLAUDE.md + skills by @kodiii in #188
• fix(graph): Go resolver — fan out package imports to all sibling files by @kodiii in #189
• docs(bench): refresh summary.md — Go resolver fix lifts graphReachability 0.81→0.94 by @kodiii in #190
• feat(bench): token reduction column — naive baseline vs ctxloom skeleton (24.6× overall) by @kodiii in #191
• feat(embed): pluggable embedding model — opt-in jina-code upgrade by @kodiii in #192
• feat(extractor): add 5 languages — C/C++, Scala, Lua, Elixir, Zig (13 → 18) by @kodiii in #193
• bench(embed): jina-code micro-bench — +72.5% discrimination over minilm by @kodiii in #194
• feat(setup): add 4 hosts (Zed, Gemini CLI, Kiro, OpenCode) + Cursor project-path by @kodiii in #195
• fix(setup): Continue per-server YAML — corrected MCP config path by @kodiii in #196
• fix(setup): Codex TOML — corrected MCP config path + shared-file writer by @kodiii in #197
• feat(indexer): streaming file walk + batch upserts (monorepo support) by @kodiii in #198
• fix(watcher): sync chokidar ignore list with indexer — plug FD leak (task #13) by @kodiii in #199
• chore(deps): bump @protobufjs/utf8 from 1.1.0 to 1.1.1 by @dependabot[bot] in #200
• chore(deps): bump hono from 4.12.14 to 4.12.22 by @dependabot[bot] in #201
• chore(deps): bump protobufjs from 7.5.5 to 7.6.1 by @dependabot[bot] in #202
• perf(indexer): batch ONNX inference — 3-10× faster embedding by @kodiii in #204
• docs(v1.7.0): release notes + blog post + README marketing surface by @kodiii in #205
• chore(deps): bump postcss 8.5.9 → 8.5.15 (security patches) by @kodiii in #206
• chore(ci): tune Dependabot — split security from majors, ignore risky bumps by @kodiii in #207
• release(v1.7.0): bump version + refresh bench summary timestamp by @kodiii in #223
• chore(ci): cache + pre-warm MiniLM ONNX model — eliminate flaky test race by @kodiii in #225
• chore(ci): extend Dependabot ignores — zod / express / @types/node / @types/supertest by @kodiii in #224
• chore(deps)(deps): bump yaml from 2.8.3 to 2.9.0 by @dependabot[bot] in #222
• chore(dev-deps)(deps-dev): bump the dev-tooling group across 1 directory with 2 updates by @dependabot[bot] in #217
• chore(ci)(deps): bump docker/login-action from 3 to 4 by @dependabot[bot] in #208
• chore(ci)(deps): bump docker/setup-buildx-action from 3 to 4 by @dependabot[bot] in #209
• chore(ci)(deps): bump docker/build-push-action from 6 to 7 by @dependabot[bot] in #212
• chore(ci): round-2 Dependabot ignores — lancedb minor / react-router / recharts / jsdom / GH Actions majors by @kodiii in #239
• chore(deps)(deps): bump fast-uri from 3.1.0 to 3.1.2 by @dependabot[bot] in #228
• chore(ci): round-3 Dependabot ignores — tailwind 4 / arrow / open / @types/express / web-tree-sitter minor by @kodiii in #248
• chore(deps)(deps): bump express from 4.22.1 to 4.22.2 by @dependabot[bot] in #242
• chore(deps)(deps): bump @octokit/rest from 21.1.1 to 22.0.1 by @dependabot[bot] in #249

New Contributors

• @dependabot[bot] made their first contribution in #200

Full Changelog: v1.5.6...v1.7.0

v1.5.3

What's Changed

• chore: release v1.5.2 — dashboard Budget page by @kodiii in #160
• docs(install): Phase 0 host-adapter verification against vendor docs by @kodiii in #161
• bench(v1.6.0): scaffold honest-benchmark harness with spike gate by @kodiii in #162
• fix(ast): emit import nodes for CommonJS require() calls by @kodiii in #163
• bench: fix corpus + predict (PR #162 lost these via squash-merge) by @kodiii in #164
• fix(bench): prefer non-test source for entry-point selection by @kodiii in #165
• fix(graph): resolveImport must skip directories (require('..') bug) by @kodiii in #166
• fix(bench): define __dirname/__filename for ESM by @kodiii in #167
• fix(graph): resolve absolute Python imports (was eating first char) by @kodiii in #168
• fix(bench): index merge commit instead of parent (post-PR state) by @kodiii in #169
• feat(bench): add source-file recall — separate structural ceiling from graph quality by @kodiii in #170
• experiment(bench): calibrate to depth=1 (direct importers only) by @kodiii in #171
• chore: release v1.5.3 — silent-failure graph-quality hotfix by @kodiii in #172

