Repository Analysis

JuliusBrussee/caveman

🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman

4.9 Likely human-written View on GitHub

Analysis Overview

This report presents the forensic synthetic code analysis of JuliusBrussee/caveman, a Go project with 101,595 GitHub stars. SynthScan v2.0 examined 244,489 lines of code across 1000 source files, recording 866 pattern matches distributed across 17 syntactic categories. The overall adjusted score of 4.9 places this repository in the Likely human-written band.

The scanner applied 160+ deterministic lexical heuristics, multi-line block detectors, abstract syntax tree depth profilers, and a cross-file Jaccard similarity matrix to construct a statistically normalised synthetic code estimate. All matches are individually weighted by severity coefficient and contextual multiplier before summation, and the resulting headline score is temporally discounted to account for the repository's development history relative to the commercial emergence of large language model coding tooling (November 2022 onward).

4.9
Adjusted Score
4.9
Raw Score
100%
Time Factor
2026-08-27
Last Push
101.6K
Stars
Go
Language
244.5K
Lines of Code
1.0K
Files
866
Pattern Hits
2026-08-29
Scan Date
0.04
HC Hit Rate

What These Metrics Mean

Adjusted Score
Primary synthetic code indicator. Raw score normalised per 1,000 lines of code and multiplied by the temporal discount factor. This is the definitive comparative metric — use it to rank repositories by AI authorship density.
Raw Score
The unmodified sum of all severity-weighted, context-multiplied pattern match scores before temporal discounting. Reflects the absolute signal strength independent of when the repository was last active.
Time Factor
The temporal discount multiplier (0–100%) applied to the raw score. Repositories last updated before ChatGPT's launch (Nov 2022) receive a 5% factor. Full signal is only assigned to repositories active in the post-adoption era (Jan 2024+).
Pattern Hits
Total count of individual pattern matches across all files and categories. A high hit count with a low score may indicate a very large codebase with isolated AI snippets; a low count with a high score indicates dense, concentrated AI signatures.
HC Hit Rate
High+Critical pattern hits per file, averaged across the repository. This orthogonal signal catches repositories where a few files are densely packed with high-severity AI tells — a strong indicator even when the normalised score appears moderate due to codebase size.
Lines of Code / Files
Total lines and files analysed. The scanner examines 94 file extensions. These denominators are used to normalise the score, enabling fair comparison between repositories of vastly different sizes.

Score History

This chart maps the temporal evolution of the adjusted synthetic code score across successive scan runs. An upward trajectory indicates ongoing incorporation of AI-generated code or expanding LLM-assisted scaffolding; a stable or declining trajectory may reflect active human refactoring, code removal, or the adoption of stricter authorship policies. The dashed secondary line (right axis) independently tracks total raw pattern hit count, which can diverge from the normalised score when codebase size changes significantly between scans.

Severity Breakdown

Classifies detected patterns by their diagnostic confidence and structural impact. CRITICAL patterns (coefficient 10) represent definitive synthetic signatures — hallucinated imports, explicit LLM attribution metadata — virtually never produced by human authors. HIGH (5) indicates strong structural tells such as cross-file repetition or cross-linguistic idioms. MEDIUM (2) covers recognisable conversational padding and AI-specific vocabulary. LOW (1) captures subtle indicators like tautological comments and generic boilerplate that require density to carry independent signal.

CRITICAL 0HIGH 38MEDIUM 132LOW 696

Directory Score Breakdown

This horizontal bar chart decomposes the repository's raw synthetic code score by top-level directory, allowing you to pinpoint precisely which modules or components carry the highest AI authorship density. Directories with disproportionately high scores relative to their size warrant targeted manual review: concentrated AI signatures often trace back to mass-generated configuration layers, auto-ported test suites, LLM-scaffolded boilerplate classes, or entire subsystems authored under heavy copilot assistance. Use this view to prioritise your human code-review effort.

Pattern Findings

The scanner identified 866 distinct pattern matches across 17 syntactic categories. Each entry below represents a discrete location in the source code where the engine recorded a statistically significant AI authorship indicator. Expand any category row to inspect the individual file paths, line numbers, code snippets, and the lexical context (CODE, COMMENT, or STRING) in which each match was detected.

Reading the findings table: The Severity column indicates the diagnostic confidence level (CRITICAL / HIGH / MEDIUM / LOW). The Context column identifies whether the match occurred inside executable code, an inline comment, or a string literal — comment-context matches receive a ×1.5 weight because LLMs systematically over-annotate. The ⚡ bolt icon marks clustered matches: three or more patterns within a 10-line window, each receiving an additional ×1.5 density multiplier as dense clusters constitute far stronger evidence of synthetic authorship than isolated hits.