Full Changelog: v1.5.2...v1.5.3

v1.5.2

What's Changed

• docs+chore: dogfood ctxloom init --host=all + document the --host flag by @kodiii in #158
• feat(dashboard): add Budget page for budget-telemetry events by @kodiii in #159

Full Changelog: v1.5.1...v1.5.2

v1.5.1

What's Changed

• chore: release v1.5.1 — v1.5.0 dogfood hardening cohort by @kodiii in #157

Full Changelog: v1.5.0...v1.5.1

v1.5.0

What's Changed

• feat(agent-harness): Phase 4a — server-enforced task-tool budget by @kodiii in #152
• feat(agent-harness): Phase 4b — telemetry-learned next_tool_suggestions (opt-in) by @kodiii in #153
• feat(agent-harness): Phase 4d — cross-agent host matrix by @kodiii in #154
• feat(agent-harness): Phase 4c — pr-bot suggests Phase 3 skills in review comments by @kodiii in #155
• chore: release v1.5.0 — Phase 4 agent-harness completion by @kodiii in #156

Full Changelog: v1.4.0...v1.5.0

v1.4.0

What's Changed

• chore: release v1.3.1 by @kodiii in #147
• feat(agent-harness): Phase 1 — self-guiding API (v1.4.0 candidate) by @kodiii in #148
• feat(agent-harness): Phase 2 — ctxloom init writes HMAC-signed agent rules + hooks by @kodiii in #149
• feat(agent-harness): Phase 3 — prepackaged Claude Code skills by @kodiii in #150
• chore: release v1.4.0 — Agent-First Harness by @kodiii in #151

Full Changelog: v1.3.1...v1.4.0

v1.3.1

What's Changed

• chore(pr-bot): pin example workflow to ctxloom-pro@1.3.0 (closes #107 follow-up) by @kodiii in #131
• test(pr-bot): pin-drift guard + README pin consistency (closes ARCH-131-1/2 + TEST-131-1) by @kodiii in #132
• test(pr-bot): live-registry drift detection for agent-spec tool references (closes TEST-131-3) by @kodiii in #133
• feat(agents): opt the 4 specialist specs into the Phase B budget surface (closes Part 1 of #107 follow-up) by @kodiii in #134
• feat(budget): persistent event collector + ctxloom budget-stats CLI by @kodiii in #135
• fix(tests): scope Budget.test.ts emitTelemetry tests to a temp telemetry dir (closes TEST-135-2 + PERF-135-1) by @kodiii in #136
• refactor(budget): dedup percentile + mark internal helpers @internal (closes ARCH-135-2 + ARCH-135-3) by @kodiii in #137
• test(cli): integration tests for ctxloom budget-stats + add to license-gate bypass (closes TEST-135-3) by @kodiii in #138
• refactor(budget): injectable TelemetrySink (ARCH-135-1) by @kodiii in #139
• test+fix(budget): harden TelemetrySink contract (M1+L3+L4 from #139 dogfood) by @kodiii in #140
• feat(budget): thread TelemetrySink via ServerContext (closes #141) by @kodiii in #145
• harden(eventCollector): sanitize env path + once-flag warn + NaN dates (closes #142, #143, #144) by @kodiii in #146

Full Changelog: v1.3.0...v1.3.1

v1.3.0

What's Changed

• docs: add Response Budgets section (B2 closeout for #106) by @kodiii in #128
• test(budget): drift detection for README defaults table (closes TEST-128-1) by @kodiii in #129
• docs: B3 release-prep — CHANGELOG (1.2.6, 1.2.7, 1.3.0 draft) + skeleton-first explainer + AI-REVIEWS budget section by @kodiii in #130

Full Changelog: v1.2.7...v1.3.0