Decorative Section Separators104 hits · 301 pts
SeverityFileLineSnippetContext
MEDIUMextension/popup.js22// ── pixel-flame logo (the CaveMark, 8×8) — white on onyx, no colour ─────────COMMENT
MEDIUMextension/popup.js44// ── settings wiring ──────────────────────────────────────────────────────COMMENT
MEDIUMproxy/providers/gemini/content_compress_test.go56// ── live-zone cost ────────────────────────────────────────────────────────────COMMENT
MEDIUMproxy/providers/openai/content_compress_test.go163// ── live-zone cost ────────────────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js79// ── Argv ───────────────────────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js181// ── Color helpers ──────────────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js191// ── Env guards ─────────────────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js215// ── Provider matrix ────────────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js287// ── Detection ─────────────────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js392// ── Repo root resolution ───────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js405// ── Run helpers ────────────────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js464// ── Per-provider installers ────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js630// ── hermes native install ──────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js692// ── opencode native install ───────────────────────────────────────────────COMMENT
MEDIUMbin/install.js959// ── OpenClaw native install ───────────────────────────────────────────────COMMENT
MEDIUMbin/install.js991// ── Hooks installer ────────────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js1121// ── MCP shrink wiring ─────────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js1158// ── Init writers (per-repo rule files) ────────────────────────────────────COMMENT
MEDIUMbin/install.js1197// ── HTTPS download via stdlib ─────────────────────────────────────────────COMMENT
MEDIUMbin/install.js1223// ── Integrity verification for downloaded hooks (#262) ─────────────────────COMMENT
MEDIUMbin/install.js1223// ── Integrity verification for downloaded hooks (#262) ─────────────────────COMMENT
MEDIUMbin/install.js1253// ── Uninstall ─────────────────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js1482// ── Interactive prompt (TTY-only) ─────────────────────────────────────────COMMENT
MEDIUMbin/install.js1500// ── --list ─────────────────────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js1518// ── Help ───────────────────────────────────────────────────────────────────COMMENT
MEDIUMbin/install.js1569// ── Main ───────────────────────────────────────────────────────────────────COMMENT
MEDIUMbin/lib/openclaw.js140// ── Frontmatter helpers ───────────────────────────────────────────────────COMMENT
MEDIUMbin/lib/openclaw.js178// ── Bootstrap snippet load ────────────────────────────────────────────────COMMENT
MEDIUMbin/lib/openclaw.js215// ── SOUL.md marker-block append/strip ─────────────────────────────────────COMMENT
MEDIUMbin/lib/openclaw.js311// ── Public API ────────────────────────────────────────────────────────────COMMENT
MEDIUMbin/lib/settings.js26// ── stripJsonComments ──────────────────────────────────────────────────────COMMENT
MEDIUMbin/lib/settings.js64// ── stripTrailingCommas ────────────────────────────────────────────────────COMMENT
MEDIUMbin/lib/settings.js95// ── readSettings ───────────────────────────────────────────────────────────COMMENT
MEDIUMbin/lib/settings.js124// ── writeSettings ──────────────────────────────────────────────────────────COMMENT
MEDIUMbin/lib/settings.js139// ── validateHookFields ────────────────────────────────────────────────────COMMENT
MEDIUMbin/lib/settings.js161// ── Idempotency probe ──────────────────────────────────────────────────────COMMENT
MEDIUMbin/lib/settings.js171// ── addCommandHook ────────────────────────────────────────────────────────COMMENT
MEDIUMbin/lib/settings.js190// ── Managed hook scripts ──────────────────────────────────────────────────COMMENT
MEDIUMbin/lib/settings.js229// ── removeCavemanHooks ────────────────────────────────────────────────────COMMENT
MEDIUMbin/lib/settings.js258// ── rewriteLegacyManagedHookCommands ──────────────────────────────────────COMMENT
MEDIUMbin/lib/settings.js290// ── pruneOrphanedManagedHooks ─────────────────────────────────────────────COMMENT
MEDIUMbin/lib/settings.js376// ── claudeConfigDir ───────────────────────────────────────────────────────COMMENT
MEDIUMtests/test_caveman_stats.js464// ── Output-reduction share (never a "usage"/"budget" claim) ────────────────COMMENT
MEDIUMtests/test_caveman_stats.js536// ── Mid-session mode-change attribution (#601) ─────────────────────────────COMMENT
MEDIUMtests/test_caveman_stats.js536// ── Mid-session mode-change attribution (#601) ─────────────────────────────COMMENT
MEDIUMtests/test_caveman_stats.js637// ── Rule-overhead + net (#145/#677) ────────────────────────────────────────COMMENT
MEDIUMtests/test_caveman_stats.js637// ── Rule-overhead + net (#145/#677) ────────────────────────────────────────COMMENT
MEDIUMtests/test_mcp_shrink.js182// ── Packaging (#597) ────────────────────────────────────────────────────────COMMENT
MEDIUMtests/test_mcp_shrink.js182// ── Packaging (#597) ────────────────────────────────────────────────────────COMMENT
MEDIUMtests/test_repo_local_config.js159// ── #634: optional startDir param (backward compatible) ────────────────────COMMENT
MEDIUMtests/test_repo_local_config.js159// ── #634: optional startDir param (backward compatible) ────────────────────COMMENT
MEDIUMtests/test_mode_tracker.py96 # ── #598: deactivation word orders ──────────────────────────────────COMMENT
MEDIUMtests/test_mode_tracker.py135 # ── #598: activation guards ─────────────────────────────────────────COMMENT
MEDIUMtests/test_mode_tracker.py174 # ── slash commands ──────────────────────────────────────────────────COMMENT
MEDIUMtests/test_mode_tracker.py185 # ── #599: one-shot independent modes ────────────────────────────────COMMENT
MEDIUMtests/test_mode_tracker.py58 # ── hook budget: act on the first complete payload, not on EOF ──────COMMENT
MEDIUMtests/test_cavecrew_model_overrides.js30// ── patchFrontmatterModel ──────────────────────────────────────────────────COMMENT
MEDIUMtests/test_cavecrew_model_overrides.js155// ── resolvePluginRoot ──────────────────────────────────────────────────────COMMENT
MEDIUMtests/test_cavecrew_model_overrides.js165// ── applyOverrides ─────────────────────────────────────────────────────────COMMENT
MEDIUMtests/test_cavecrew_model_overrides.js283// ── Summary ────────────────────────────────────────────────────────────────COMMENT
44 more matches not shown…
Hyper-Verbose Identifiers320 hits · 276 pts
SeverityFileLineSnippetContext
LOWbin/lib/opencode-agent.js56function transformOpencodeAgentFrontmatter(content) {CODE
LOWbin/lib/settings.js264function rewriteLegacyManagedHookCommands(settings, absoluteNode, platform = process.platform) {CODE
LOWbin/lib/settings.js307function pruneOrphanedManagedHooks(settings, configDir, platform = process.platform) {CODE
LOW…ns/caveman/skills/caveman-compress/scripts/validate.py112def extract_indented_code_blocks(text):CODE
LOWtests/verify_repo.py116def verify_shipped_skills_are_documented() -> None:CODE
LOWtests/verify_repo.py157def verify_skill_frontmatter_upload_compatibility() -> None:CODE
LOWtests/verify_repo.py239def verify_manifests_and_syntax() -> None:CODE
LOWtests/verify_repo.py638def verify_license_boundaries() -> None:CODE
LOWtests/verify_repo.py672def verify_untrusted_git_invocations() -> None:CODE
LOWtests/test_detect.py68 def test_known_names_case_insensitive(self):STRING
LOWtests/test_detect.py74 def test_cmakelists_txt_not_compressible_despite_txt_extension(self):STRING
LOWtests/test_detect.py80 def test_shebang_script_is_code(self):STRING
LOWtests/test_detect.py86 def test_extensionless_prose_still_compressible(self):STRING
LOWtests/test_detect.py92 def test_markdown_still_compressible(self):STRING
LOWtests/test_mode_tracker.py98 def test_turn_caveman_mode_off_deactivates(self):CODE
LOWtests/test_mode_tracker.py104 def test_turn_caveman_off_deactivates(self):CODE
LOWtests/test_mode_tracker.py109 def test_turn_off_caveman_deactivates(self):CODE
LOWtests/test_mode_tracker.py114 def test_stop_caveman_multiline_deactivates(self):CODE
LOWtests/test_mode_tracker.py120 def test_normal_mode_command_deactivates(self):CODE
LOWtests/test_mode_tracker.py125 def test_back_to_normal_mode_deactivates(self):CODE
LOWtests/test_mode_tracker.py130 def test_vim_normal_mode_does_not_deactivate(self):CODE
LOWtests/test_mode_tracker.py137 def test_enable_caveman_with_stop_elsewhere_activates(self):CODE
LOWtests/test_mode_tracker.py144 def test_question_does_not_activate(self):CODE
LOWtests/test_mode_tracker.py150 def test_scoped_brevity_does_not_activate(self):CODE
LOWtests/test_mode_tracker.py154 def test_unscoped_brevity_activates(self):CODE
LOWtests/test_mode_tracker.py158 def test_activate_caveman_still_works(self):CODE
LOWtests/test_mode_tracker.py162 def test_turn_on_caveman_mode_still_works(self):CODE
LOWtests/test_mode_tracker.py166 def test_talk_like_caveman_still_works(self):CODE
LOWtests/test_mode_tracker.py170 def test_bare_caveman_mode_still_works(self):CODE
LOWtests/test_mode_tracker.py176 def test_slash_caveman_level_switch(self):CODE
LOWtests/test_mode_tracker.py187 def test_commit_restores_prior_level_on_next_prompt(self):CODE
LOWtests/test_mode_tracker.py195 def test_commit_with_no_prior_mode_deactivates_after(self):CODE
LOWtests/test_mode_tracker.py202 def test_chained_independent_modes_keep_original_prev(self):CODE
LOWtests/test_mode_tracker.py210 def test_namespaced_commit_and_review_recognized(self):CODE
LOWtests/test_mode_tracker.py219 def test_no_reinforcement_during_independent_turn(self):CODE
LOWtests/test_mode_tracker.py224 def test_deactivation_clears_saved_prev(self):CODE
LOWtests/test_mode_tracker.py60 def test_acts_on_first_complete_payload_without_waiting_for_eof(self):CODE
LOWtests/test_hooks.py40 def test_install_upgrades_old_two_file_install(self):CODE
LOWtests/test_hooks.py61 def test_install_reconfigures_missing_statusline(self):CODE
LOWtests/test_hooks.py108 def test_uninstall_preserves_custom_statusline(self):CODE
LOWtests/test_hooks.py159 def test_activate_does_not_nudge_when_custom_statusline_exists(self):CODE
LOWtests/test_hooks.py184 def test_activate_emits_skill_md_not_fallback_from_repo_layout(self):CODE
LOWtests/test_hooks.py197 def test_activate_finds_skill_beside_config_dir_hooks(self):CODE
LOWtests/test_hooks.py218 def test_activate_prefers_claude_plugin_root(self):CODE
LOWtests/test_compress_safety.py46 def test_empty_compressed_output_does_not_touch_disk(self):CODE
LOWtests/test_compress_safety.py56 def test_whitespace_only_compressed_output_does_not_touch_disk(self):CODE
LOWtests/test_compress_safety.py66 def test_identical_compressed_output_does_not_touch_disk(self):CODE
LOWtests/test_compress_safety.py76 def test_real_compression_writes_backup_and_target(self):CODE
LOWtests/test_compress_safety.py117 def test_write_text_atomic_leaves_destination_untouched_on_encode_failure(self):CODE
LOWtests/test_compress_safety.py135 def test_forced_primary_write_failure_leaves_original_and_backup_intact(self):CODE
LOWtests/test_compress_safety.py168 def test_permission_preserved_across_compression(self):CODE
LOWtests/test_compress_safety.py183 def test_retry_preamble_output_rejected_and_not_written(self):CODE
LOWtests/test_compress_safety.py233 def test_crlf_line_endings_survive_the_round_trip(self):CODE
LOWtests/test_compress_safety.py253 def test_one_crlf_line_does_not_convert_an_lf_document(self):CODE
LOWtests/test_compress_safety.py290 def test_two_separate_blocks_are_left_alone(self):CODE
LOWtests/test_compress_safety.py294 def test_a_real_wrapper_is_stripped(self):CODE
LOWtests/test_compress_safety.py298 def test_a_longer_wrapper_around_inner_fences_is_stripped(self):CODE
LOWtests/test_validate_inline.py26 def test_indented_fence_markers_not_leaked_as_inline(self):CODE
LOWtests/test_validate_inline.py29 def test_nested_fence_file_validates_against_itself(self):CODE
LOWtests/test_validate_inline.py34 def test_deeply_indented_fence_markers_not_leaked(self):CODE
260 more matches not shown…
Over-Commented Block266 hits · 257 pts
SeverityFileLineSnippetContext
LOWinstall.sh1#!/usr/bin/env bashCOMMENT
LOWrewriter/rewriter.go1// Package rewriter is the aged-zone reflection rewriter: an external LLMCOMMENT
LOWrewriter/rewriter.go21// not.COMMENT
LOWrewriter/rewriter.go61COMMENT
LOWrewriter/rewriter.go101 WindowBytes [][]byte // steps s-3..s serialized, per the paper's a+b windowCOMMENT
LOWrewriter/tokens.go1package rewriterCOMMENT
LOWrewriter/gate.go1package rewriterCOMMENT
LOWrewriter/gate.go41 "warn", "err!", "not found", "no such file", "timed out", "timeout",COMMENT
LOWrewriter/gate.go61COMMENT
LOWrewriter/gate.go81// on failure lines.COMMENT
LOWrewriter/prompt.go1package rewriterCOMMENT
LOWcacheengine/types.go141COMMENT
LOWcacheengine/doc.go1// Package cacheengine plans and applies provider-native prompt caching throughCOMMENT
LOWproxy/doc.go1// Package proxy is the public, byte-safe Caveman reverse proxy: the `caveman`COMMENT
LOWproxy/cmd/caveman-proxy/main.go1// Command caveman-proxy is the standalone byte-safe proxy binary. `caveman start`COMMENT
LOWproxy/providers/toolzone.go1package providersCOMMENT
LOWproxy/providers/adapter.go41COMMENT
LOWproxy/providers/adapter.go61 // Scheme records how the key arrived when it was passed through from theCOMMENT
LOWproxy/providers/adapter.go161COMMENT
LOWproxy/providers/adapter.go241 // the cache fields, message_delta carries ONLY the final output_tokens —COMMENT
LOWproxy/providers/adapter.go261 RawUsage json.RawMessageCOMMENT
LOWproxy/providers/adapter.go321 // body upstream. ok=false opts a provider/route out, so compress mode passes theCOMMENT
LOWproxy/providers/adapter.go361// injected tool, `mcp__caveman__caveman_retrieve` under Claude Code's MCPCOMMENT
LOWproxy/providers/adapter.go1081 case int64:COMMENT
LOWproxy/providers/adapter.go1101//COMMENT
LOWproxy/providers/gemini/content_compress.go81 }COMMENT
LOWproxy/providers/gemini/explicit_cache.go21// ApplyProviderNativeTransforms is an unconditional byte-safe pass-through.COMMENT
LOWproxy/providers/bedrock/cache_points.go81COMMENT
LOWproxy/providers/bedrock/cache_points.go101 if enabled, ok := entry.Capabilities["prompt_cache"].(bool); ok && enabled {COMMENT
LOWproxy/providers/bedrock/bedrock.go1package bedrockCOMMENT
LOWproxy/providers/bedrock/usage.go21COMMENT
LOWproxy/providers/anthropic/content_compress.go101// compression) plus the frozen user/system blocks at or below the cache_controlCOMMENT
LOWproxy/providers/anthropic/breakpoint_plan.go21COMMENT
LOWproxy/providers/anthropic/breakpoint_plan.go41// changes no model-visible bytes; on any parse anomaly the body is left unchanged.COMMENT
LOWproxy/providers/anthropic/breakpoint_plan.go241// The documented semantics (platform.claude.com/docs/en/build-with-claude/prompt-caching):COMMENT
LOWproxy/providers/anthropic/cogs_bench_test.go1package anthropicCOMMENT
LOWproxy/providers/vertex/vertex.go1package vertexCOMMENT
LOWproxy/providers/vertex/vertex.go21// thoughtsTokenCount) and Claude snake_case (usage.input_tokens/COMMENT
LOWproxy/providers/openai/cache_key.go21 providers.BaseCOMMENT
LOWproxy/providers/openai/count_tokens.go61// unprojectableResponsesFields name SERVER-HELD prompt surface: content theCOMMENT
LOWproxy/providers/openai/count_tokens.go81//COMMENT
LOWproxy/providers/openai/reasoning_effort.go41//COMMENT
LOWproxy/providers/openai/reasoning_effort.go141 }COMMENT
LOWproxy/providers/openai/session_cache_key.go21// sessionCacheKeyDomain separates this hash from any other sha256 of the sameCOMMENT
LOWproxy/providers/openai/output_brevity.go1package openaiCOMMENT
LOWproxy/providers/openai/stream_usage.go1package openaiCOMMENT
LOWproxy/providers/openai/stream_usage.go21// OpenAI-compatible client already knows to either consume or ignore anCOMMENT
LOWproxy/providers/openai/stream_usage.go61 }COMMENT
LOWproxy/internal/standalone/standalone.go121type Options struct {COMMENT
LOWproxy/internal/config/config.go21COMMENT
LOWproxy/internal/config/config.go41 // "off" disables it; unknown values fail closed to "off". There isCOMMENT
LOWproxy/internal/nativeruntime/runtime.go1161// elision engine keeps the head and tail rows, keeps one representative of everyCOMMENT
LOWproxy/internal/gateway/server.go1// Package gateway is the byte-safe request lifecycle of the public standaloneCOMMENT
LOWproxy/internal/gateway/server.go81// original block for CCR recovery and returns its deterministic content-addressedCOMMENT
LOWproxy/internal/gateway/server.go101// deterministic; empty query must match CompressSegment behavior.COMMENT
LOWproxy/internal/gateway/server.go121// recovery row is ever written — the estimate is a pure measurement of whatCOMMENT
LOWproxy/internal/gateway/server.go141// the implementation may return only the elided sections most relevant to theCOMMENT
LOWproxy/internal/gateway/server.go161// rewrite it could not reproduce on the next turn.COMMENT
LOWproxy/internal/gateway/server.go261 // only a prefix; those rows keep hashes empty and mark this false.COMMENT
LOWproxy/internal/gateway/server.go301 // on compress candidates), and unlike cacheguard it tolerates append-onlyCOMMENT
206 more matches not shown…
Cross-Language Confusion (JS/TS)35 hits · 160 pts
SeverityFileLineSnippetContext
HIGHpackages/cli/src/index.ts7117 return NoneCODE
HIGHpackages/cli/src/index.ts7121 return NoneCODE
HIGHpackages/cli/src/index.ts8914 print({ logged_out: true, remote_session_revoked: true, external_token_cleared: false, external_token_source: "CAVE_TCODE
HIGHpackages/cli/src/index.ts8923 print({ logged_out: true });CODE
HIGHpackages/cli/src/index.ts12081 return FalseCODE
HIGHpackages/cli/src/index.ts12092 return FalseCODE
HIGHpackages/cli/src/index.ts12098 return NoneCODE
HIGHpackages/cli/src/index.ts12101 return NoneCODE
HIGHpackages/cli/src/index.ts12103 return NoneCODE
HIGHpackages/cli/src/index.ts12105 return NoneCODE
HIGHpackages/cli/src/index.ts12114 return NoneCODE
HIGHpackages/cli/src/index.ts12117 return NoneCODE
HIGHpackages/cli/src/index.ts692 print({CODE
HIGHpackages/cli/src/index.ts712 print({ telemetry: "on", anonymous_id: anonymousId, note: "env override active (DO_NOT_TRACK/CAVEMAN_TELEMETRY) — noCODE
HIGHpackages/cli/src/index.ts715 print({ telemetry: "on", anonymous_id: anonymousId });CODE
HIGHpackages/cli/src/index.ts735 print({ telemetry: "off", anonymous_id: "none" });CODE
HIGHpackages/cli/src/index.ts1692 print({ skills: AGENT_SKILL_METADATA, suites: AGENT_SKILL_SUITES });CODE
HIGHpackages/cli/src/index.ts2370 print({CODE
HIGHpackages/cli/src/index.ts2980 print({CODE
HIGHpackages/cli/src/index.ts7097 return FalseCODE
HIGHpackages/cli/src/index.ts7100 return FalseCODE
HIGHpackages/cli/src/index.ts8051 print({ ...result, repair: result.components.shared_runtime ? "caveman start" : "caveman setup --install", trust: "nCODE
HIGHpackages/cli/src/index.ts8073 print({CODE
HIGHpackages/cli/src/index.ts8852 print({ authenticated: true, baseURL, gateway_url: gateway || null, organization_id: organizationId ?? null, tokenCODE
HIGHpackages/cli/src/index.ts11201 print(receipt);CODE
HIGHpackages/cli/src/index.ts11265 print(explanation);CODE
HIGHpackages/cli/src/index.ts13892 print({ linked: "codex", source: "local_rate_limits", token_store: "none", refresh });CODE
HIGHpackages/cli/src/index.ts13920 print({ linked: "claude", token_store: tokenStore, basis: "linked_api", refresh });CODE
HIGHpackages/cli/src/index.ts14501 print({ ...view, native_integrations: integrations });CODE
HIGHpackages/cli/src/index.ts14523 print({CODE
HIGHpackages/cli/src/index.ts14549 print(body);CODE
HIGHpackages/cli/src/index.ts14581 print(await response.json());CODE
HIGHpackages/cli/src/index.ts14628 print(body);CODE
HIGHpackages/cli/src/index.ts14933 print({CODE
HIGHpackages/cli/src/index.ts14972 print({ exported: out, receipts: Array.isArray(bundle.receipts) ? bundle.receipts.length : 0 });CODE
Unused Imports47 hits · 44 pts
SeverityFileLineSnippetContext
LOWevals/measure.py17CODE
LOWevals/plot.py12CODE
LOWevals/llm_run.py27CODE
LOW…eengine/cachebench/scripts/lmcache_parquet_to_jsonl.py4CODE
LOWtests/verify_repo.py4CODE
LOWtests/verify_repo.py398CODE
LOWtests/verify_repo.py400CODE
LOWshared/provider-catalog/validate_catalog.py2CODE
LOWpackages/sdk/python/tests/test_trace_continuity.py3CODE
LOWpackages/sdk/python/tests/test_context_pack.py1CODE
LOWpackages/sdk/python/tests/test_exporter.py1CODE
LOWpackages/sdk/python/tests/test_sdk.py1CODE
LOWpackages/sdk/python/tests/test_shared_context.py3CODE
LOWpackages/sdk/python/tests/test_runtime_policy.py16CODE
LOWpackages/sdk/python/tests/test_runtime_policy.py23CODE
LOWpackages/sdk/python/tests/test_parity.py12CODE
LOWpackages/sdk/python/tests/test_tool_events.py3CODE
LOWpackages/sdk/python/tests/test_task_profile.py1CODE
LOWpackages/sdk/python/tests/test_assembly.py1CODE
LOWpackages/sdk/python/tests/test_structural.py6CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py1CODE
LOWpackages/sdk/python/caveman_cloud/core.py1CODE
LOWmem/py/cavemem.py8CODE
LOWmem/py/tests/test_cavemem.py1CODE
Excessive Try-Catch Wrapping42 hits · 41 pts
SeverityFileLineSnippetContext
LOWevals/measure.py33 except Exception:CODE
LOWevals/plot.py29 except Exception:CODE
LOWevals/llm_run.py44 except Exception:CODE
LOWevals/llm_run.py84 except Exception:CODE
MEDIUMevals/llm_run.py77def claude_version() -> str:CODE
LOWplugins/caveman/skills/caveman-compress/scripts/cli.py19 except Exception:CODE
LOWplugins/caveman/skills/caveman-compress/scripts/cli.py79 except Exception as e:CODE
LOW…ns/caveman/skills/caveman-compress/scripts/compress.py24 except Exception:CODE
LOW…ns/caveman/skills/caveman-compress/scripts/compress.py190 except Exception:CODE
LOW…ns/caveman/skills/caveman-compress/scripts/compress.py254 except Exception:CODE
LOWtests/verify_repo.py24 except Exception:CODE
LOWtests/test_mode_tracker.py91 except Exception:CODE
LOWbenchmarks/run.py21 except Exception:CODE
MEDIUMpackages/cli/src/index.ts7044def _digest(value):CODE
MEDIUMpackages/cli/src/index.ts7052def _task_type(value):CODE
MEDIUMpackages/cli/src/index.ts7076def _task_terms(value):CODE
MEDIUMpackages/cli/src/index.ts7104def _call(event, payload=None):CODE
MEDIUMpackages/cli/src/index.ts12071def _int_env(name, default):CODE
MEDIUMpackages/cli/src/index.ts12078def _eligible(command):CODE
MEDIUMpackages/cli/src/index.ts12095def _transform_terminal_output(command=None, output=None, **kwargs):CODE
LOWpackages/cli/src/index.ts7120 except Exception:CODE
LOWpackages/cli/src/index.ts12074 except Exception:CODE
LOWpackages/cli/src/index.ts12091 except Exception:CODE
LOWpackages/cli/src/index.ts12116 except Exception:CODE
LOWpackages/cli/src/index.ts7048 except Exception:CODE
LOWpackages/cli/src/index.ts7055 except Exception:CODE
LOWpackages/cli/src/index.ts7079 except Exception:CODE
LOWpackages/sdk/python/caveman_cloud/core.py804 except Exception: # noqa: BLE001 — fail-closed: any failure ⇒ pass-through.CODE
LOWpackages/sdk/python/caveman_cloud/core.py1009 except Exception: # noqa: BLE001 — lossy selector must fail closed to caller-owned context.CODE
LOWpackages/sdk/python/caveman_cloud/core.py1925 except Exception: # noqa: BLE001 — a verifier must fail closed, never raise.CODE
LOWpackages/sdk/python/caveman_cloud/core.py2141 except Exception: # noqa: BLE001 — observability must never break a decision.CODE
LOWpackages/sdk/python/caveman_cloud/core.py2297 except Exception: # noqa: BLE001 — the agent's path must not depend on this call.CODE
LOWpackages/sdk/python/caveman_cloud/core.py2303 except Exception: # noqa: BLE001CODE
LOWpackages/sdk/python/caveman_cloud/core.py2356 except Exception: # noqa: BLE001CODE
LOWpackages/sdk/python/caveman_cloud/core.py2386 except Exception: # noqa: BLE001CODE
LOWpackages/sdk/python/caveman_cloud/core.py2418 except Exception: # noqa: BLE001 — decide() is total: a bug here must not break the agent.CODE
LOWpackages/sdk/python/caveman_cloud/core.py2422 except Exception: # noqa: BLE001 — observability never breaks a decision.CODE
LOWskills/caveman-compress/scripts/cli.py19 except Exception:CODE
LOWskills/caveman-compress/scripts/cli.py79 except Exception as e:CODE
LOWskills/caveman-compress/scripts/compress.py24 except Exception:CODE
LOWskills/caveman-compress/scripts/compress.py190 except Exception:CODE
LOWskills/caveman-compress/scripts/compress.py254 except Exception:CODE
Synthetic Comment Markers3 hits · 34 pts
SeverityFileLineSnippetContext
HIGHpackages/cli/src/index.ts3703// - valid entitlement (now < expires_at): compression as requested, "entitled".COMMENT
HIGHpackages/cli/src/index.ts3704// - lapsed ≤7 days: compression as requested, "grace" (caller kicks a refresh).COMMENT
HIGHpackages/cli/src/index.ts3705// - no/expired/unreadable entitlement: compression as requested, "unentitled".COMMENT
Modern AI Meta-Vocabulary12 hits · 32 pts
SeverityFileLineSnippetContext
MEDIUMcacheengine/cachebench/README.md69[LMCache Agentic Traces](https://huggingface.co/datasets/sammshen/lmcache-agentic-traces)CODE
MEDIUM…hebench/results/lmcache-agentic-traces-2026-08-10.json8 "url": "https://huggingface.co/datasets/sammshen/lmcache-agentic-traces",CODE
MEDIUM…ine/cachebench/scripts/fetch_lmcache_agentic_traces.sh6cachebench_base_url="https://huggingface.co/datasets/sammshen/lmcache-agentic-traces/resolve/${cachebench_revision}/dataCODE
MEDIUMproxy/providers/gemini/explicit_cache.go12// runtime scaffold. It is not a current policy flag or optimization label.COMMENT
MEDIUMproxy/providers/anthropic/cogs_bench_test.go7// body, which prices a million agentic requests at roughly EUR 0.13 of ScalewayCOMMENT
MEDIUMproxy/providers/anthropic/cogs_bench_test.go43// message history — the shape an agentic caller actually sends.COMMENT
MEDIUMproxy/internal/gateway/proxy.go1489// The historical gemini-explicit-cache scaffold is permanently absent: it neverCOMMENT
MEDIUMshared/platform/optimizers/optimizers_test.go86 // gemini-explicit-cache was a runtime-policy scaffold, never a money-bearingCOMMENT
MEDIUMshared/platform/optimizers/optimizers.go27// telemetry join key. The old scaffold never created a cachedContents resource,COMMENT
MEDIUMpackages/cli/src/agent-skills.generated.ts17 "caveman-manage": "---\nname: caveman-manage\ndescription: >\n Inspect Caveman Cloud's eval-gated experiment lifecyclCODE
MEDIUMsrc/hooks/caveman-stats.js388// limit usage — input + cache tokens dominate agentic sessions, count againstCOMMENT
MEDIUMsrc/hooks/caveman-stats.js514 // on the line above, and input + cache tokens (which dominate agenticCOMMENT
AI Slop Vocabulary7 hits · 18 pts
SeverityFileLineSnippetContext
MEDIUMbenchmarks/run.py61# the eval harness uses (evals/llm_run.py TERSE_PREFIX) so the two agree.COMMENT
MEDIUMpackages/cli/src/index.ts10666// CLI, robust to PATH: a resolved `caveman`/`cave`, else this very script's node.COMMENT
MEDIUMengine/compressors/json.go54 anomalyZ float64 // robust |z| (MAD-based) at/above which an item is an anomalyCODE
MEDIUMengine/compressors/json.go174 // Numeric anomalies (robust MAD-z per numeric leaf path + element length).COMMENT
MEDIUMengine/compressors/json.go329// computes a robust MAD-based z over the elements that have that path and flagsCOMMENT
MEDIUMengine/compressors/json.go388// medianAndScale returns the median and a robust scale (MAD) of xs, falling backCOMMENT
MEDIUMengine/compressors/json.go575// and accepts it when the mean shift across k exceeds zThresh robust scales.COMMENT
Deep Nesting9 hits · 8 pts
SeverityFileLineSnippetContext
LOW…ns/caveman/skills/caveman-compress/scripts/validate.py112CODE
LOW…gins/caveman/skills/caveman-compress/scripts/detect.py56CODE
LOWtests/verify_repo.py82CODE
LOWpackages/sdk/python/tests/test_exporter.py20CODE
LOWpackages/sdk/python/caveman_cloud/core.py403CODE
LOWpackages/sdk/python/caveman_cloud/core.py1745CODE
LOWpackages/sdk/python/caveman_cloud/core.py2307CODE
LOWskills/caveman-compress/scripts/validate.py112CODE
LOWskills/caveman-compress/scripts/detect.py56CODE
Redundant / Tautological Comments5 hits · 6 pts
SeverityFileLineSnippetContext
LOWplugins/caveman/skills/caveman-compress/scripts/cli.py55 # Check if compressibleCOMMENT
LOW…ns/caveman/skills/caveman-compress/scripts/compress.py407 # Check if backup already exists to prevent accidental overwritingSTRING
LOWskills/caveman-compress/scripts/cli.py55 # Check if compressibleCOMMENT
LOWskills/caveman-compress/scripts/compress.py407 # Check if backup already exists to prevent accidental overwritingSTRING
LOWsrc/hooks/install.sh48# Check if already installed (unless --force). Older installs only had two hookCOMMENT
Fake / Example Data3 hits · 4 pts
SeverityFileLineSnippetContext
LOWengine/compressors/testdata/axtree_cdp_local_page.json614 "type": "placeholder",CODE
LOWengine/compressors/testdata/axtree_cdp_local_page.json615 "attribute": "placeholder",CODE
LOWengine/compressors/testdata/axtree_cdp_local_page.json619 "type": "placeholder",CODE
Modern Structural Boilerplate3 hits · 3 pts
SeverityFileLineSnippetContext
LOW…ns/caveman/skills/caveman-compress/scripts/__init__.py7__all__ = ["cli", "compress", "detect", "validate"]CODE
LOWpackages/sdk/python/caveman_cloud/__init__.py28__all__ = [CODE
LOWskills/caveman-compress/scripts/__init__.py7__all__ = ["cli", "compress", "detect", "validate"]CODE
Slop Phrases1 hit · 2 pts
SeverityFileLineSnippetContext
MEDIUMproxy/routing/confidence_test.go17 "As an AI language model, I am unable to do that.",CODE
Verbosity Indicators4 hits · 2 pts
SeverityFileLineSnippetContext
LOW…ns/caveman/skills/caveman-compress/scripts/compress.py425 # Step 1: Compress (body only, frontmatter excluded)STRING
LOW…ns/caveman/skills/caveman-compress/scripts/compress.py461 # Step 2: Validate + RetrySTRING
LOWskills/caveman-compress/scripts/compress.py425 # Step 1: Compress (body only, frontmatter excluded)STRING
LOWskills/caveman-compress/scripts/compress.py461 # Step 2: Validate + RetrySTRING
Structural Annotation Overuse4 hits · 2 pts
SeverityFileLineSnippetContext
LOW…ns/caveman/skills/caveman-compress/scripts/compress.py425 # Step 1: Compress (body only, frontmatter excluded)STRING
LOW…ns/caveman/skills/caveman-compress/scripts/compress.py461 # Step 2: Validate + RetrySTRING
LOWskills/caveman-compress/scripts/compress.py425 # Step 1: Compress (body only, frontmatter excluded)STRING
LOWskills/caveman-compress/scripts/compress.py461 # Step 2: Validate + RetrySTRING
AI Structural Patterns1 hit · 1 pts
SeverityFileLineSnippetContext
LOWpackages/sdk/python/caveman_cloud/core.py1639CODE